diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8652 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.982278481012658,
+  "eval_steps": 500,
+  "global_step": 1230,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004050632911392405,
+      "grad_norm": 6.452088848190726,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 1.0447,
+      "step": 1
+    },
+    {
+      "epoch": 0.00810126582278481,
+      "grad_norm": 6.521517400959134,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 1.0567,
+      "step": 2
+    },
+    {
+      "epoch": 0.012151898734177215,
+      "grad_norm": 6.605400243860161,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 1.0575,
+      "step": 3
+    },
+    {
+      "epoch": 0.01620253164556962,
+      "grad_norm": 6.3685015946017565,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 1.0441,
+      "step": 4
+    },
+    {
+      "epoch": 0.020253164556962026,
+      "grad_norm": 5.896795996007261,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 1.0194,
+      "step": 5
+    },
+    {
+      "epoch": 0.02430379746835443,
+      "grad_norm": 5.768909788051986,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.0112,
+      "step": 6
+    },
+    {
+      "epoch": 0.028354430379746835,
+      "grad_norm": 4.532996446557441,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 1.0053,
+      "step": 7
+    },
+    {
+      "epoch": 0.03240506329113924,
+      "grad_norm": 4.262178810227945,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 0.9883,
+      "step": 8
+    },
+    {
+      "epoch": 0.03645569620253165,
+      "grad_norm": 2.7383256082556033,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 0.9672,
+      "step": 9
+    },
+    {
+      "epoch": 0.04050632911392405,
+      "grad_norm": 2.6104817735772348,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.9679,
+      "step": 10
+    },
+    {
+      "epoch": 0.044556962025316456,
+      "grad_norm": 2.4652486776677858,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.9473,
+      "step": 11
+    },
+    {
+      "epoch": 0.04860759493670886,
+      "grad_norm": 3.7623968693307117,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.9301,
+      "step": 12
+    },
+    {
+      "epoch": 0.052658227848101265,
+      "grad_norm": 4.16257539621541,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 0.9432,
+      "step": 13
+    },
+    {
+      "epoch": 0.05670886075949367,
+      "grad_norm": 4.1614194580649615,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 0.9423,
+      "step": 14
+    },
+    {
+      "epoch": 0.060759493670886074,
+      "grad_norm": 3.7440218110005317,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.9269,
+      "step": 15
+    },
+    {
+      "epoch": 0.06481012658227848,
+      "grad_norm": 2.7414105185530366,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 0.8869,
+      "step": 16
+    },
+    {
+      "epoch": 0.06886075949367089,
+      "grad_norm": 2.603607911545143,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 0.8884,
+      "step": 17
+    },
+    {
+      "epoch": 0.0729113924050633,
+      "grad_norm": 2.184008928127837,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.8793,
+      "step": 18
+    },
+    {
+      "epoch": 0.0769620253164557,
+      "grad_norm": 1.5674401230193302,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.8458,
+      "step": 19
+    },
+    {
+      "epoch": 0.0810126582278481,
+      "grad_norm": 1.1449927985859576,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.8388,
+      "step": 20
+    },
+    {
+      "epoch": 0.08506329113924051,
+      "grad_norm": 1.1495500940358077,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.8321,
+      "step": 21
+    },
+    {
+      "epoch": 0.08911392405063291,
+      "grad_norm": 1.2559623447113588,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 0.8085,
+      "step": 22
+    },
+    {
+      "epoch": 0.09316455696202532,
+      "grad_norm": 1.1296481628195054,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 0.8127,
+      "step": 23
+    },
+    {
+      "epoch": 0.09721518987341772,
+      "grad_norm": 0.9758082471872678,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.7963,
+      "step": 24
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 0.9090407110445023,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.7735,
+      "step": 25
+    },
+    {
+      "epoch": 0.10531645569620253,
+      "grad_norm": 0.902158239518975,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 0.7627,
+      "step": 26
+    },
+    {
+      "epoch": 0.10936708860759493,
+      "grad_norm": 0.9181729098490561,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.7848,
+      "step": 27
+    },
+    {
+      "epoch": 0.11341772151898734,
+      "grad_norm": 0.7859631929558692,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 0.776,
+      "step": 28
+    },
+    {
+      "epoch": 0.11746835443037974,
+      "grad_norm": 0.8235908387422828,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 0.7818,
+      "step": 29
+    },
+    {
+      "epoch": 0.12151898734177215,
+      "grad_norm": 0.7794891023372312,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.7531,
+      "step": 30
+    },
+    {
+      "epoch": 0.12556962025316457,
+      "grad_norm": 0.7957291883131091,
+      "learning_rate": 1.008130081300813e-05,
+      "loss": 0.7519,
+      "step": 31
+    },
+    {
+      "epoch": 0.12962025316455697,
+      "grad_norm": 0.7543855121307167,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 0.7663,
+      "step": 32
+    },
+    {
+      "epoch": 0.13367088607594937,
+      "grad_norm": 0.7015018429304976,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.7531,
+      "step": 33
+    },
+    {
+      "epoch": 0.13772151898734178,
+      "grad_norm": 0.7997164443360236,
+      "learning_rate": 1.1056910569105692e-05,
+      "loss": 0.7477,
+      "step": 34
+    },
+    {
+      "epoch": 0.14177215189873418,
+      "grad_norm": 0.7404233637343466,
+      "learning_rate": 1.1382113821138213e-05,
+      "loss": 0.7341,
+      "step": 35
+    },
+    {
+      "epoch": 0.1458227848101266,
+      "grad_norm": 0.6504349103027512,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.7302,
+      "step": 36
+    },
+    {
+      "epoch": 0.149873417721519,
+      "grad_norm": 0.8002893890047319,
+      "learning_rate": 1.2032520325203254e-05,
+      "loss": 0.7298,
+      "step": 37
+    },
+    {
+      "epoch": 0.1539240506329114,
+      "grad_norm": 0.6065791822002672,
+      "learning_rate": 1.2357723577235774e-05,
+      "loss": 0.7558,
+      "step": 38
+    },
+    {
+      "epoch": 0.1579746835443038,
+      "grad_norm": 0.6283785579453666,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.7272,
+      "step": 39
+    },
+    {
+      "epoch": 0.1620253164556962,
+      "grad_norm": 0.6077912642488584,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.7305,
+      "step": 40
+    },
+    {
+      "epoch": 0.1660759493670886,
+      "grad_norm": 0.6097035211047714,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7182,
+      "step": 41
+    },
+    {
+      "epoch": 0.17012658227848101,
+      "grad_norm": 0.548234584314678,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.7299,
+      "step": 42
+    },
+    {
+      "epoch": 0.17417721518987342,
+      "grad_norm": 0.6410853337107066,
+      "learning_rate": 1.3983739837398376e-05,
+      "loss": 0.7202,
+      "step": 43
+    },
+    {
+      "epoch": 0.17822784810126582,
+      "grad_norm": 0.6546590259848933,
+      "learning_rate": 1.4308943089430896e-05,
+      "loss": 0.7062,
+      "step": 44
+    },
+    {
+      "epoch": 0.18227848101265823,
+      "grad_norm": 0.5920014034747922,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.697,
+      "step": 45
+    },
+    {
+      "epoch": 0.18632911392405063,
+      "grad_norm": 0.6529556009652181,
+      "learning_rate": 1.4959349593495935e-05,
+      "loss": 0.7294,
+      "step": 46
+    },
+    {
+      "epoch": 0.19037974683544304,
+      "grad_norm": 0.6206324726172597,
+      "learning_rate": 1.528455284552846e-05,
+      "loss": 0.7188,
+      "step": 47
+    },
+    {
+      "epoch": 0.19443037974683544,
+      "grad_norm": 0.5563344578409094,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.7113,
+      "step": 48
+    },
+    {
+      "epoch": 0.19848101265822785,
+      "grad_norm": 0.6566299753347189,
+      "learning_rate": 1.5934959349593496e-05,
+      "loss": 0.6906,
+      "step": 49
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 0.5892130059159871,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.7138,
+      "step": 50
+    },
+    {
+      "epoch": 0.20658227848101265,
+      "grad_norm": 0.6008830419174666,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.6686,
+      "step": 51
+    },
+    {
+      "epoch": 0.21063291139240506,
+      "grad_norm": 0.4797280372590014,
+      "learning_rate": 1.691056910569106e-05,
+      "loss": 0.6933,
+      "step": 52
+    },
+    {
+      "epoch": 0.21468354430379746,
+      "grad_norm": 0.5625000634571587,
+      "learning_rate": 1.7235772357723578e-05,
+      "loss": 0.693,
+      "step": 53
+    },
+    {
+      "epoch": 0.21873417721518987,
+      "grad_norm": 0.44754520003207926,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.7011,
+      "step": 54
+    },
+    {
+      "epoch": 0.22278481012658227,
+      "grad_norm": 0.5715504144446111,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 0.7008,
+      "step": 55
+    },
+    {
+      "epoch": 0.22683544303797468,
+      "grad_norm": 0.6430440181561486,
+      "learning_rate": 1.821138211382114e-05,
+      "loss": 0.6942,
+      "step": 56
+    },
+    {
+      "epoch": 0.23088607594936708,
+      "grad_norm": 0.4541449376139326,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.6976,
+      "step": 57
+    },
+    {
+      "epoch": 0.23493670886075949,
+      "grad_norm": 0.6051254044134815,
+      "learning_rate": 1.886178861788618e-05,
+      "loss": 0.7019,
+      "step": 58
+    },
+    {
+      "epoch": 0.2389873417721519,
+      "grad_norm": 0.6535188153603061,
+      "learning_rate": 1.91869918699187e-05,
+      "loss": 0.6994,
+      "step": 59
+    },
+    {
+      "epoch": 0.2430379746835443,
+      "grad_norm": 0.6996453268057852,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.684,
+      "step": 60
+    },
+    {
+      "epoch": 0.2470886075949367,
+      "grad_norm": 0.7822527661573047,
+      "learning_rate": 1.983739837398374e-05,
+      "loss": 0.6892,
+      "step": 61
+    },
+    {
+      "epoch": 0.25113924050632913,
+      "grad_norm": 0.7546161636038426,
+      "learning_rate": 2.016260162601626e-05,
+      "loss": 0.6781,
+      "step": 62
+    },
+    {
+      "epoch": 0.25518987341772154,
+      "grad_norm": 0.563310330614363,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 0.6763,
+      "step": 63
+    },
+    {
+      "epoch": 0.25924050632911394,
+      "grad_norm": 0.6035273096615805,
+      "learning_rate": 2.0813008130081303e-05,
+      "loss": 0.6834,
+      "step": 64
+    },
+    {
+      "epoch": 0.26329113924050634,
+      "grad_norm": 0.7137129818695536,
+      "learning_rate": 2.1138211382113822e-05,
+      "loss": 0.6996,
+      "step": 65
+    },
+    {
+      "epoch": 0.26734177215189875,
+      "grad_norm": 0.8184595931762164,
+      "learning_rate": 2.1463414634146344e-05,
+      "loss": 0.7031,
+      "step": 66
+    },
+    {
+      "epoch": 0.27139240506329115,
+      "grad_norm": 0.6913879682461097,
+      "learning_rate": 2.1788617886178863e-05,
+      "loss": 0.6695,
+      "step": 67
+    },
+    {
+      "epoch": 0.27544303797468356,
+      "grad_norm": 0.4873803955287146,
+      "learning_rate": 2.2113821138211385e-05,
+      "loss": 0.6947,
+      "step": 68
+    },
+    {
+      "epoch": 0.27949367088607596,
+      "grad_norm": 0.7021888816466603,
+      "learning_rate": 2.2439024390243907e-05,
+      "loss": 0.674,
+      "step": 69
+    },
+    {
+      "epoch": 0.28354430379746837,
+      "grad_norm": 0.7989860784138514,
+      "learning_rate": 2.2764227642276426e-05,
+      "loss": 0.6815,
+      "step": 70
+    },
+    {
+      "epoch": 0.28759493670886077,
+      "grad_norm": 0.6732646232763189,
+      "learning_rate": 2.3089430894308948e-05,
+      "loss": 0.6687,
+      "step": 71
+    },
+    {
+      "epoch": 0.2916455696202532,
+      "grad_norm": 0.8646331078179385,
+      "learning_rate": 2.3414634146341463e-05,
+      "loss": 0.689,
+      "step": 72
+    },
+    {
+      "epoch": 0.2956962025316456,
+      "grad_norm": 1.3019502568600194,
+      "learning_rate": 2.3739837398373985e-05,
+      "loss": 0.6687,
+      "step": 73
+    },
+    {
+      "epoch": 0.299746835443038,
+      "grad_norm": 0.7820590511775346,
+      "learning_rate": 2.4065040650406507e-05,
+      "loss": 0.6827,
+      "step": 74
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 0.7393228554526755,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.6739,
+      "step": 75
+    },
+    {
+      "epoch": 0.3078481012658228,
+      "grad_norm": 0.7548721439293152,
+      "learning_rate": 2.4715447154471548e-05,
+      "loss": 0.6892,
+      "step": 76
+    },
+    {
+      "epoch": 0.3118987341772152,
+      "grad_norm": 1.0113235864660404,
+      "learning_rate": 2.5040650406504066e-05,
+      "loss": 0.6769,
+      "step": 77
+    },
+    {
+      "epoch": 0.3159493670886076,
+      "grad_norm": 1.360211501950147,
+      "learning_rate": 2.536585365853659e-05,
+      "loss": 0.6752,
+      "step": 78
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6595769921562289,
+      "learning_rate": 2.569105691056911e-05,
+      "loss": 0.6633,
+      "step": 79
+    },
+    {
+      "epoch": 0.3240506329113924,
+      "grad_norm": 0.7956562908093638,
+      "learning_rate": 2.601626016260163e-05,
+      "loss": 0.6682,
+      "step": 80
+    },
+    {
+      "epoch": 0.3281012658227848,
+      "grad_norm": 0.9548585312886568,
+      "learning_rate": 2.634146341463415e-05,
+      "loss": 0.6757,
+      "step": 81
+    },
+    {
+      "epoch": 0.3321518987341772,
+      "grad_norm": 0.7445850134719754,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.6878,
+      "step": 82
+    },
+    {
+      "epoch": 0.3362025316455696,
+      "grad_norm": 0.7253909753267015,
+      "learning_rate": 2.699186991869919e-05,
+      "loss": 0.6708,
+      "step": 83
+    },
+    {
+      "epoch": 0.34025316455696203,
+      "grad_norm": 0.9796313875124703,
+      "learning_rate": 2.731707317073171e-05,
+      "loss": 0.6641,
+      "step": 84
+    },
+    {
+      "epoch": 0.34430379746835443,
+      "grad_norm": 1.413947607830908,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 0.6719,
+      "step": 85
+    },
+    {
+      "epoch": 0.34835443037974684,
+      "grad_norm": 0.6915298808000039,
+      "learning_rate": 2.796747967479675e-05,
+      "loss": 0.6747,
+      "step": 86
+    },
+    {
+      "epoch": 0.35240506329113924,
+      "grad_norm": 1.4154032753659296,
+      "learning_rate": 2.829268292682927e-05,
+      "loss": 0.6662,
+      "step": 87
+    },
+    {
+      "epoch": 0.35645569620253165,
+      "grad_norm": 0.8183771656919938,
+      "learning_rate": 2.8617886178861792e-05,
+      "loss": 0.6556,
+      "step": 88
+    },
+    {
+      "epoch": 0.36050632911392405,
+      "grad_norm": 1.213988528595901,
+      "learning_rate": 2.8943089430894314e-05,
+      "loss": 0.6728,
+      "step": 89
+    },
+    {
+      "epoch": 0.36455696202531646,
+      "grad_norm": 1.0819874570948689,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.6666,
+      "step": 90
+    },
+    {
+      "epoch": 0.36860759493670886,
+      "grad_norm": 0.914248834075912,
+      "learning_rate": 2.959349593495935e-05,
+      "loss": 0.6876,
+      "step": 91
+    },
+    {
+      "epoch": 0.37265822784810126,
+      "grad_norm": 1.0133362759090452,
+      "learning_rate": 2.991869918699187e-05,
+      "loss": 0.6706,
+      "step": 92
+    },
+    {
+      "epoch": 0.37670886075949367,
+      "grad_norm": 1.806592675978055,
+      "learning_rate": 3.0243902439024392e-05,
+      "loss": 0.6682,
+      "step": 93
+    },
+    {
+      "epoch": 0.3807594936708861,
+      "grad_norm": 0.7614409968752955,
+      "learning_rate": 3.056910569105692e-05,
+      "loss": 0.6707,
+      "step": 94
+    },
+    {
+      "epoch": 0.3848101265822785,
+      "grad_norm": 2.23432705499668,
+      "learning_rate": 3.089430894308943e-05,
+      "loss": 0.6751,
+      "step": 95
+    },
+    {
+      "epoch": 0.3888607594936709,
+      "grad_norm": 1.478174203599363,
+      "learning_rate": 3.1219512195121955e-05,
+      "loss": 0.6691,
+      "step": 96
+    },
+    {
+      "epoch": 0.3929113924050633,
+      "grad_norm": 1.891250318516487,
+      "learning_rate": 3.154471544715447e-05,
+      "loss": 0.6706,
+      "step": 97
+    },
+    {
+      "epoch": 0.3969620253164557,
+      "grad_norm": 1.7918029054815314,
+      "learning_rate": 3.186991869918699e-05,
+      "loss": 0.6576,
+      "step": 98
+    },
+    {
+      "epoch": 0.4010126582278481,
+      "grad_norm": 1.1706416925295287,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 0.6718,
+      "step": 99
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 1.9846087299741704,
+      "learning_rate": 3.2520325203252037e-05,
+      "loss": 0.661,
+      "step": 100
+    },
+    {
+      "epoch": 0.4091139240506329,
+      "grad_norm": 1.2071100664598156,
+      "learning_rate": 3.284552845528456e-05,
+      "loss": 0.673,
+      "step": 101
+    },
+    {
+      "epoch": 0.4131645569620253,
+      "grad_norm": 2.1329941029623547,
+      "learning_rate": 3.3170731707317074e-05,
+      "loss": 0.6682,
+      "step": 102
+    },
+    {
+      "epoch": 0.4172151898734177,
+      "grad_norm": 1.6863710530998095,
+      "learning_rate": 3.3495934959349596e-05,
+      "loss": 0.677,
+      "step": 103
+    },
+    {
+      "epoch": 0.4212658227848101,
+      "grad_norm": 1.7096647435885,
+      "learning_rate": 3.382113821138212e-05,
+      "loss": 0.6675,
+      "step": 104
+    },
+    {
+      "epoch": 0.4253164556962025,
+      "grad_norm": 1.3351792421070592,
+      "learning_rate": 3.414634146341463e-05,
+      "loss": 0.6719,
+      "step": 105
+    },
+    {
+      "epoch": 0.4293670886075949,
+      "grad_norm": 1.8217804699906697,
+      "learning_rate": 3.4471544715447155e-05,
+      "loss": 0.6576,
+      "step": 106
+    },
+    {
+      "epoch": 0.43341772151898733,
+      "grad_norm": 0.9586450200464353,
+      "learning_rate": 3.479674796747968e-05,
+      "loss": 0.648,
+      "step": 107
+    },
+    {
+      "epoch": 0.43746835443037974,
+      "grad_norm": 1.759653161553532,
+      "learning_rate": 3.51219512195122e-05,
+      "loss": 0.6689,
+      "step": 108
+    },
+    {
+      "epoch": 0.44151898734177214,
+      "grad_norm": 1.0903019300012775,
+      "learning_rate": 3.544715447154472e-05,
+      "loss": 0.6444,
+      "step": 109
+    },
+    {
+      "epoch": 0.44556962025316454,
+      "grad_norm": 2.1536320527101465,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 0.6825,
+      "step": 110
+    },
+    {
+      "epoch": 0.44962025316455695,
+      "grad_norm": 1.6542345264329172,
+      "learning_rate": 3.609756097560976e-05,
+      "loss": 0.6751,
+      "step": 111
+    },
+    {
+      "epoch": 0.45367088607594935,
+      "grad_norm": 1.6682161474499615,
+      "learning_rate": 3.642276422764228e-05,
+      "loss": 0.6673,
+      "step": 112
+    },
+    {
+      "epoch": 0.45772151898734176,
+      "grad_norm": 1.467070120959557,
+      "learning_rate": 3.67479674796748e-05,
+      "loss": 0.6649,
+      "step": 113
+    },
+    {
+      "epoch": 0.46177215189873416,
+      "grad_norm": 1.2264821631124763,
+      "learning_rate": 3.7073170731707325e-05,
+      "loss": 0.6521,
+      "step": 114
+    },
+    {
+      "epoch": 0.46582278481012657,
+      "grad_norm": 1.319087473995399,
+      "learning_rate": 3.739837398373984e-05,
+      "loss": 0.6641,
+      "step": 115
+    },
+    {
+      "epoch": 0.46987341772151897,
+      "grad_norm": 0.9547341828928255,
+      "learning_rate": 3.772357723577236e-05,
+      "loss": 0.655,
+      "step": 116
+    },
+    {
+      "epoch": 0.4739240506329114,
+      "grad_norm": 1.3038474457619562,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 0.6571,
+      "step": 117
+    },
+    {
+      "epoch": 0.4779746835443038,
+      "grad_norm": 1.2731985312443859,
+      "learning_rate": 3.83739837398374e-05,
+      "loss": 0.6561,
+      "step": 118
+    },
+    {
+      "epoch": 0.4820253164556962,
+      "grad_norm": 0.9354998504716119,
+      "learning_rate": 3.869918699186992e-05,
+      "loss": 0.6556,
+      "step": 119
+    },
+    {
+      "epoch": 0.4860759493670886,
+      "grad_norm": 1.4219202402758675,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.6529,
+      "step": 120
+    },
+    {
+      "epoch": 0.490126582278481,
+      "grad_norm": 0.9517066377754545,
+      "learning_rate": 3.9349593495934966e-05,
+      "loss": 0.6544,
+      "step": 121
+    },
+    {
+      "epoch": 0.4941772151898734,
+      "grad_norm": 1.2966422599945817,
+      "learning_rate": 3.967479674796748e-05,
+      "loss": 0.6587,
+      "step": 122
+    },
+    {
+      "epoch": 0.4982278481012658,
+      "grad_norm": 0.9663907026308576,
+      "learning_rate": 4e-05,
+      "loss": 0.6453,
+      "step": 123
+    },
+    {
+      "epoch": 0.5022784810126583,
+      "grad_norm": 1.0914036432193444,
+      "learning_rate": 3.999991946137476e-05,
+      "loss": 0.6513,
+      "step": 124
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 1.1032638323674273,
+      "learning_rate": 3.999967784614766e-05,
+      "loss": 0.6568,
+      "step": 125
+    },
+    {
+      "epoch": 0.5103797468354431,
+      "grad_norm": 1.0063056772337013,
+      "learning_rate": 3.9999275156264656e-05,
+      "loss": 0.6524,
+      "step": 126
+    },
+    {
+      "epoch": 0.5144303797468355,
+      "grad_norm": 0.9241185168159832,
+      "learning_rate": 3.999871139496895e-05,
+      "loss": 0.6387,
+      "step": 127
+    },
+    {
+      "epoch": 0.5184810126582279,
+      "grad_norm": 0.9796917341865563,
+      "learning_rate": 3.9997986566800995e-05,
+      "loss": 0.6661,
+      "step": 128
+    },
+    {
+      "epoch": 0.5225316455696203,
+      "grad_norm": 1.0272963105799917,
+      "learning_rate": 3.999710067759846e-05,
+      "loss": 0.6542,
+      "step": 129
+    },
+    {
+      "epoch": 0.5265822784810127,
+      "grad_norm": 1.140751722148569,
+      "learning_rate": 3.999605373449617e-05,
+      "loss": 0.6478,
+      "step": 130
+    },
+    {
+      "epoch": 0.5306329113924051,
+      "grad_norm": 1.0207173297351482,
+      "learning_rate": 3.9994845745926075e-05,
+      "loss": 0.6611,
+      "step": 131
+    },
+    {
+      "epoch": 0.5346835443037975,
+      "grad_norm": 1.2988309656474655,
+      "learning_rate": 3.999347672161713e-05,
+      "loss": 0.6658,
+      "step": 132
+    },
+    {
+      "epoch": 0.5387341772151899,
+      "grad_norm": 1.0613189751410246,
+      "learning_rate": 3.999194667259528e-05,
+      "loss": 0.6323,
+      "step": 133
+    },
+    {
+      "epoch": 0.5427848101265823,
+      "grad_norm": 1.19896227253905,
+      "learning_rate": 3.999025561118334e-05,
+      "loss": 0.6464,
+      "step": 134
+    },
+    {
+      "epoch": 0.5468354430379747,
+      "grad_norm": 0.9228285440487731,
+      "learning_rate": 3.998840355100086e-05,
+      "loss": 0.6512,
+      "step": 135
+    },
+    {
+      "epoch": 0.5508860759493671,
+      "grad_norm": 1.3813734463259375,
+      "learning_rate": 3.998639050696409e-05,
+      "loss": 0.6479,
+      "step": 136
+    },
+    {
+      "epoch": 0.5549367088607595,
+      "grad_norm": 1.0439398060705125,
+      "learning_rate": 3.998421649528582e-05,
+      "loss": 0.6512,
+      "step": 137
+    },
+    {
+      "epoch": 0.5589873417721519,
+      "grad_norm": 1.2081672543066029,
+      "learning_rate": 3.9981881533475234e-05,
+      "loss": 0.6415,
+      "step": 138
+    },
+    {
+      "epoch": 0.5630379746835443,
+      "grad_norm": 1.3285921422341929,
+      "learning_rate": 3.997938564033779e-05,
+      "loss": 0.6559,
+      "step": 139
+    },
+    {
+      "epoch": 0.5670886075949367,
+      "grad_norm": 0.7857391139622759,
+      "learning_rate": 3.9976728835975064e-05,
+      "loss": 0.6585,
+      "step": 140
+    },
+    {
+      "epoch": 0.5711392405063291,
+      "grad_norm": 1.2710476679836518,
+      "learning_rate": 3.9973911141784605e-05,
+      "loss": 0.6604,
+      "step": 141
+    },
+    {
+      "epoch": 0.5751898734177215,
+      "grad_norm": 0.6892713976859774,
+      "learning_rate": 3.997093258045973e-05,
+      "loss": 0.6319,
+      "step": 142
+    },
+    {
+      "epoch": 0.579240506329114,
+      "grad_norm": 0.9589066748939733,
+      "learning_rate": 3.996779317598936e-05,
+      "loss": 0.6461,
+      "step": 143
+    },
+    {
+      "epoch": 0.5832911392405064,
+      "grad_norm": 0.7139300974949718,
+      "learning_rate": 3.996449295365782e-05,
+      "loss": 0.6369,
+      "step": 144
+    },
+    {
+      "epoch": 0.5873417721518988,
+      "grad_norm": 0.8149131377876515,
+      "learning_rate": 3.996103194004467e-05,
+      "loss": 0.6391,
+      "step": 145
+    },
+    {
+      "epoch": 0.5913924050632912,
+      "grad_norm": 0.8034723461468668,
+      "learning_rate": 3.995741016302441e-05,
+      "loss": 0.6675,
+      "step": 146
+    },
+    {
+      "epoch": 0.5954430379746836,
+      "grad_norm": 0.6686801660090204,
+      "learning_rate": 3.9953627651766364e-05,
+      "loss": 0.6533,
+      "step": 147
+    },
+    {
+      "epoch": 0.599493670886076,
+      "grad_norm": 0.6042588125843888,
+      "learning_rate": 3.9949684436734325e-05,
+      "loss": 0.6443,
+      "step": 148
+    },
+    {
+      "epoch": 0.6035443037974684,
+      "grad_norm": 0.6448845428120026,
+      "learning_rate": 3.994558054968643e-05,
+      "loss": 0.6388,
+      "step": 149
+    },
+    {
+      "epoch": 0.6075949367088608,
+      "grad_norm": 0.6811191122083278,
+      "learning_rate": 3.994131602367481e-05,
+      "loss": 0.6426,
+      "step": 150
+    },
+    {
+      "epoch": 0.6116455696202532,
+      "grad_norm": 0.6417686533073053,
+      "learning_rate": 3.9936890893045376e-05,
+      "loss": 0.6525,
+      "step": 151
+    },
+    {
+      "epoch": 0.6156962025316456,
+      "grad_norm": 0.612067542483574,
+      "learning_rate": 3.993230519343752e-05,
+      "loss": 0.635,
+      "step": 152
+    },
+    {
+      "epoch": 0.619746835443038,
+      "grad_norm": 0.7803297124568453,
+      "learning_rate": 3.992755896178383e-05,
+      "loss": 0.6555,
+      "step": 153
+    },
+    {
+      "epoch": 0.6237974683544304,
+      "grad_norm": 0.8118985689906424,
+      "learning_rate": 3.992265223630981e-05,
+      "loss": 0.6359,
+      "step": 154
+    },
+    {
+      "epoch": 0.6278481012658228,
+      "grad_norm": 0.7841474767181533,
+      "learning_rate": 3.991758505653355e-05,
+      "loss": 0.6349,
+      "step": 155
+    },
+    {
+      "epoch": 0.6318987341772152,
+      "grad_norm": 0.5974488584901587,
+      "learning_rate": 3.991235746326543e-05,
+      "loss": 0.6274,
+      "step": 156
+    },
+    {
+      "epoch": 0.6359493670886076,
+      "grad_norm": 0.7883800552239252,
+      "learning_rate": 3.9906969498607745e-05,
+      "loss": 0.6315,
+      "step": 157
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5388167759177698,
+      "learning_rate": 3.990142120595444e-05,
+      "loss": 0.6355,
+      "step": 158
+    },
+    {
+      "epoch": 0.6440506329113924,
+      "grad_norm": 0.8001751218496637,
+      "learning_rate": 3.98957126299907e-05,
+      "loss": 0.6302,
+      "step": 159
+    },
+    {
+      "epoch": 0.6481012658227848,
+      "grad_norm": 0.7695681342630445,
+      "learning_rate": 3.9889843816692596e-05,
+      "loss": 0.639,
+      "step": 160
+    },
+    {
+      "epoch": 0.6521518987341772,
+      "grad_norm": 0.5199225568508513,
+      "learning_rate": 3.9883814813326766e-05,
+      "loss": 0.6282,
+      "step": 161
+    },
+    {
+      "epoch": 0.6562025316455696,
+      "grad_norm": 0.7487207902891283,
+      "learning_rate": 3.9877625668449956e-05,
+      "loss": 0.6512,
+      "step": 162
+    },
+    {
+      "epoch": 0.660253164556962,
+      "grad_norm": 0.6669119753974444,
+      "learning_rate": 3.98712764319087e-05,
+      "loss": 0.6254,
+      "step": 163
+    },
+    {
+      "epoch": 0.6643037974683544,
+      "grad_norm": 0.49401633800094236,
+      "learning_rate": 3.9864767154838864e-05,
+      "loss": 0.6359,
+      "step": 164
+    },
+    {
+      "epoch": 0.6683544303797468,
+      "grad_norm": 0.6421135601793051,
+      "learning_rate": 3.9858097889665277e-05,
+      "loss": 0.6556,
+      "step": 165
+    },
+    {
+      "epoch": 0.6724050632911392,
+      "grad_norm": 0.6329425851176634,
+      "learning_rate": 3.985126869010129e-05,
+      "loss": 0.6303,
+      "step": 166
+    },
+    {
+      "epoch": 0.6764556962025317,
+      "grad_norm": 0.8353126746438074,
+      "learning_rate": 3.984427961114833e-05,
+      "loss": 0.6279,
+      "step": 167
+    },
+    {
+      "epoch": 0.6805063291139241,
+      "grad_norm": 0.7935621313010524,
+      "learning_rate": 3.9837130709095475e-05,
+      "loss": 0.6221,
+      "step": 168
+    },
+    {
+      "epoch": 0.6845569620253165,
+      "grad_norm": 0.6963875756861462,
+      "learning_rate": 3.982982204151901e-05,
+      "loss": 0.6332,
+      "step": 169
+    },
+    {
+      "epoch": 0.6886075949367089,
+      "grad_norm": 0.8078167621712001,
+      "learning_rate": 3.982235366728193e-05,
+      "loss": 0.6319,
+      "step": 170
+    },
+    {
+      "epoch": 0.6926582278481013,
+      "grad_norm": 0.8645788363322172,
+      "learning_rate": 3.9814725646533505e-05,
+      "loss": 0.6411,
+      "step": 171
+    },
+    {
+      "epoch": 0.6967088607594937,
+      "grad_norm": 0.9248142962810029,
+      "learning_rate": 3.9806938040708746e-05,
+      "loss": 0.6338,
+      "step": 172
+    },
+    {
+      "epoch": 0.7007594936708861,
+      "grad_norm": 1.0179133821538575,
+      "learning_rate": 3.9798990912527976e-05,
+      "loss": 0.6295,
+      "step": 173
+    },
+    {
+      "epoch": 0.7048101265822785,
+      "grad_norm": 0.8176638642103785,
+      "learning_rate": 3.979088432599627e-05,
+      "loss": 0.6385,
+      "step": 174
+    },
+    {
+      "epoch": 0.7088607594936709,
+      "grad_norm": 0.4774284771159163,
+      "learning_rate": 3.9782618346402964e-05,
+      "loss": 0.6424,
+      "step": 175
+    },
+    {
+      "epoch": 0.7129113924050633,
+      "grad_norm": 1.0051078278985945,
+      "learning_rate": 3.977419304032111e-05,
+      "loss": 0.6291,
+      "step": 176
+    },
+    {
+      "epoch": 0.7169620253164557,
+      "grad_norm": 1.0151594640547004,
+      "learning_rate": 3.976560847560697e-05,
+      "loss": 0.6381,
+      "step": 177
+    },
+    {
+      "epoch": 0.7210126582278481,
+      "grad_norm": 0.610856520901865,
+      "learning_rate": 3.9756864721399456e-05,
+      "loss": 0.6414,
+      "step": 178
+    },
+    {
+      "epoch": 0.7250632911392405,
+      "grad_norm": 0.4833324249240733,
+      "learning_rate": 3.974796184811956e-05,
+      "loss": 0.6281,
+      "step": 179
+    },
+    {
+      "epoch": 0.7291139240506329,
+      "grad_norm": 0.7994548829689329,
+      "learning_rate": 3.973889992746979e-05,
+      "loss": 0.639,
+      "step": 180
+    },
+    {
+      "epoch": 0.7331645569620253,
+      "grad_norm": 0.7946628115595455,
+      "learning_rate": 3.972967903243361e-05,
+      "loss": 0.6233,
+      "step": 181
+    },
+    {
+      "epoch": 0.7372151898734177,
+      "grad_norm": 0.6257294355964683,
+      "learning_rate": 3.972029923727486e-05,
+      "loss": 0.6343,
+      "step": 182
+    },
+    {
+      "epoch": 0.7412658227848101,
+      "grad_norm": 0.46117199980140844,
+      "learning_rate": 3.971076061753709e-05,
+      "loss": 0.6286,
+      "step": 183
+    },
+    {
+      "epoch": 0.7453164556962025,
+      "grad_norm": 0.6010999767123585,
+      "learning_rate": 3.9701063250043066e-05,
+      "loss": 0.6212,
+      "step": 184
+    },
+    {
+      "epoch": 0.7493670886075949,
+      "grad_norm": 0.6725557206519044,
+      "learning_rate": 3.969120721289402e-05,
+      "loss": 0.6333,
+      "step": 185
+    },
+    {
+      "epoch": 0.7534177215189873,
+      "grad_norm": 0.5837863953031328,
+      "learning_rate": 3.9681192585469146e-05,
+      "loss": 0.6405,
+      "step": 186
+    },
+    {
+      "epoch": 0.7574683544303797,
+      "grad_norm": 0.5465735369863751,
+      "learning_rate": 3.9671019448424865e-05,
+      "loss": 0.6249,
+      "step": 187
+    },
+    {
+      "epoch": 0.7615189873417721,
+      "grad_norm": 0.5536638059954762,
+      "learning_rate": 3.966068788369422e-05,
+      "loss": 0.6357,
+      "step": 188
+    },
+    {
+      "epoch": 0.7655696202531646,
+      "grad_norm": 0.6514184616296875,
+      "learning_rate": 3.965019797448622e-05,
+      "loss": 0.6277,
+      "step": 189
+    },
+    {
+      "epoch": 0.769620253164557,
+      "grad_norm": 0.6667313124136925,
+      "learning_rate": 3.963954980528515e-05,
+      "loss": 0.6295,
+      "step": 190
+    },
+    {
+      "epoch": 0.7736708860759494,
+      "grad_norm": 0.4325642572551252,
+      "learning_rate": 3.9628743461849905e-05,
+      "loss": 0.6287,
+      "step": 191
+    },
+    {
+      "epoch": 0.7777215189873418,
+      "grad_norm": 0.5716228519275894,
+      "learning_rate": 3.961777903121329e-05,
+      "loss": 0.6402,
+      "step": 192
+    },
+    {
+      "epoch": 0.7817721518987342,
+      "grad_norm": 0.552478027885477,
+      "learning_rate": 3.960665660168131e-05,
+      "loss": 0.6399,
+      "step": 193
+    },
+    {
+      "epoch": 0.7858227848101266,
+      "grad_norm": 0.45935236554299774,
+      "learning_rate": 3.9595376262832485e-05,
+      "loss": 0.6465,
+      "step": 194
+    },
+    {
+      "epoch": 0.789873417721519,
+      "grad_norm": 0.48301774296638655,
+      "learning_rate": 3.9583938105517127e-05,
+      "loss": 0.6369,
+      "step": 195
+    },
+    {
+      "epoch": 0.7939240506329114,
+      "grad_norm": 0.493985013410535,
+      "learning_rate": 3.957234222185657e-05,
+      "loss": 0.6406,
+      "step": 196
+    },
+    {
+      "epoch": 0.7979746835443038,
+      "grad_norm": 0.5693281510121665,
+      "learning_rate": 3.9560588705242474e-05,
+      "loss": 0.6417,
+      "step": 197
+    },
+    {
+      "epoch": 0.8020253164556962,
+      "grad_norm": 0.6816163645298046,
+      "learning_rate": 3.954867765033605e-05,
+      "loss": 0.636,
+      "step": 198
+    },
+    {
+      "epoch": 0.8060759493670886,
+      "grad_norm": 0.5509891164502074,
+      "learning_rate": 3.953660915306728e-05,
+      "loss": 0.6388,
+      "step": 199
+    },
+    {
+      "epoch": 0.810126582278481,
+      "grad_norm": 0.6297857513777854,
+      "learning_rate": 3.952438331063419e-05,
+      "loss": 0.6147,
+      "step": 200
+    },
+    {
+      "epoch": 0.8141772151898734,
+      "grad_norm": 0.6134929964865111,
+      "learning_rate": 3.951200022150205e-05,
+      "loss": 0.6287,
+      "step": 201
+    },
+    {
+      "epoch": 0.8182278481012658,
+      "grad_norm": 0.508174055637192,
+      "learning_rate": 3.949945998540253e-05,
+      "loss": 0.6274,
+      "step": 202
+    },
+    {
+      "epoch": 0.8222784810126582,
+      "grad_norm": 0.582865982434805,
+      "learning_rate": 3.9486762703332993e-05,
+      "loss": 0.6239,
+      "step": 203
+    },
+    {
+      "epoch": 0.8263291139240506,
+      "grad_norm": 0.4889545572917407,
+      "learning_rate": 3.947390847755559e-05,
+      "loss": 0.6204,
+      "step": 204
+    },
+    {
+      "epoch": 0.830379746835443,
+      "grad_norm": 0.501656233473206,
+      "learning_rate": 3.946089741159648e-05,
+      "loss": 0.6216,
+      "step": 205
+    },
+    {
+      "epoch": 0.8344303797468354,
+      "grad_norm": 0.5532156040541987,
+      "learning_rate": 3.944772961024501e-05,
+      "loss": 0.6096,
+      "step": 206
+    },
+    {
+      "epoch": 0.8384810126582278,
+      "grad_norm": 0.5694778773435663,
+      "learning_rate": 3.943440517955285e-05,
+      "loss": 0.6394,
+      "step": 207
+    },
+    {
+      "epoch": 0.8425316455696202,
+      "grad_norm": 0.4508551136392328,
+      "learning_rate": 3.9420924226833126e-05,
+      "loss": 0.6261,
+      "step": 208
+    },
+    {
+      "epoch": 0.8465822784810126,
+      "grad_norm": 0.46517827503888587,
+      "learning_rate": 3.9407286860659566e-05,
+      "loss": 0.6226,
+      "step": 209
+    },
+    {
+      "epoch": 0.850632911392405,
+      "grad_norm": 0.5350229808276998,
+      "learning_rate": 3.9393493190865657e-05,
+      "loss": 0.6312,
+      "step": 210
+    },
+    {
+      "epoch": 0.8546835443037974,
+      "grad_norm": 0.519487003914067,
+      "learning_rate": 3.937954332854371e-05,
+      "loss": 0.6083,
+      "step": 211
+    },
+    {
+      "epoch": 0.8587341772151899,
+      "grad_norm": 0.5099419838314018,
+      "learning_rate": 3.9365437386044016e-05,
+      "loss": 0.615,
+      "step": 212
+    },
+    {
+      "epoch": 0.8627848101265823,
+      "grad_norm": 0.49316951694411215,
+      "learning_rate": 3.935117547697387e-05,
+      "loss": 0.6308,
+      "step": 213
+    },
+    {
+      "epoch": 0.8668354430379747,
+      "grad_norm": 0.5295535483120498,
+      "learning_rate": 3.933675771619675e-05,
+      "loss": 0.6328,
+      "step": 214
+    },
+    {
+      "epoch": 0.8708860759493671,
+      "grad_norm": 0.6222677877123625,
+      "learning_rate": 3.932218421983131e-05,
+      "loss": 0.6136,
+      "step": 215
+    },
+    {
+      "epoch": 0.8749367088607595,
+      "grad_norm": 0.7061027332106123,
+      "learning_rate": 3.9307455105250484e-05,
+      "loss": 0.62,
+      "step": 216
+    },
+    {
+      "epoch": 0.8789873417721519,
+      "grad_norm": 0.5660830534018042,
+      "learning_rate": 3.929257049108054e-05,
+      "loss": 0.621,
+      "step": 217
+    },
+    {
+      "epoch": 0.8830379746835443,
+      "grad_norm": 0.4025936062136094,
+      "learning_rate": 3.927753049720011e-05,
+      "loss": 0.6185,
+      "step": 218
+    },
+    {
+      "epoch": 0.8870886075949367,
+      "grad_norm": 0.44649716707035286,
+      "learning_rate": 3.9262335244739234e-05,
+      "loss": 0.6229,
+      "step": 219
+    },
+    {
+      "epoch": 0.8911392405063291,
+      "grad_norm": 0.5192572408458983,
+      "learning_rate": 3.92469848560784e-05,
+      "loss": 0.6184,
+      "step": 220
+    },
+    {
+      "epoch": 0.8951898734177215,
+      "grad_norm": 0.4570854855886648,
+      "learning_rate": 3.923147945484751e-05,
+      "loss": 0.6081,
+      "step": 221
+    },
+    {
+      "epoch": 0.8992405063291139,
+      "grad_norm": 0.45870729756310225,
+      "learning_rate": 3.9215819165924956e-05,
+      "loss": 0.6149,
+      "step": 222
+    },
+    {
+      "epoch": 0.9032911392405063,
+      "grad_norm": 0.4176111340291989,
+      "learning_rate": 3.920000411543654e-05,
+      "loss": 0.6274,
+      "step": 223
+    },
+    {
+      "epoch": 0.9073417721518987,
+      "grad_norm": 0.42098624456752476,
+      "learning_rate": 3.9184034430754495e-05,
+      "loss": 0.6134,
+      "step": 224
+    },
+    {
+      "epoch": 0.9113924050632911,
+      "grad_norm": 0.37244221564945695,
+      "learning_rate": 3.916791024049648e-05,
+      "loss": 0.6214,
+      "step": 225
+    },
+    {
+      "epoch": 0.9154430379746835,
+      "grad_norm": 0.5103605340746195,
+      "learning_rate": 3.91516316745245e-05,
+      "loss": 0.6305,
+      "step": 226
+    },
+    {
+      "epoch": 0.9194936708860759,
+      "grad_norm": 0.41584426140661734,
+      "learning_rate": 3.913519886394389e-05,
+      "loss": 0.6218,
+      "step": 227
+    },
+    {
+      "epoch": 0.9235443037974683,
+      "grad_norm": 0.42989572401323883,
+      "learning_rate": 3.911861194110225e-05,
+      "loss": 0.625,
+      "step": 228
+    },
+    {
+      "epoch": 0.9275949367088607,
+      "grad_norm": 0.504496243388637,
+      "learning_rate": 3.910187103958837e-05,
+      "loss": 0.6269,
+      "step": 229
+    },
+    {
+      "epoch": 0.9316455696202531,
+      "grad_norm": 0.4996867561539127,
+      "learning_rate": 3.908497629423117e-05,
+      "loss": 0.6287,
+      "step": 230
+    },
+    {
+      "epoch": 0.9356962025316455,
+      "grad_norm": 0.37166753512099604,
+      "learning_rate": 3.9067927841098614e-05,
+      "loss": 0.6628,
+      "step": 231
+    },
+    {
+      "epoch": 0.9397468354430379,
+      "grad_norm": 0.49976083212719596,
+      "learning_rate": 3.9050725817496594e-05,
+      "loss": 0.62,
+      "step": 232
+    },
+    {
+      "epoch": 0.9437974683544303,
+      "grad_norm": 0.5536630234042523,
+      "learning_rate": 3.9033370361967844e-05,
+      "loss": 0.6264,
+      "step": 233
+    },
+    {
+      "epoch": 0.9478481012658228,
+      "grad_norm": 0.4565903007165249,
+      "learning_rate": 3.901586161429081e-05,
+      "loss": 0.6224,
+      "step": 234
+    },
+    {
+      "epoch": 0.9518987341772152,
+      "grad_norm": 0.49198610374537827,
+      "learning_rate": 3.8998199715478545e-05,
+      "loss": 0.629,
+      "step": 235
+    },
+    {
+      "epoch": 0.9559493670886076,
+      "grad_norm": 0.46468731116655543,
+      "learning_rate": 3.8980384807777564e-05,
+      "loss": 0.6148,
+      "step": 236
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5247949073078567,
+      "learning_rate": 3.896241703466667e-05,
+      "loss": 0.6222,
+      "step": 237
+    },
+    {
+      "epoch": 0.9640506329113924,
+      "grad_norm": 0.4623581548021075,
+      "learning_rate": 3.894429654085585e-05,
+      "loss": 0.6289,
+      "step": 238
+    },
+    {
+      "epoch": 0.9681012658227848,
+      "grad_norm": 0.4576156432209765,
+      "learning_rate": 3.892602347228505e-05,
+      "loss": 0.607,
+      "step": 239
+    },
+    {
+      "epoch": 0.9721518987341772,
+      "grad_norm": 0.4425320696784023,
+      "learning_rate": 3.890759797612307e-05,
+      "loss": 0.6115,
+      "step": 240
+    },
+    {
+      "epoch": 0.9762025316455696,
+      "grad_norm": 0.46659742124115505,
+      "learning_rate": 3.888902020076632e-05,
+      "loss": 0.6073,
+      "step": 241
+    },
+    {
+      "epoch": 0.980253164556962,
+      "grad_norm": 0.4319041710789876,
+      "learning_rate": 3.887029029583764e-05,
+      "loss": 0.6317,
+      "step": 242
+    },
+    {
+      "epoch": 0.9843037974683544,
+      "grad_norm": 0.4807322068552483,
+      "learning_rate": 3.8851408412185125e-05,
+      "loss": 0.6209,
+      "step": 243
+    },
+    {
+      "epoch": 0.9883544303797468,
+      "grad_norm": 0.47289685342464033,
+      "learning_rate": 3.8832374701880855e-05,
+      "loss": 0.614,
+      "step": 244
+    },
+    {
+      "epoch": 0.9924050632911392,
+      "grad_norm": 0.526538965773254,
+      "learning_rate": 3.881318931821972e-05,
+      "loss": 0.6133,
+      "step": 245
+    },
+    {
+      "epoch": 0.9964556962025316,
+      "grad_norm": 0.5099834937974786,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.6184,
+      "step": 246
+    },
+    {
+      "epoch": 1.000506329113924,
+      "grad_norm": 0.6066124815827333,
+      "learning_rate": 3.8774364150112955e-05,
+      "loss": 0.6026,
+      "step": 247
+    },
+    {
+      "epoch": 1.0045569620253165,
+      "grad_norm": 0.6605002089221907,
+      "learning_rate": 3.8754724678359884e-05,
+      "loss": 0.5537,
+      "step": 248
+    },
+    {
+      "epoch": 1.0086075949367088,
+      "grad_norm": 0.6234034791350893,
+      "learning_rate": 3.873493415863256e-05,
+      "loss": 0.5746,
+      "step": 249
+    },
+    {
+      "epoch": 1.0126582278481013,
+      "grad_norm": 0.5849005813538919,
+      "learning_rate": 3.871499275032111e-05,
+      "loss": 0.6032,
+      "step": 250
+    },
+    {
+      "epoch": 1.0167088607594936,
+      "grad_norm": 0.4932810631516911,
+      "learning_rate": 3.869490061403091e-05,
+      "loss": 0.5817,
+      "step": 251
+    },
+    {
+      "epoch": 1.0207594936708861,
+      "grad_norm": 0.4338601103145693,
+      "learning_rate": 3.867465791158124e-05,
+      "loss": 0.5795,
+      "step": 252
+    },
+    {
+      "epoch": 1.0248101265822784,
+      "grad_norm": 0.4253392762937026,
+      "learning_rate": 3.865426480600407e-05,
+      "loss": 0.5664,
+      "step": 253
+    },
+    {
+      "epoch": 1.028860759493671,
+      "grad_norm": 0.537037255979096,
+      "learning_rate": 3.863372146154264e-05,
+      "loss": 0.5761,
+      "step": 254
+    },
+    {
+      "epoch": 1.0329113924050632,
+      "grad_norm": 0.45419331912812816,
+      "learning_rate": 3.861302804365024e-05,
+      "loss": 0.5939,
+      "step": 255
+    },
+    {
+      "epoch": 1.0369620253164558,
+      "grad_norm": 0.509756467951194,
+      "learning_rate": 3.85921847189888e-05,
+      "loss": 0.5764,
+      "step": 256
+    },
+    {
+      "epoch": 1.041012658227848,
+      "grad_norm": 0.6138751525210029,
+      "learning_rate": 3.85711916554276e-05,
+      "loss": 0.57,
+      "step": 257
+    },
+    {
+      "epoch": 1.0450632911392406,
+      "grad_norm": 0.5407105553508496,
+      "learning_rate": 3.85500490220419e-05,
+      "loss": 0.5667,
+      "step": 258
+    },
+    {
+      "epoch": 1.0491139240506329,
+      "grad_norm": 0.5159426225531744,
+      "learning_rate": 3.852875698911154e-05,
+      "loss": 0.5695,
+      "step": 259
+    },
+    {
+      "epoch": 1.0531645569620254,
+      "grad_norm": 0.4471848859292597,
+      "learning_rate": 3.850731572811963e-05,
+      "loss": 0.5727,
+      "step": 260
+    },
+    {
+      "epoch": 1.0572151898734177,
+      "grad_norm": 0.4011658557149292,
+      "learning_rate": 3.848572541175116e-05,
+      "loss": 0.5741,
+      "step": 261
+    },
+    {
+      "epoch": 1.0612658227848102,
+      "grad_norm": 0.5193793657243576,
+      "learning_rate": 3.846398621389154e-05,
+      "loss": 0.5722,
+      "step": 262
+    },
+    {
+      "epoch": 1.0653164556962025,
+      "grad_norm": 0.4781546711939762,
+      "learning_rate": 3.84420983096253e-05,
+      "loss": 0.5723,
+      "step": 263
+    },
+    {
+      "epoch": 1.069367088607595,
+      "grad_norm": 0.37208750156890275,
+      "learning_rate": 3.8420061875234606e-05,
+      "loss": 0.5782,
+      "step": 264
+    },
+    {
+      "epoch": 1.0734177215189873,
+      "grad_norm": 0.47371454390146545,
+      "learning_rate": 3.839787708819787e-05,
+      "loss": 0.5604,
+      "step": 265
+    },
+    {
+      "epoch": 1.0774683544303798,
+      "grad_norm": 0.5350098116054309,
+      "learning_rate": 3.8375544127188325e-05,
+      "loss": 0.5869,
+      "step": 266
+    },
+    {
+      "epoch": 1.081518987341772,
+      "grad_norm": 0.5547302533152668,
+      "learning_rate": 3.8353063172072564e-05,
+      "loss": 0.5717,
+      "step": 267
+    },
+    {
+      "epoch": 1.0855696202531646,
+      "grad_norm": 0.5415052099500521,
+      "learning_rate": 3.8330434403909105e-05,
+      "loss": 0.5683,
+      "step": 268
+    },
+    {
+      "epoch": 1.089620253164557,
+      "grad_norm": 0.5400680954057001,
+      "learning_rate": 3.8307658004946934e-05,
+      "loss": 0.5761,
+      "step": 269
+    },
+    {
+      "epoch": 1.0936708860759494,
+      "grad_norm": 0.5507342085011718,
+      "learning_rate": 3.8284734158624046e-05,
+      "loss": 0.5742,
+      "step": 270
+    },
+    {
+      "epoch": 1.0977215189873417,
+      "grad_norm": 0.5883042476215907,
+      "learning_rate": 3.826166304956594e-05,
+      "loss": 0.5592,
+      "step": 271
+    },
+    {
+      "epoch": 1.1017721518987342,
+      "grad_norm": 0.5781462611068698,
+      "learning_rate": 3.8238444863584164e-05,
+      "loss": 0.5722,
+      "step": 272
+    },
+    {
+      "epoch": 1.1058227848101265,
+      "grad_norm": 0.4477064708482778,
+      "learning_rate": 3.821507978767479e-05,
+      "loss": 0.5656,
+      "step": 273
+    },
+    {
+      "epoch": 1.109873417721519,
+      "grad_norm": 0.5259267911467675,
+      "learning_rate": 3.819156801001693e-05,
+      "loss": 0.5784,
+      "step": 274
+    },
+    {
+      "epoch": 1.1139240506329113,
+      "grad_norm": 0.6095760328564332,
+      "learning_rate": 3.816790971997121e-05,
+      "loss": 0.5734,
+      "step": 275
+    },
+    {
+      "epoch": 1.1179746835443038,
+      "grad_norm": 0.587690409847593,
+      "learning_rate": 3.8144105108078246e-05,
+      "loss": 0.5798,
+      "step": 276
+    },
+    {
+      "epoch": 1.1220253164556961,
+      "grad_norm": 0.7130312965436388,
+      "learning_rate": 3.81201543660571e-05,
+      "loss": 0.5728,
+      "step": 277
+    },
+    {
+      "epoch": 1.1260759493670887,
+      "grad_norm": 0.63746954300719,
+      "learning_rate": 3.809605768680377e-05,
+      "loss": 0.5524,
+      "step": 278
+    },
+    {
+      "epoch": 1.130126582278481,
+      "grad_norm": 0.4000850523746339,
+      "learning_rate": 3.807181526438958e-05,
+      "loss": 0.5667,
+      "step": 279
+    },
+    {
+      "epoch": 1.1341772151898735,
+      "grad_norm": 0.5591827629516113,
+      "learning_rate": 3.8047427294059697e-05,
+      "loss": 0.5759,
+      "step": 280
+    },
+    {
+      "epoch": 1.1382278481012658,
+      "grad_norm": 0.4094661887700723,
+      "learning_rate": 3.802289397223145e-05,
+      "loss": 0.5722,
+      "step": 281
+    },
+    {
+      "epoch": 1.1422784810126583,
+      "grad_norm": 0.4572110949392109,
+      "learning_rate": 3.7998215496492854e-05,
+      "loss": 0.5615,
+      "step": 282
+    },
+    {
+      "epoch": 1.1463291139240506,
+      "grad_norm": 0.3977633995322179,
+      "learning_rate": 3.797339206560096e-05,
+      "loss": 0.5458,
+      "step": 283
+    },
+    {
+      "epoch": 1.150379746835443,
+      "grad_norm": 0.38138664330242805,
+      "learning_rate": 3.794842387948027e-05,
+      "loss": 0.5643,
+      "step": 284
+    },
+    {
+      "epoch": 1.1544303797468354,
+      "grad_norm": 0.4755726347994244,
+      "learning_rate": 3.7923311139221114e-05,
+      "loss": 0.5722,
+      "step": 285
+    },
+    {
+      "epoch": 1.158481012658228,
+      "grad_norm": 0.47709530544757917,
+      "learning_rate": 3.7898054047078054e-05,
+      "loss": 0.5691,
+      "step": 286
+    },
+    {
+      "epoch": 1.1625316455696202,
+      "grad_norm": 0.47762646615491283,
+      "learning_rate": 3.787265280646825e-05,
+      "loss": 0.5658,
+      "step": 287
+    },
+    {
+      "epoch": 1.1665822784810127,
+      "grad_norm": 0.5491659649179814,
+      "learning_rate": 3.7847107621969786e-05,
+      "loss": 0.5631,
+      "step": 288
+    },
+    {
+      "epoch": 1.170632911392405,
+      "grad_norm": 0.48364674564276666,
+      "learning_rate": 3.7821418699320064e-05,
+      "loss": 0.5699,
+      "step": 289
+    },
+    {
+      "epoch": 1.1746835443037975,
+      "grad_norm": 0.4825446593328885,
+      "learning_rate": 3.7795586245414145e-05,
+      "loss": 0.5586,
+      "step": 290
+    },
+    {
+      "epoch": 1.1787341772151898,
+      "grad_norm": 0.7473473671032612,
+      "learning_rate": 3.776961046830306e-05,
+      "loss": 0.5671,
+      "step": 291
+    },
+    {
+      "epoch": 1.1827848101265823,
+      "grad_norm": 1.16868810568441,
+      "learning_rate": 3.774349157719215e-05,
+      "loss": 0.605,
+      "step": 292
+    },
+    {
+      "epoch": 1.1868354430379746,
+      "grad_norm": 0.3980676742987899,
+      "learning_rate": 3.7717229782439365e-05,
+      "loss": 0.5686,
+      "step": 293
+    },
+    {
+      "epoch": 1.1908860759493671,
+      "grad_norm": 0.5728559500543624,
+      "learning_rate": 3.769082529555359e-05,
+      "loss": 0.5578,
+      "step": 294
+    },
+    {
+      "epoch": 1.1949367088607594,
+      "grad_norm": 0.5926597762249155,
+      "learning_rate": 3.766427832919294e-05,
+      "loss": 0.5681,
+      "step": 295
+    },
+    {
+      "epoch": 1.198987341772152,
+      "grad_norm": 0.6686467361225584,
+      "learning_rate": 3.7637589097163024e-05,
+      "loss": 0.5652,
+      "step": 296
+    },
+    {
+      "epoch": 1.2030379746835442,
+      "grad_norm": 0.5179930853027336,
+      "learning_rate": 3.761075781441526e-05,
+      "loss": 0.5779,
+      "step": 297
+    },
+    {
+      "epoch": 1.2070886075949367,
+      "grad_norm": 0.4704171615195804,
+      "learning_rate": 3.75837846970451e-05,
+      "loss": 0.5677,
+      "step": 298
+    },
+    {
+      "epoch": 1.211139240506329,
+      "grad_norm": 0.47367652394594917,
+      "learning_rate": 3.755666996229032e-05,
+      "loss": 0.5601,
+      "step": 299
+    },
+    {
+      "epoch": 1.2151898734177216,
+      "grad_norm": 0.5250742006069392,
+      "learning_rate": 3.752941382852927e-05,
+      "loss": 0.5788,
+      "step": 300
+    },
+    {
+      "epoch": 1.2192405063291138,
+      "grad_norm": 0.486393264220946,
+      "learning_rate": 3.7502016515279115e-05,
+      "loss": 0.5745,
+      "step": 301
+    },
+    {
+      "epoch": 1.2232911392405064,
+      "grad_norm": 0.44920110083486264,
+      "learning_rate": 3.7474478243194043e-05,
+      "loss": 0.5601,
+      "step": 302
+    },
+    {
+      "epoch": 1.2273417721518987,
+      "grad_norm": 0.5688567103231287,
+      "learning_rate": 3.744679923406351e-05,
+      "loss": 0.5709,
+      "step": 303
+    },
+    {
+      "epoch": 1.2313924050632912,
+      "grad_norm": 0.5189976033824729,
+      "learning_rate": 3.741897971081043e-05,
+      "loss": 0.5657,
+      "step": 304
+    },
+    {
+      "epoch": 1.2354430379746835,
+      "grad_norm": 0.45579170672321057,
+      "learning_rate": 3.739101989748946e-05,
+      "loss": 0.5721,
+      "step": 305
+    },
+    {
+      "epoch": 1.239493670886076,
+      "grad_norm": 0.4174350679711346,
+      "learning_rate": 3.7362920019285066e-05,
+      "loss": 0.5678,
+      "step": 306
+    },
+    {
+      "epoch": 1.2435443037974683,
+      "grad_norm": 0.46572327764629745,
+      "learning_rate": 3.73346803025098e-05,
+      "loss": 0.5752,
+      "step": 307
+    },
+    {
+      "epoch": 1.2475949367088608,
+      "grad_norm": 0.39044313127442015,
+      "learning_rate": 3.730630097460247e-05,
+      "loss": 0.5671,
+      "step": 308
+    },
+    {
+      "epoch": 1.251645569620253,
+      "grad_norm": 0.3737396671535663,
+      "learning_rate": 3.727778226412628e-05,
+      "loss": 0.5749,
+      "step": 309
+    },
+    {
+      "epoch": 1.2556962025316456,
+      "grad_norm": 0.3960516711080633,
+      "learning_rate": 3.7249124400767006e-05,
+      "loss": 0.5669,
+      "step": 310
+    },
+    {
+      "epoch": 1.259746835443038,
+      "grad_norm": 0.39892755817814496,
+      "learning_rate": 3.722032761533114e-05,
+      "loss": 0.5625,
+      "step": 311
+    },
+    {
+      "epoch": 1.2637974683544304,
+      "grad_norm": 0.425764277944176,
+      "learning_rate": 3.719139213974403e-05,
+      "loss": 0.5771,
+      "step": 312
+    },
+    {
+      "epoch": 1.2678481012658227,
+      "grad_norm": 0.4620174905570317,
+      "learning_rate": 3.7162318207048006e-05,
+      "loss": 0.564,
+      "step": 313
+    },
+    {
+      "epoch": 1.2718987341772152,
+      "grad_norm": 0.3615238318476161,
+      "learning_rate": 3.713310605140055e-05,
+      "loss": 0.5874,
+      "step": 314
+    },
+    {
+      "epoch": 1.2759493670886077,
+      "grad_norm": 0.43121819661274124,
+      "learning_rate": 3.710375590807233e-05,
+      "loss": 0.5834,
+      "step": 315
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.4070198175108835,
+      "learning_rate": 3.7074268013445365e-05,
+      "loss": 0.5682,
+      "step": 316
+    },
+    {
+      "epoch": 1.2840506329113923,
+      "grad_norm": 0.41573889402046377,
+      "learning_rate": 3.7044642605011114e-05,
+      "loss": 0.5624,
+      "step": 317
+    },
+    {
+      "epoch": 1.2881012658227848,
+      "grad_norm": 0.46242738414496803,
+      "learning_rate": 3.701487992136854e-05,
+      "loss": 0.5769,
+      "step": 318
+    },
+    {
+      "epoch": 1.2921518987341774,
+      "grad_norm": 0.40551619862314386,
+      "learning_rate": 3.69849802022222e-05,
+      "loss": 0.5678,
+      "step": 319
+    },
+    {
+      "epoch": 1.2962025316455696,
+      "grad_norm": 0.5087882589611792,
+      "learning_rate": 3.6954943688380334e-05,
+      "loss": 0.554,
+      "step": 320
+    },
+    {
+      "epoch": 1.300253164556962,
+      "grad_norm": 0.40736883808966173,
+      "learning_rate": 3.692477062175289e-05,
+      "loss": 0.5647,
+      "step": 321
+    },
+    {
+      "epoch": 1.3043037974683545,
+      "grad_norm": 0.3903432896112121,
+      "learning_rate": 3.689446124534958e-05,
+      "loss": 0.5631,
+      "step": 322
+    },
+    {
+      "epoch": 1.308354430379747,
+      "grad_norm": 0.44043526741993333,
+      "learning_rate": 3.686401580327799e-05,
+      "loss": 0.567,
+      "step": 323
+    },
+    {
+      "epoch": 1.3124050632911393,
+      "grad_norm": 0.42958677501686693,
+      "learning_rate": 3.683343454074149e-05,
+      "loss": 0.5691,
+      "step": 324
+    },
+    {
+      "epoch": 1.3164556962025316,
+      "grad_norm": 0.49845393125305054,
+      "learning_rate": 3.6802717704037386e-05,
+      "loss": 0.5454,
+      "step": 325
+    },
+    {
+      "epoch": 1.320506329113924,
+      "grad_norm": 0.5360290095283241,
+      "learning_rate": 3.6771865540554855e-05,
+      "loss": 0.5623,
+      "step": 326
+    },
+    {
+      "epoch": 1.3245569620253166,
+      "grad_norm": 0.5173417831116621,
+      "learning_rate": 3.674087829877297e-05,
+      "loss": 0.5579,
+      "step": 327
+    },
+    {
+      "epoch": 1.3286075949367089,
+      "grad_norm": 0.41216732183872273,
+      "learning_rate": 3.6709756228258735e-05,
+      "loss": 0.561,
+      "step": 328
+    },
+    {
+      "epoch": 1.3326582278481012,
+      "grad_norm": 0.35849627783984206,
+      "learning_rate": 3.667849957966501e-05,
+      "loss": 0.5701,
+      "step": 329
+    },
+    {
+      "epoch": 1.3367088607594937,
+      "grad_norm": 0.4483147750717391,
+      "learning_rate": 3.6647108604728546e-05,
+      "loss": 0.5709,
+      "step": 330
+    },
+    {
+      "epoch": 1.3407594936708862,
+      "grad_norm": 0.4402365244265339,
+      "learning_rate": 3.661558355626795e-05,
+      "loss": 0.5785,
+      "step": 331
+    },
+    {
+      "epoch": 1.3448101265822785,
+      "grad_norm": 0.3918997439231543,
+      "learning_rate": 3.658392468818163e-05,
+      "loss": 0.5661,
+      "step": 332
+    },
+    {
+      "epoch": 1.3488607594936708,
+      "grad_norm": 0.4122859000332447,
+      "learning_rate": 3.655213225544574e-05,
+      "loss": 0.575,
+      "step": 333
+    },
+    {
+      "epoch": 1.3529113924050633,
+      "grad_norm": 0.41940816042486617,
+      "learning_rate": 3.652020651411218e-05,
+      "loss": 0.5637,
+      "step": 334
+    },
+    {
+      "epoch": 1.3569620253164558,
+      "grad_norm": 0.5026616395886866,
+      "learning_rate": 3.6488147721306474e-05,
+      "loss": 0.5781,
+      "step": 335
+    },
+    {
+      "epoch": 1.3610126582278481,
+      "grad_norm": 0.5578754556343364,
+      "learning_rate": 3.645595613522574e-05,
+      "loss": 0.5682,
+      "step": 336
+    },
+    {
+      "epoch": 1.3650632911392404,
+      "grad_norm": 0.39047685761175327,
+      "learning_rate": 3.642363201513657e-05,
+      "loss": 0.5708,
+      "step": 337
+    },
+    {
+      "epoch": 1.369113924050633,
+      "grad_norm": 0.49272617493238935,
+      "learning_rate": 3.6391175621373006e-05,
+      "loss": 0.5684,
+      "step": 338
+    },
+    {
+      "epoch": 1.3731645569620254,
+      "grad_norm": 0.42420968499289136,
+      "learning_rate": 3.6358587215334355e-05,
+      "loss": 0.5589,
+      "step": 339
+    },
+    {
+      "epoch": 1.3772151898734177,
+      "grad_norm": 0.5133086778003473,
+      "learning_rate": 3.632586705948318e-05,
+      "loss": 0.5556,
+      "step": 340
+    },
+    {
+      "epoch": 1.38126582278481,
+      "grad_norm": 0.48308936899528593,
+      "learning_rate": 3.629301541734311e-05,
+      "loss": 0.5746,
+      "step": 341
+    },
+    {
+      "epoch": 1.3853164556962025,
+      "grad_norm": 0.4255735543993257,
+      "learning_rate": 3.626003255349676e-05,
+      "loss": 0.5712,
+      "step": 342
+    },
+    {
+      "epoch": 1.389367088607595,
+      "grad_norm": 0.35689843513706776,
+      "learning_rate": 3.622691873358357e-05,
+      "loss": 0.5578,
+      "step": 343
+    },
+    {
+      "epoch": 1.3934177215189874,
+      "grad_norm": 0.4167672996764173,
+      "learning_rate": 3.61936742242977e-05,
+      "loss": 0.5666,
+      "step": 344
+    },
+    {
+      "epoch": 1.3974683544303796,
+      "grad_norm": 0.4233487562827185,
+      "learning_rate": 3.6160299293385864e-05,
+      "loss": 0.5437,
+      "step": 345
+    },
+    {
+      "epoch": 1.4015189873417722,
+      "grad_norm": 0.35628853247052517,
+      "learning_rate": 3.612679420964516e-05,
+      "loss": 0.5768,
+      "step": 346
+    },
+    {
+      "epoch": 1.4055696202531647,
+      "grad_norm": 0.41147055865225896,
+      "learning_rate": 3.609315924292092e-05,
+      "loss": 0.5783,
+      "step": 347
+    },
+    {
+      "epoch": 1.409620253164557,
+      "grad_norm": 0.37489715436513704,
+      "learning_rate": 3.6059394664104554e-05,
+      "loss": 0.5766,
+      "step": 348
+    },
+    {
+      "epoch": 1.4136708860759493,
+      "grad_norm": 0.3883672659943733,
+      "learning_rate": 3.602550074513133e-05,
+      "loss": 0.5665,
+      "step": 349
+    },
+    {
+      "epoch": 1.4177215189873418,
+      "grad_norm": 0.3766506532124595,
+      "learning_rate": 3.599147775897822e-05,
+      "loss": 0.5697,
+      "step": 350
+    },
+    {
+      "epoch": 1.4217721518987343,
+      "grad_norm": 0.418985063389108,
+      "learning_rate": 3.595732597966167e-05,
+      "loss": 0.5603,
+      "step": 351
+    },
+    {
+      "epoch": 1.4258227848101266,
+      "grad_norm": 0.4074712259700754,
+      "learning_rate": 3.592304568223542e-05,
+      "loss": 0.5641,
+      "step": 352
+    },
+    {
+      "epoch": 1.4298734177215189,
+      "grad_norm": 0.35433955500736897,
+      "learning_rate": 3.588863714278826e-05,
+      "loss": 0.5608,
+      "step": 353
+    },
+    {
+      "epoch": 1.4339240506329114,
+      "grad_norm": 0.375999825167048,
+      "learning_rate": 3.585410063844186e-05,
+      "loss": 0.5668,
+      "step": 354
+    },
+    {
+      "epoch": 1.437974683544304,
+      "grad_norm": 0.39117775203351396,
+      "learning_rate": 3.581943644734846e-05,
+      "loss": 0.5632,
+      "step": 355
+    },
+    {
+      "epoch": 1.4420253164556962,
+      "grad_norm": 0.47349514564218886,
+      "learning_rate": 3.578464484868869e-05,
+      "loss": 0.5643,
+      "step": 356
+    },
+    {
+      "epoch": 1.4460759493670885,
+      "grad_norm": 0.4130216113696812,
+      "learning_rate": 3.5749726122669316e-05,
+      "loss": 0.5692,
+      "step": 357
+    },
+    {
+      "epoch": 1.450126582278481,
+      "grad_norm": 0.4259102734984914,
+      "learning_rate": 3.5714680550520943e-05,
+      "loss": 0.5684,
+      "step": 358
+    },
+    {
+      "epoch": 1.4541772151898735,
+      "grad_norm": 0.3953318842617152,
+      "learning_rate": 3.5679508414495794e-05,
+      "loss": 0.572,
+      "step": 359
+    },
+    {
+      "epoch": 1.4582278481012658,
+      "grad_norm": 0.41260474237078004,
+      "learning_rate": 3.564420999786543e-05,
+      "loss": 0.5647,
+      "step": 360
+    },
+    {
+      "epoch": 1.4622784810126581,
+      "grad_norm": 0.427749968962693,
+      "learning_rate": 3.560878558491842e-05,
+      "loss": 0.5649,
+      "step": 361
+    },
+    {
+      "epoch": 1.4663291139240506,
+      "grad_norm": 0.48808233789785943,
+      "learning_rate": 3.5573235460958145e-05,
+      "loss": 0.5669,
+      "step": 362
+    },
+    {
+      "epoch": 1.4703797468354431,
+      "grad_norm": 0.3906325327915913,
+      "learning_rate": 3.553755991230039e-05,
+      "loss": 0.5625,
+      "step": 363
+    },
+    {
+      "epoch": 1.4744303797468354,
+      "grad_norm": 0.5791161394783736,
+      "learning_rate": 3.5501759226271144e-05,
+      "loss": 0.575,
+      "step": 364
+    },
+    {
+      "epoch": 1.4784810126582277,
+      "grad_norm": 0.601943797848734,
+      "learning_rate": 3.546583369120419e-05,
+      "loss": 0.5794,
+      "step": 365
+    },
+    {
+      "epoch": 1.4825316455696202,
+      "grad_norm": 0.3669302851551387,
+      "learning_rate": 3.5429783596438864e-05,
+      "loss": 0.5692,
+      "step": 366
+    },
+    {
+      "epoch": 1.4865822784810128,
+      "grad_norm": 0.4519965906665417,
+      "learning_rate": 3.539360923231766e-05,
+      "loss": 0.5547,
+      "step": 367
+    },
+    {
+      "epoch": 1.490632911392405,
+      "grad_norm": 1.4220997747774635,
+      "learning_rate": 3.535731089018394e-05,
+      "loss": 0.5868,
+      "step": 368
+    },
+    {
+      "epoch": 1.4946835443037974,
+      "grad_norm": 0.4458779344462051,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.5705,
+      "step": 369
+    },
+    {
+      "epoch": 1.4987341772151899,
+      "grad_norm": 0.6665337719364763,
+      "learning_rate": 3.528434344224253e-05,
+      "loss": 0.5766,
+      "step": 370
+    },
+    {
+      "epoch": 1.5027848101265824,
+      "grad_norm": 1.7202064711660423,
+      "learning_rate": 3.524767492410464e-05,
+      "loss": 0.6013,
+      "step": 371
+    },
+    {
+      "epoch": 1.5068354430379747,
+      "grad_norm": 0.7437882922553338,
+      "learning_rate": 3.521088360328908e-05,
+      "loss": 0.5669,
+      "step": 372
+    },
+    {
+      "epoch": 1.510886075949367,
+      "grad_norm": 0.8573499979830769,
+      "learning_rate": 3.517396977610811e-05,
+      "loss": 0.5725,
+      "step": 373
+    },
+    {
+      "epoch": 1.5149367088607595,
+      "grad_norm": 0.6512124740362442,
+      "learning_rate": 3.5136933739860595e-05,
+      "loss": 0.5878,
+      "step": 374
+    },
+    {
+      "epoch": 1.518987341772152,
+      "grad_norm": 0.6149121736736269,
+      "learning_rate": 3.509977579282971e-05,
+      "loss": 0.5706,
+      "step": 375
+    },
+    {
+      "epoch": 1.5230379746835443,
+      "grad_norm": 0.6077005899574783,
+      "learning_rate": 3.5062496234280424e-05,
+      "loss": 0.5612,
+      "step": 376
+    },
+    {
+      "epoch": 1.5270886075949366,
+      "grad_norm": 0.5039451259314034,
+      "learning_rate": 3.502509536445719e-05,
+      "loss": 0.5661,
+      "step": 377
+    },
+    {
+      "epoch": 1.531139240506329,
+      "grad_norm": 0.6144754964028479,
+      "learning_rate": 3.498757348458147e-05,
+      "loss": 0.5775,
+      "step": 378
+    },
+    {
+      "epoch": 1.5351898734177216,
+      "grad_norm": 0.5632496523583074,
+      "learning_rate": 3.4949930896849324e-05,
+      "loss": 0.5887,
+      "step": 379
+    },
+    {
+      "epoch": 1.539240506329114,
+      "grad_norm": 0.44432243344902017,
+      "learning_rate": 3.491216790442899e-05,
+      "loss": 0.5712,
+      "step": 380
+    },
+    {
+      "epoch": 1.5432911392405062,
+      "grad_norm": 0.6295874424545577,
+      "learning_rate": 3.487428481145839e-05,
+      "loss": 0.5708,
+      "step": 381
+    },
+    {
+      "epoch": 1.5473417721518987,
+      "grad_norm": 0.5130111832503208,
+      "learning_rate": 3.483628192304278e-05,
+      "loss": 0.5511,
+      "step": 382
+    },
+    {
+      "epoch": 1.5513924050632912,
+      "grad_norm": 0.4125246964221208,
+      "learning_rate": 3.479815954525219e-05,
+      "loss": 0.5516,
+      "step": 383
+    },
+    {
+      "epoch": 1.5554430379746835,
+      "grad_norm": 0.567916753078382,
+      "learning_rate": 3.475991798511899e-05,
+      "loss": 0.5772,
+      "step": 384
+    },
+    {
+      "epoch": 1.5594936708860758,
+      "grad_norm": 0.46411853516799334,
+      "learning_rate": 3.4721557550635464e-05,
+      "loss": 0.5638,
+      "step": 385
+    },
+    {
+      "epoch": 1.5635443037974683,
+      "grad_norm": 0.493282968616619,
+      "learning_rate": 3.468307855075128e-05,
+      "loss": 0.5731,
+      "step": 386
+    },
+    {
+      "epoch": 1.5675949367088609,
+      "grad_norm": 0.3675151978955606,
+      "learning_rate": 3.4644481295371005e-05,
+      "loss": 0.5596,
+      "step": 387
+    },
+    {
+      "epoch": 1.5716455696202531,
+      "grad_norm": 0.428556178725583,
+      "learning_rate": 3.460576609535163e-05,
+      "loss": 0.5657,
+      "step": 388
+    },
+    {
+      "epoch": 1.5756962025316454,
+      "grad_norm": 0.4283416365100608,
+      "learning_rate": 3.456693326250006e-05,
+      "loss": 0.5663,
+      "step": 389
+    },
+    {
+      "epoch": 1.579746835443038,
+      "grad_norm": 0.36125219170450307,
+      "learning_rate": 3.452798310957058e-05,
+      "loss": 0.5548,
+      "step": 390
+    },
+    {
+      "epoch": 1.5837974683544305,
+      "grad_norm": 0.3744441427017394,
+      "learning_rate": 3.4488915950262386e-05,
+      "loss": 0.5672,
+      "step": 391
+    },
+    {
+      "epoch": 1.5878481012658228,
+      "grad_norm": 0.4131466900499676,
+      "learning_rate": 3.4449732099216985e-05,
+      "loss": 0.5689,
+      "step": 392
+    },
+    {
+      "epoch": 1.591898734177215,
+      "grad_norm": 0.8277610725325351,
+      "learning_rate": 3.441043187201574e-05,
+      "loss": 0.5676,
+      "step": 393
+    },
+    {
+      "epoch": 1.5959493670886076,
+      "grad_norm": 0.3534497857572687,
+      "learning_rate": 3.437101558517728e-05,
+      "loss": 0.5653,
+      "step": 394
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.45743694590094824,
+      "learning_rate": 3.433148355615496e-05,
+      "loss": 0.5682,
+      "step": 395
+    },
+    {
+      "epoch": 1.6040506329113924,
+      "grad_norm": 0.38875415550741277,
+      "learning_rate": 3.4291836103334294e-05,
+      "loss": 0.57,
+      "step": 396
+    },
+    {
+      "epoch": 1.6081012658227847,
+      "grad_norm": 0.37892924806115186,
+      "learning_rate": 3.425207354603043e-05,
+      "loss": 0.5677,
+      "step": 397
+    },
+    {
+      "epoch": 1.6121518987341772,
+      "grad_norm": 0.3375834352522977,
+      "learning_rate": 3.421219620448553e-05,
+      "loss": 0.5624,
+      "step": 398
+    },
+    {
+      "epoch": 1.6162025316455697,
+      "grad_norm": 0.3111736709677703,
+      "learning_rate": 3.417220439986623e-05,
+      "loss": 0.5599,
+      "step": 399
+    },
+    {
+      "epoch": 1.620253164556962,
+      "grad_norm": 0.4089967577016983,
+      "learning_rate": 3.4132098454261024e-05,
+      "loss": 0.5641,
+      "step": 400
+    },
+    {
+      "epoch": 1.6243037974683543,
+      "grad_norm": 0.3531256468097575,
+      "learning_rate": 3.4091878690677676e-05,
+      "loss": 0.5597,
+      "step": 401
+    },
+    {
+      "epoch": 1.6283544303797468,
+      "grad_norm": 0.37800094736598294,
+      "learning_rate": 3.405154543304065e-05,
+      "loss": 0.5632,
+      "step": 402
+    },
+    {
+      "epoch": 1.6324050632911393,
+      "grad_norm": 1.3418278334370821,
+      "learning_rate": 3.401109900618843e-05,
+      "loss": 0.5602,
+      "step": 403
+    },
+    {
+      "epoch": 1.6364556962025316,
+      "grad_norm": 1.0354847920633978,
+      "learning_rate": 3.3970539735870996e-05,
+      "loss": 0.5609,
+      "step": 404
+    },
+    {
+      "epoch": 1.640506329113924,
+      "grad_norm": 0.3898421925828762,
+      "learning_rate": 3.392986794874714e-05,
+      "loss": 0.5678,
+      "step": 405
+    },
+    {
+      "epoch": 1.6445569620253164,
+      "grad_norm": 0.31117064440139647,
+      "learning_rate": 3.388908397238184e-05,
+      "loss": 0.5663,
+      "step": 406
+    },
+    {
+      "epoch": 1.648607594936709,
+      "grad_norm": 0.45148559066248106,
+      "learning_rate": 3.384818813524362e-05,
+      "loss": 0.5796,
+      "step": 407
+    },
+    {
+      "epoch": 1.6526582278481012,
+      "grad_norm": 9.929239173599935,
+      "learning_rate": 3.380718076670195e-05,
+      "loss": 0.5856,
+      "step": 408
+    },
+    {
+      "epoch": 1.6567088607594935,
+      "grad_norm": 0.7599134892418711,
+      "learning_rate": 3.376606219702454e-05,
+      "loss": 0.5638,
+      "step": 409
+    },
+    {
+      "epoch": 1.660759493670886,
+      "grad_norm": 1.2693461098304806,
+      "learning_rate": 3.372483275737468e-05,
+      "loss": 0.5666,
+      "step": 410
+    },
+    {
+      "epoch": 1.6648101265822786,
+      "grad_norm": 0.6540662756166754,
+      "learning_rate": 3.368349277980861e-05,
+      "loss": 0.5708,
+      "step": 411
+    },
+    {
+      "epoch": 1.6688607594936709,
+      "grad_norm": 1.0107411618989088,
+      "learning_rate": 3.3642042597272844e-05,
+      "loss": 0.5844,
+      "step": 412
+    },
+    {
+      "epoch": 1.6729113924050631,
+      "grad_norm": 0.6271521401798261,
+      "learning_rate": 3.360048254360144e-05,
+      "loss": 0.562,
+      "step": 413
+    },
+    {
+      "epoch": 1.6769620253164557,
+      "grad_norm": 0.6902786165958873,
+      "learning_rate": 3.355881295351336e-05,
+      "loss": 0.5556,
+      "step": 414
+    },
+    {
+      "epoch": 1.6810126582278482,
+      "grad_norm": 0.9193943806853043,
+      "learning_rate": 3.351703416260975e-05,
+      "loss": 0.5778,
+      "step": 415
+    },
+    {
+      "epoch": 1.6850632911392405,
+      "grad_norm": 0.5673546490034937,
+      "learning_rate": 3.347514650737126e-05,
+      "loss": 0.5654,
+      "step": 416
+    },
+    {
+      "epoch": 1.6891139240506328,
+      "grad_norm": 0.7754871824227723,
+      "learning_rate": 3.3433150325155295e-05,
+      "loss": 0.5692,
+      "step": 417
+    },
+    {
+      "epoch": 1.6931645569620253,
+      "grad_norm": 0.43946850026359097,
+      "learning_rate": 3.339104595419334e-05,
+      "loss": 0.578,
+      "step": 418
+    },
+    {
+      "epoch": 1.6972151898734178,
+      "grad_norm": 0.9105484867051511,
+      "learning_rate": 3.3348833733588204e-05,
+      "loss": 0.5762,
+      "step": 419
+    },
+    {
+      "epoch": 1.70126582278481,
+      "grad_norm": 0.5039305978898586,
+      "learning_rate": 3.3306514003311305e-05,
+      "loss": 0.562,
+      "step": 420
+    },
+    {
+      "epoch": 1.7053164556962024,
+      "grad_norm": 0.6674425299636102,
+      "learning_rate": 3.326408710419996e-05,
+      "loss": 0.5489,
+      "step": 421
+    },
+    {
+      "epoch": 1.709367088607595,
+      "grad_norm": 0.6767802802870119,
+      "learning_rate": 3.322155337795454e-05,
+      "loss": 0.5704,
+      "step": 422
+    },
+    {
+      "epoch": 1.7134177215189874,
+      "grad_norm": 0.5028137852709147,
+      "learning_rate": 3.317891316713587e-05,
+      "loss": 0.5783,
+      "step": 423
+    },
+    {
+      "epoch": 1.7174683544303797,
+      "grad_norm": 0.6848301668319898,
+      "learning_rate": 3.313616681516231e-05,
+      "loss": 0.5698,
+      "step": 424
+    },
+    {
+      "epoch": 1.721518987341772,
+      "grad_norm": 0.4659968545857974,
+      "learning_rate": 3.309331466630713e-05,
+      "loss": 0.5692,
+      "step": 425
+    },
+    {
+      "epoch": 1.7255696202531645,
+      "grad_norm": 0.5709105341502976,
+      "learning_rate": 3.305035706569563e-05,
+      "loss": 0.5648,
+      "step": 426
+    },
+    {
+      "epoch": 1.729620253164557,
+      "grad_norm": 0.5061048569077146,
+      "learning_rate": 3.3007294359302433e-05,
+      "loss": 0.5621,
+      "step": 427
+    },
+    {
+      "epoch": 1.7336708860759493,
+      "grad_norm": 0.5206105285594833,
+      "learning_rate": 3.296412689394864e-05,
+      "loss": 0.5759,
+      "step": 428
+    },
+    {
+      "epoch": 1.7377215189873416,
+      "grad_norm": 0.49326542508378257,
+      "learning_rate": 3.292085501729909e-05,
+      "loss": 0.5618,
+      "step": 429
+    },
+    {
+      "epoch": 1.7417721518987341,
+      "grad_norm": 0.45801425934510104,
+      "learning_rate": 3.2877479077859534e-05,
+      "loss": 0.5618,
+      "step": 430
+    },
+    {
+      "epoch": 1.7458227848101266,
+      "grad_norm": 0.5722503200720526,
+      "learning_rate": 3.283399942497381e-05,
+      "loss": 0.5788,
+      "step": 431
+    },
+    {
+      "epoch": 1.749873417721519,
+      "grad_norm": 0.4163916214901573,
+      "learning_rate": 3.279041640882108e-05,
+      "loss": 0.5508,
+      "step": 432
+    },
+    {
+      "epoch": 1.7539240506329112,
+      "grad_norm": 0.5642132915462726,
+      "learning_rate": 3.2746730380412964e-05,
+      "loss": 0.5749,
+      "step": 433
+    },
+    {
+      "epoch": 1.7579746835443038,
+      "grad_norm": 0.6419555147455109,
+      "learning_rate": 3.2702941691590726e-05,
+      "loss": 0.5601,
+      "step": 434
+    },
+    {
+      "epoch": 1.7620253164556963,
+      "grad_norm": 0.3453087191549191,
+      "learning_rate": 3.265905069502244e-05,
+      "loss": 0.5461,
+      "step": 435
+    },
+    {
+      "epoch": 1.7660759493670886,
+      "grad_norm": 0.4964223602565925,
+      "learning_rate": 3.261505774420016e-05,
+      "loss": 0.5662,
+      "step": 436
+    },
+    {
+      "epoch": 1.7701265822784809,
+      "grad_norm": 0.6363863621018495,
+      "learning_rate": 3.257096319343707e-05,
+      "loss": 0.5777,
+      "step": 437
+    },
+    {
+      "epoch": 1.7741772151898734,
+      "grad_norm": 0.42953342156204893,
+      "learning_rate": 3.2526767397864614e-05,
+      "loss": 0.5727,
+      "step": 438
+    },
+    {
+      "epoch": 1.7782278481012659,
+      "grad_norm": 0.34440645488939453,
+      "learning_rate": 3.248247071342966e-05,
+      "loss": 0.5588,
+      "step": 439
+    },
+    {
+      "epoch": 1.7822784810126582,
+      "grad_norm": 0.39788208583302115,
+      "learning_rate": 3.243807349689161e-05,
+      "loss": 0.5634,
+      "step": 440
+    },
+    {
+      "epoch": 1.7863291139240505,
+      "grad_norm": 0.4254896378512519,
+      "learning_rate": 3.2393576105819544e-05,
+      "loss": 0.5705,
+      "step": 441
+    },
+    {
+      "epoch": 1.790379746835443,
+      "grad_norm": 1.1332624383890137,
+      "learning_rate": 3.2348978898589333e-05,
+      "loss": 0.5714,
+      "step": 442
+    },
+    {
+      "epoch": 1.7944303797468355,
+      "grad_norm": 0.44327791604094874,
+      "learning_rate": 3.230428223438075e-05,
+      "loss": 0.5548,
+      "step": 443
+    },
+    {
+      "epoch": 1.7984810126582278,
+      "grad_norm": 0.405510068003038,
+      "learning_rate": 3.225948647317459e-05,
+      "loss": 0.5627,
+      "step": 444
+    },
+    {
+      "epoch": 1.80253164556962,
+      "grad_norm": 0.3622377742170301,
+      "learning_rate": 3.2214591975749745e-05,
+      "loss": 0.5571,
+      "step": 445
+    },
+    {
+      "epoch": 1.8065822784810126,
+      "grad_norm": 0.4360877847943863,
+      "learning_rate": 3.216959910368034e-05,
+      "loss": 0.5742,
+      "step": 446
+    },
+    {
+      "epoch": 1.8106329113924051,
+      "grad_norm": 0.3673282039804778,
+      "learning_rate": 3.212450821933277e-05,
+      "loss": 0.5584,
+      "step": 447
+    },
+    {
+      "epoch": 1.8146835443037974,
+      "grad_norm": 0.3854936205173935,
+      "learning_rate": 3.207931968586281e-05,
+      "loss": 0.5642,
+      "step": 448
+    },
+    {
+      "epoch": 1.81873417721519,
+      "grad_norm": 0.40515647241291103,
+      "learning_rate": 3.203403386721272e-05,
+      "loss": 0.5768,
+      "step": 449
+    },
+    {
+      "epoch": 1.8227848101265822,
+      "grad_norm": 0.46354680137141885,
+      "learning_rate": 3.1988651128108245e-05,
+      "loss": 0.5631,
+      "step": 450
+    },
+    {
+      "epoch": 1.8268354430379747,
+      "grad_norm": 0.41110804567539866,
+      "learning_rate": 3.194317183405573e-05,
+      "loss": 0.5594,
+      "step": 451
+    },
+    {
+      "epoch": 1.8308860759493673,
+      "grad_norm": 0.3699900257891482,
+      "learning_rate": 3.189759635133914e-05,
+      "loss": 0.5732,
+      "step": 452
+    },
+    {
+      "epoch": 1.8349367088607595,
+      "grad_norm": 0.4208854538069655,
+      "learning_rate": 3.185192504701718e-05,
+      "loss": 0.5545,
+      "step": 453
+    },
+    {
+      "epoch": 1.8389873417721518,
+      "grad_norm": 0.35386286491641755,
+      "learning_rate": 3.1806158288920234e-05,
+      "loss": 0.5748,
+      "step": 454
+    },
+    {
+      "epoch": 1.8430379746835444,
+      "grad_norm": 0.39383619860923175,
+      "learning_rate": 3.1760296445647477e-05,
+      "loss": 0.5748,
+      "step": 455
+    },
+    {
+      "epoch": 1.8470886075949369,
+      "grad_norm": 0.41059681846896523,
+      "learning_rate": 3.1714339886563896e-05,
+      "loss": 0.5572,
+      "step": 456
+    },
+    {
+      "epoch": 1.8511392405063292,
+      "grad_norm": 0.36857323600848035,
+      "learning_rate": 3.166828898179731e-05,
+      "loss": 0.5637,
+      "step": 457
+    },
+    {
+      "epoch": 1.8551898734177215,
+      "grad_norm": 0.40734828580571564,
+      "learning_rate": 3.162214410223536e-05,
+      "loss": 0.5616,
+      "step": 458
+    },
+    {
+      "epoch": 1.859240506329114,
+      "grad_norm": 0.38745108128079525,
+      "learning_rate": 3.157590561952257e-05,
+      "loss": 0.5513,
+      "step": 459
+    },
+    {
+      "epoch": 1.8632911392405065,
+      "grad_norm": 0.42357959128797024,
+      "learning_rate": 3.152957390605732e-05,
+      "loss": 0.5632,
+      "step": 460
+    },
+    {
+      "epoch": 1.8673417721518988,
+      "grad_norm": 0.32386291563780195,
+      "learning_rate": 3.148314933498886e-05,
+      "loss": 0.5633,
+      "step": 461
+    },
+    {
+      "epoch": 1.871392405063291,
+      "grad_norm": 0.406316313438326,
+      "learning_rate": 3.143663228021431e-05,
+      "loss": 0.5688,
+      "step": 462
+    },
+    {
+      "epoch": 1.8754430379746836,
+      "grad_norm": 0.3110960985932442,
+      "learning_rate": 3.1390023116375624e-05,
+      "loss": 0.5619,
+      "step": 463
+    },
+    {
+      "epoch": 1.879493670886076,
+      "grad_norm": 0.3755076052012993,
+      "learning_rate": 3.134332221885661e-05,
+      "loss": 0.5585,
+      "step": 464
+    },
+    {
+      "epoch": 1.8835443037974684,
+      "grad_norm": 0.4513263273171889,
+      "learning_rate": 3.129652996377987e-05,
+      "loss": 0.5624,
+      "step": 465
+    },
+    {
+      "epoch": 1.8875949367088607,
+      "grad_norm": 0.43026977320819215,
+      "learning_rate": 3.12496467280038e-05,
+      "loss": 0.5652,
+      "step": 466
+    },
+    {
+      "epoch": 1.8916455696202532,
+      "grad_norm": 0.3272565995248847,
+      "learning_rate": 3.120267288911952e-05,
+      "loss": 0.5641,
+      "step": 467
+    },
+    {
+      "epoch": 1.8956962025316457,
+      "grad_norm": 0.3961544778527916,
+      "learning_rate": 3.11556088254479e-05,
+      "loss": 0.5635,
+      "step": 468
+    },
+    {
+      "epoch": 1.899746835443038,
+      "grad_norm": 0.31471908661194753,
+      "learning_rate": 3.11084549160364e-05,
+      "loss": 0.5577,
+      "step": 469
+    },
+    {
+      "epoch": 1.9037974683544303,
+      "grad_norm": 0.35488153809381234,
+      "learning_rate": 3.106121154065615e-05,
+      "loss": 0.5583,
+      "step": 470
+    },
+    {
+      "epoch": 1.9078481012658228,
+      "grad_norm": 0.32559278506846306,
+      "learning_rate": 3.1013879079798805e-05,
+      "loss": 0.5762,
+      "step": 471
+    },
+    {
+      "epoch": 1.9118987341772153,
+      "grad_norm": 0.38906847640101616,
+      "learning_rate": 3.096645791467348e-05,
+      "loss": 0.5624,
+      "step": 472
+    },
+    {
+      "epoch": 1.9159493670886076,
+      "grad_norm": 0.3138808457934149,
+      "learning_rate": 3.091894842720373e-05,
+      "loss": 0.545,
+      "step": 473
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.377509150601867,
+      "learning_rate": 3.0871351000024425e-05,
+      "loss": 0.5625,
+      "step": 474
+    },
+    {
+      "epoch": 1.9240506329113924,
+      "grad_norm": 0.3792364707998045,
+      "learning_rate": 3.0823666016478716e-05,
+      "loss": 0.5504,
+      "step": 475
+    },
+    {
+      "epoch": 1.928101265822785,
+      "grad_norm": 0.3402547738791823,
+      "learning_rate": 3.0775893860614896e-05,
+      "loss": 0.5602,
+      "step": 476
+    },
+    {
+      "epoch": 1.9321518987341773,
+      "grad_norm": 0.3418504249356052,
+      "learning_rate": 3.0728034917183336e-05,
+      "loss": 0.5624,
+      "step": 477
+    },
+    {
+      "epoch": 1.9362025316455695,
+      "grad_norm": 0.30273325950136903,
+      "learning_rate": 3.06800895716334e-05,
+      "loss": 0.5652,
+      "step": 478
+    },
+    {
+      "epoch": 1.940253164556962,
+      "grad_norm": 0.35204304073192977,
+      "learning_rate": 3.063205821011029e-05,
+      "loss": 0.556,
+      "step": 479
+    },
+    {
+      "epoch": 1.9443037974683546,
+      "grad_norm": 0.3260604284008201,
+      "learning_rate": 3.0583941219452016e-05,
+      "loss": 0.5645,
+      "step": 480
+    },
+    {
+      "epoch": 1.9483544303797469,
+      "grad_norm": 0.36420259487699597,
+      "learning_rate": 3.053573898718618e-05,
+      "loss": 0.5563,
+      "step": 481
+    },
+    {
+      "epoch": 1.9524050632911392,
+      "grad_norm": 0.32383675149273644,
+      "learning_rate": 3.0487451901526956e-05,
+      "loss": 0.5568,
+      "step": 482
+    },
+    {
+      "epoch": 1.9564556962025317,
+      "grad_norm": 0.3565153071252402,
+      "learning_rate": 3.0439080351371875e-05,
+      "loss": 0.5683,
+      "step": 483
+    },
+    {
+      "epoch": 1.9605063291139242,
+      "grad_norm": 0.3401199474578724,
+      "learning_rate": 3.0390624726298764e-05,
+      "loss": 0.5633,
+      "step": 484
+    },
+    {
+      "epoch": 1.9645569620253165,
+      "grad_norm": 1.0751224474860739,
+      "learning_rate": 3.034208541656255e-05,
+      "loss": 0.5711,
+      "step": 485
+    },
+    {
+      "epoch": 1.9686075949367088,
+      "grad_norm": 0.3427723957749214,
+      "learning_rate": 3.029346281309218e-05,
+      "loss": 0.5702,
+      "step": 486
+    },
+    {
+      "epoch": 1.9726582278481013,
+      "grad_norm": 0.38715388582590093,
+      "learning_rate": 3.0244757307487415e-05,
+      "loss": 0.5691,
+      "step": 487
+    },
+    {
+      "epoch": 1.9767088607594938,
+      "grad_norm": 0.3746274075845667,
+      "learning_rate": 3.019596929201569e-05,
+      "loss": 0.5571,
+      "step": 488
+    },
+    {
+      "epoch": 1.980759493670886,
+      "grad_norm": 0.3487876234068379,
+      "learning_rate": 3.0147099159608985e-05,
+      "loss": 0.5804,
+      "step": 489
+    },
+    {
+      "epoch": 1.9848101265822784,
+      "grad_norm": 0.4253339527400132,
+      "learning_rate": 3.0098147303860616e-05,
+      "loss": 0.576,
+      "step": 490
+    },
+    {
+      "epoch": 1.988860759493671,
+      "grad_norm": 0.38864391456369235,
+      "learning_rate": 3.0049114119022117e-05,
+      "loss": 0.5598,
+      "step": 491
+    },
+    {
+      "epoch": 1.9929113924050634,
+      "grad_norm": 0.3738113521257081,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5579,
+      "step": 492
+    },
+    {
+      "epoch": 1.9969620253164557,
+      "grad_norm": 0.3409590855141507,
+      "learning_rate": 2.995080534235264e-05,
+      "loss": 0.5663,
+      "step": 493
+    },
+    {
+      "epoch": 2.001012658227848,
+      "grad_norm": 0.37903392383031337,
+      "learning_rate": 2.9901530542287044e-05,
+      "loss": 0.546,
+      "step": 494
+    },
+    {
+      "epoch": 2.0050632911392405,
+      "grad_norm": 0.566614106852766,
+      "learning_rate": 2.9852175996655676e-05,
+      "loss": 0.505,
+      "step": 495
+    },
+    {
+      "epoch": 2.009113924050633,
+      "grad_norm": 0.4480382958409841,
+      "learning_rate": 2.980274210295326e-05,
+      "loss": 0.5009,
+      "step": 496
+    },
+    {
+      "epoch": 2.013164556962025,
+      "grad_norm": 0.9221116698304787,
+      "learning_rate": 2.9753229259313578e-05,
+      "loss": 0.4923,
+      "step": 497
+    },
+    {
+      "epoch": 2.0172151898734176,
+      "grad_norm": 0.4303852719369365,
+      "learning_rate": 2.9703637864506274e-05,
+      "loss": 0.5005,
+      "step": 498
+    },
+    {
+      "epoch": 2.02126582278481,
+      "grad_norm": 0.5420582775593321,
+      "learning_rate": 2.965396831793362e-05,
+      "loss": 0.4972,
+      "step": 499
+    },
+    {
+      "epoch": 2.0253164556962027,
+      "grad_norm": 0.420852190783598,
+      "learning_rate": 2.9604221019627316e-05,
+      "loss": 0.5121,
+      "step": 500
+    },
+    {
+      "epoch": 2.0293670886075947,
+      "grad_norm": 0.5293031894054138,
+      "learning_rate": 2.955439637024526e-05,
+      "loss": 0.5059,
+      "step": 501
+    },
+    {
+      "epoch": 2.0334177215189873,
+      "grad_norm": 0.623334063000591,
+      "learning_rate": 2.9504494771068334e-05,
+      "loss": 0.4974,
+      "step": 502
+    },
+    {
+      "epoch": 2.0374683544303798,
+      "grad_norm": 0.4463119488075968,
+      "learning_rate": 2.9454516623997156e-05,
+      "loss": 0.4951,
+      "step": 503
+    },
+    {
+      "epoch": 2.0415189873417723,
+      "grad_norm": 0.5516435364410915,
+      "learning_rate": 2.9404462331548847e-05,
+      "loss": 0.4989,
+      "step": 504
+    },
+    {
+      "epoch": 2.0455696202531644,
+      "grad_norm": 0.497102336495903,
+      "learning_rate": 2.93543322968538e-05,
+      "loss": 0.4925,
+      "step": 505
+    },
+    {
+      "epoch": 2.049620253164557,
+      "grad_norm": 0.3838607996683134,
+      "learning_rate": 2.9304126923652428e-05,
+      "loss": 0.483,
+      "step": 506
+    },
+    {
+      "epoch": 2.0536708860759494,
+      "grad_norm": 0.44141992011761844,
+      "learning_rate": 2.9253846616291896e-05,
+      "loss": 0.4894,
+      "step": 507
+    },
+    {
+      "epoch": 2.057721518987342,
+      "grad_norm": 0.35115091963097866,
+      "learning_rate": 2.9203491779722896e-05,
+      "loss": 0.4759,
+      "step": 508
+    },
+    {
+      "epoch": 2.061772151898734,
+      "grad_norm": 0.449132840990915,
+      "learning_rate": 2.9153062819496357e-05,
+      "loss": 0.499,
+      "step": 509
+    },
+    {
+      "epoch": 2.0658227848101265,
+      "grad_norm": 0.3489185330585189,
+      "learning_rate": 2.9102560141760178e-05,
+      "loss": 0.4923,
+      "step": 510
+    },
+    {
+      "epoch": 2.069873417721519,
+      "grad_norm": 0.4106106791281783,
+      "learning_rate": 2.9051984153256004e-05,
+      "loss": 0.4993,
+      "step": 511
+    },
+    {
+      "epoch": 2.0739240506329115,
+      "grad_norm": 0.35159341624029883,
+      "learning_rate": 2.900133526131588e-05,
+      "loss": 0.4944,
+      "step": 512
+    },
+    {
+      "epoch": 2.0779746835443036,
+      "grad_norm": 0.39949005153429706,
+      "learning_rate": 2.8950613873859025e-05,
+      "loss": 0.5128,
+      "step": 513
+    },
+    {
+      "epoch": 2.082025316455696,
+      "grad_norm": 0.3431699090955081,
+      "learning_rate": 2.8899820399388515e-05,
+      "loss": 0.4909,
+      "step": 514
+    },
+    {
+      "epoch": 2.0860759493670886,
+      "grad_norm": 0.3908359269076817,
+      "learning_rate": 2.8848955246988012e-05,
+      "loss": 0.4971,
+      "step": 515
+    },
+    {
+      "epoch": 2.090126582278481,
+      "grad_norm": 0.323332208753449,
+      "learning_rate": 2.879801882631847e-05,
+      "loss": 0.4978,
+      "step": 516
+    },
+    {
+      "epoch": 2.094177215189873,
+      "grad_norm": 0.3994047013716915,
+      "learning_rate": 2.8747011547614808e-05,
+      "loss": 0.5105,
+      "step": 517
+    },
+    {
+      "epoch": 2.0982278481012657,
+      "grad_norm": 0.3380579293694711,
+      "learning_rate": 2.8695933821682635e-05,
+      "loss": 0.4932,
+      "step": 518
+    },
+    {
+      "epoch": 2.1022784810126582,
+      "grad_norm": 0.38228604016855183,
+      "learning_rate": 2.864478605989494e-05,
+      "loss": 0.4861,
+      "step": 519
+    },
+    {
+      "epoch": 2.1063291139240508,
+      "grad_norm": 0.32077032548388,
+      "learning_rate": 2.8593568674188765e-05,
+      "loss": 0.5013,
+      "step": 520
+    },
+    {
+      "epoch": 2.110379746835443,
+      "grad_norm": 0.3132848090533015,
+      "learning_rate": 2.8542282077061892e-05,
+      "loss": 0.4867,
+      "step": 521
+    },
+    {
+      "epoch": 2.1144303797468353,
+      "grad_norm": 0.3558118752907399,
+      "learning_rate": 2.8490926681569523e-05,
+      "loss": 0.4868,
+      "step": 522
+    },
+    {
+      "epoch": 2.118481012658228,
+      "grad_norm": 0.31060271361841985,
+      "learning_rate": 2.8439502901320956e-05,
+      "loss": 0.4946,
+      "step": 523
+    },
+    {
+      "epoch": 2.1225316455696204,
+      "grad_norm": 0.34519962417334543,
+      "learning_rate": 2.8388011150476237e-05,
+      "loss": 0.5019,
+      "step": 524
+    },
+    {
+      "epoch": 2.1265822784810124,
+      "grad_norm": 0.34466736083873073,
+      "learning_rate": 2.8336451843742866e-05,
+      "loss": 0.5078,
+      "step": 525
+    },
+    {
+      "epoch": 2.130632911392405,
+      "grad_norm": 0.35076536348357035,
+      "learning_rate": 2.8284825396372387e-05,
+      "loss": 0.4999,
+      "step": 526
+    },
+    {
+      "epoch": 2.1346835443037975,
+      "grad_norm": 0.36339437274478875,
+      "learning_rate": 2.8233132224157132e-05,
+      "loss": 0.4906,
+      "step": 527
+    },
+    {
+      "epoch": 2.13873417721519,
+      "grad_norm": 0.3426847976608821,
+      "learning_rate": 2.8181372743426805e-05,
+      "loss": 0.505,
+      "step": 528
+    },
+    {
+      "epoch": 2.1427848101265825,
+      "grad_norm": 0.3502227325739476,
+      "learning_rate": 2.8129547371045128e-05,
+      "loss": 0.4833,
+      "step": 529
+    },
+    {
+      "epoch": 2.1468354430379746,
+      "grad_norm": 0.33316347409660435,
+      "learning_rate": 2.8077656524406534e-05,
+      "loss": 0.4936,
+      "step": 530
+    },
+    {
+      "epoch": 2.150886075949367,
+      "grad_norm": 0.3108943245451696,
+      "learning_rate": 2.802570062143278e-05,
+      "loss": 0.4992,
+      "step": 531
+    },
+    {
+      "epoch": 2.1549367088607596,
+      "grad_norm": 0.35430685205459594,
+      "learning_rate": 2.7973680080569555e-05,
+      "loss": 0.5046,
+      "step": 532
+    },
+    {
+      "epoch": 2.1589873417721517,
+      "grad_norm": 0.37907203256632505,
+      "learning_rate": 2.792159532078314e-05,
+      "loss": 0.4964,
+      "step": 533
+    },
+    {
+      "epoch": 2.163037974683544,
+      "grad_norm": 0.31984773001781613,
+      "learning_rate": 2.7869446761557033e-05,
+      "loss": 0.4867,
+      "step": 534
+    },
+    {
+      "epoch": 2.1670886075949367,
+      "grad_norm": 0.3233218233095579,
+      "learning_rate": 2.781723482288857e-05,
+      "loss": 0.4941,
+      "step": 535
+    },
+    {
+      "epoch": 2.1711392405063292,
+      "grad_norm": 0.625218578654047,
+      "learning_rate": 2.7764959925285517e-05,
+      "loss": 0.5036,
+      "step": 536
+    },
+    {
+      "epoch": 2.1751898734177217,
+      "grad_norm": 0.31858695646291735,
+      "learning_rate": 2.771262248976272e-05,
+      "loss": 0.5044,
+      "step": 537
+    },
+    {
+      "epoch": 2.179240506329114,
+      "grad_norm": 0.33472230063276465,
+      "learning_rate": 2.7660222937838677e-05,
+      "loss": 0.4915,
+      "step": 538
+    },
+    {
+      "epoch": 2.1832911392405063,
+      "grad_norm": 0.36831499583408384,
+      "learning_rate": 2.7607761691532186e-05,
+      "loss": 0.4994,
+      "step": 539
+    },
+    {
+      "epoch": 2.187341772151899,
+      "grad_norm": 0.322226962344922,
+      "learning_rate": 2.7555239173358916e-05,
+      "loss": 0.494,
+      "step": 540
+    },
+    {
+      "epoch": 2.191392405063291,
+      "grad_norm": 0.32174842133786397,
+      "learning_rate": 2.7502655806328e-05,
+      "loss": 0.4995,
+      "step": 541
+    },
+    {
+      "epoch": 2.1954430379746834,
+      "grad_norm": 0.31468397054323705,
+      "learning_rate": 2.7450012013938648e-05,
+      "loss": 0.4804,
+      "step": 542
+    },
+    {
+      "epoch": 2.199493670886076,
+      "grad_norm": 0.30630149921413063,
+      "learning_rate": 2.739730822017673e-05,
+      "loss": 0.4996,
+      "step": 543
+    },
+    {
+      "epoch": 2.2035443037974685,
+      "grad_norm": 0.2965062713287259,
+      "learning_rate": 2.7344544849511355e-05,
+      "loss": 0.498,
+      "step": 544
+    },
+    {
+      "epoch": 2.207594936708861,
+      "grad_norm": 0.3628375287335248,
+      "learning_rate": 2.7291722326891456e-05,
+      "loss": 0.5019,
+      "step": 545
+    },
+    {
+      "epoch": 2.211645569620253,
+      "grad_norm": 0.3397142712710508,
+      "learning_rate": 2.723884107774236e-05,
+      "loss": 0.5016,
+      "step": 546
+    },
+    {
+      "epoch": 2.2156962025316456,
+      "grad_norm": 0.3000955907327731,
+      "learning_rate": 2.718590152796239e-05,
+      "loss": 0.5097,
+      "step": 547
+    },
+    {
+      "epoch": 2.219746835443038,
+      "grad_norm": 0.2998923876320044,
+      "learning_rate": 2.71329041039194e-05,
+      "loss": 0.4776,
+      "step": 548
+    },
+    {
+      "epoch": 2.22379746835443,
+      "grad_norm": 0.3218252186599272,
+      "learning_rate": 2.7079849232447357e-05,
+      "loss": 0.4955,
+      "step": 549
+    },
+    {
+      "epoch": 2.2278481012658227,
+      "grad_norm": 0.2831347107953129,
+      "learning_rate": 2.7026737340842895e-05,
+      "loss": 0.5027,
+      "step": 550
+    },
+    {
+      "epoch": 2.231898734177215,
+      "grad_norm": 0.2793355643698957,
+      "learning_rate": 2.697356885686189e-05,
+      "loss": 0.4918,
+      "step": 551
+    },
+    {
+      "epoch": 2.2359493670886077,
+      "grad_norm": 0.3428309852336375,
+      "learning_rate": 2.6920344208716014e-05,
+      "loss": 0.4904,
+      "step": 552
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.281642560102733,
+      "learning_rate": 2.6867063825069252e-05,
+      "loss": 0.4942,
+      "step": 553
+    },
+    {
+      "epoch": 2.2440506329113923,
+      "grad_norm": 0.33668882364455666,
+      "learning_rate": 2.6813728135034494e-05,
+      "loss": 0.5003,
+      "step": 554
+    },
+    {
+      "epoch": 2.248101265822785,
+      "grad_norm": 0.2969347976525531,
+      "learning_rate": 2.6760337568170056e-05,
+      "loss": 0.4931,
+      "step": 555
+    },
+    {
+      "epoch": 2.2521518987341773,
+      "grad_norm": 0.28464319482300915,
+      "learning_rate": 2.6706892554476226e-05,
+      "loss": 0.5037,
+      "step": 556
+    },
+    {
+      "epoch": 2.2562025316455694,
+      "grad_norm": 0.3450143038367752,
+      "learning_rate": 2.6653393524391795e-05,
+      "loss": 0.486,
+      "step": 557
+    },
+    {
+      "epoch": 2.260253164556962,
+      "grad_norm": 0.3383346934828438,
+      "learning_rate": 2.6599840908790592e-05,
+      "loss": 0.4947,
+      "step": 558
+    },
+    {
+      "epoch": 2.2643037974683544,
+      "grad_norm": 0.2914236924504204,
+      "learning_rate": 2.6546235138978028e-05,
+      "loss": 0.4855,
+      "step": 559
+    },
+    {
+      "epoch": 2.268354430379747,
+      "grad_norm": 0.3506707138954212,
+      "learning_rate": 2.6492576646687597e-05,
+      "loss": 0.4971,
+      "step": 560
+    },
+    {
+      "epoch": 2.2724050632911394,
+      "grad_norm": 0.32333801018874886,
+      "learning_rate": 2.6438865864077425e-05,
+      "loss": 0.4972,
+      "step": 561
+    },
+    {
+      "epoch": 2.2764556962025315,
+      "grad_norm": 0.33014071390184907,
+      "learning_rate": 2.6385103223726766e-05,
+      "loss": 0.4905,
+      "step": 562
+    },
+    {
+      "epoch": 2.280506329113924,
+      "grad_norm": 0.2843367050110206,
+      "learning_rate": 2.6331289158632537e-05,
+      "loss": 0.497,
+      "step": 563
+    },
+    {
+      "epoch": 2.2845569620253166,
+      "grad_norm": 0.303002269711914,
+      "learning_rate": 2.6277424102205817e-05,
+      "loss": 0.4829,
+      "step": 564
+    },
+    {
+      "epoch": 2.2886075949367086,
+      "grad_norm": 0.2849399396375864,
+      "learning_rate": 2.6223508488268374e-05,
+      "loss": 0.4935,
+      "step": 565
+    },
+    {
+      "epoch": 2.292658227848101,
+      "grad_norm": 0.30839850794329093,
+      "learning_rate": 2.6169542751049148e-05,
+      "loss": 0.5006,
+      "step": 566
+    },
+    {
+      "epoch": 2.2967088607594937,
+      "grad_norm": 0.2878545017015629,
+      "learning_rate": 2.6115527325180754e-05,
+      "loss": 0.4893,
+      "step": 567
+    },
+    {
+      "epoch": 2.300759493670886,
+      "grad_norm": 0.27977056857275334,
+      "learning_rate": 2.606146264569603e-05,
+      "loss": 0.4958,
+      "step": 568
+    },
+    {
+      "epoch": 2.3048101265822787,
+      "grad_norm": 0.3681143341053712,
+      "learning_rate": 2.6007349148024447e-05,
+      "loss": 0.4958,
+      "step": 569
+    },
+    {
+      "epoch": 2.3088607594936708,
+      "grad_norm": 0.30467032674400957,
+      "learning_rate": 2.5953187267988694e-05,
+      "loss": 0.488,
+      "step": 570
+    },
+    {
+      "epoch": 2.3129113924050633,
+      "grad_norm": 0.3921325859705292,
+      "learning_rate": 2.5898977441801097e-05,
+      "loss": 0.4906,
+      "step": 571
+    },
+    {
+      "epoch": 2.316962025316456,
+      "grad_norm": 0.3067934407168898,
+      "learning_rate": 2.584472010606015e-05,
+      "loss": 0.4954,
+      "step": 572
+    },
+    {
+      "epoch": 2.321012658227848,
+      "grad_norm": 0.3028492677984733,
+      "learning_rate": 2.5790415697746976e-05,
+      "loss": 0.4934,
+      "step": 573
+    },
+    {
+      "epoch": 2.3250632911392404,
+      "grad_norm": 0.28418995875989905,
+      "learning_rate": 2.5736064654221808e-05,
+      "loss": 0.4834,
+      "step": 574
+    },
+    {
+      "epoch": 2.329113924050633,
+      "grad_norm": 0.2961892092387888,
+      "learning_rate": 2.568166741322048e-05,
+      "loss": 0.5012,
+      "step": 575
+    },
+    {
+      "epoch": 2.3331645569620254,
+      "grad_norm": 0.37583692065967395,
+      "learning_rate": 2.56272244128509e-05,
+      "loss": 0.5004,
+      "step": 576
+    },
+    {
+      "epoch": 2.337215189873418,
+      "grad_norm": 0.36617115225522007,
+      "learning_rate": 2.55727360915895e-05,
+      "loss": 0.4823,
+      "step": 577
+    },
+    {
+      "epoch": 2.34126582278481,
+      "grad_norm": 0.3492081022740126,
+      "learning_rate": 2.5518202888277734e-05,
+      "loss": 0.5029,
+      "step": 578
+    },
+    {
+      "epoch": 2.3453164556962025,
+      "grad_norm": 0.429300703098307,
+      "learning_rate": 2.5463625242118523e-05,
+      "loss": 0.511,
+      "step": 579
+    },
+    {
+      "epoch": 2.349367088607595,
+      "grad_norm": 0.37578646104701335,
+      "learning_rate": 2.5409003592672723e-05,
+      "loss": 0.4933,
+      "step": 580
+    },
+    {
+      "epoch": 2.353417721518987,
+      "grad_norm": 0.3197186316714807,
+      "learning_rate": 2.535433837985559e-05,
+      "loss": 0.4984,
+      "step": 581
+    },
+    {
+      "epoch": 2.3574683544303796,
+      "grad_norm": 0.3520761388494902,
+      "learning_rate": 2.529963004393324e-05,
+      "loss": 0.4984,
+      "step": 582
+    },
+    {
+      "epoch": 2.361518987341772,
+      "grad_norm": 0.2824410140009543,
+      "learning_rate": 2.524487902551908e-05,
+      "loss": 0.4968,
+      "step": 583
+    },
+    {
+      "epoch": 2.3655696202531646,
+      "grad_norm": 0.3604476428789943,
+      "learning_rate": 2.519008576557029e-05,
+      "loss": 0.5004,
+      "step": 584
+    },
+    {
+      "epoch": 2.369620253164557,
+      "grad_norm": 0.26178738349061753,
+      "learning_rate": 2.5135250705384254e-05,
+      "loss": 0.4808,
+      "step": 585
+    },
+    {
+      "epoch": 2.3736708860759492,
+      "grad_norm": 0.30916475997832304,
+      "learning_rate": 2.5080374286595007e-05,
+      "loss": 0.4923,
+      "step": 586
+    },
+    {
+      "epoch": 2.3777215189873417,
+      "grad_norm": 0.32249578778988425,
+      "learning_rate": 2.5025456951169677e-05,
+      "loss": 0.4942,
+      "step": 587
+    },
+    {
+      "epoch": 2.3817721518987343,
+      "grad_norm": 0.30464721114884125,
+      "learning_rate": 2.4970499141404942e-05,
+      "loss": 0.5033,
+      "step": 588
+    },
+    {
+      "epoch": 2.3858227848101268,
+      "grad_norm": 0.39402383971346494,
+      "learning_rate": 2.491550129992345e-05,
+      "loss": 0.4944,
+      "step": 589
+    },
+    {
+      "epoch": 2.389873417721519,
+      "grad_norm": 0.31006242281201546,
+      "learning_rate": 2.486046386967024e-05,
+      "loss": 0.4964,
+      "step": 590
+    },
+    {
+      "epoch": 2.3939240506329114,
+      "grad_norm": 0.32941342515725297,
+      "learning_rate": 2.4805387293909214e-05,
+      "loss": 0.5001,
+      "step": 591
+    },
+    {
+      "epoch": 2.397974683544304,
+      "grad_norm": 0.3358395319597829,
+      "learning_rate": 2.4750272016219552e-05,
+      "loss": 0.5026,
+      "step": 592
+    },
+    {
+      "epoch": 2.4020253164556964,
+      "grad_norm": 0.36159287143627306,
+      "learning_rate": 2.4695118480492114e-05,
+      "loss": 0.5088,
+      "step": 593
+    },
+    {
+      "epoch": 2.4060759493670885,
+      "grad_norm": 0.3777420930423377,
+      "learning_rate": 2.4639927130925898e-05,
+      "loss": 0.5128,
+      "step": 594
+    },
+    {
+      "epoch": 2.410126582278481,
+      "grad_norm": 0.34182826600308713,
+      "learning_rate": 2.458469841202444e-05,
+      "loss": 0.4893,
+      "step": 595
+    },
+    {
+      "epoch": 2.4141772151898735,
+      "grad_norm": 0.2873175109597314,
+      "learning_rate": 2.452943276859226e-05,
+      "loss": 0.5006,
+      "step": 596
+    },
+    {
+      "epoch": 2.418227848101266,
+      "grad_norm": 0.3659444933815534,
+      "learning_rate": 2.447413064573125e-05,
+      "loss": 0.497,
+      "step": 597
+    },
+    {
+      "epoch": 2.422278481012658,
+      "grad_norm": 0.2608796991139369,
+      "learning_rate": 2.4418792488837095e-05,
+      "loss": 0.4976,
+      "step": 598
+    },
+    {
+      "epoch": 2.4263291139240506,
+      "grad_norm": 0.40961814351827275,
+      "learning_rate": 2.4363418743595713e-05,
+      "loss": 0.4945,
+      "step": 599
+    },
+    {
+      "epoch": 2.430379746835443,
+      "grad_norm": 0.2786196256368595,
+      "learning_rate": 2.430800985597963e-05,
+      "loss": 0.5028,
+      "step": 600
+    },
+    {
+      "epoch": 2.4344303797468356,
+      "grad_norm": 0.4376931688714434,
+      "learning_rate": 2.4252566272244415e-05,
+      "loss": 0.4949,
+      "step": 601
+    },
+    {
+      "epoch": 2.4384810126582277,
+      "grad_norm": 0.37308030657015057,
+      "learning_rate": 2.4197088438925063e-05,
+      "loss": 0.4848,
+      "step": 602
+    },
+    {
+      "epoch": 2.44253164556962,
+      "grad_norm": 0.30672415166592243,
+      "learning_rate": 2.4141576802832417e-05,
+      "loss": 0.496,
+      "step": 603
+    },
+    {
+      "epoch": 2.4465822784810127,
+      "grad_norm": 0.6608819506629366,
+      "learning_rate": 2.408603181104957e-05,
+      "loss": 0.5142,
+      "step": 604
+    },
+    {
+      "epoch": 2.4506329113924052,
+      "grad_norm": 0.30751876593511995,
+      "learning_rate": 2.4030453910928245e-05,
+      "loss": 0.4927,
+      "step": 605
+    },
+    {
+      "epoch": 2.4546835443037973,
+      "grad_norm": 0.35286435709364855,
+      "learning_rate": 2.397484355008521e-05,
+      "loss": 0.4937,
+      "step": 606
+    },
+    {
+      "epoch": 2.45873417721519,
+      "grad_norm": 0.306475468701529,
+      "learning_rate": 2.3919201176398662e-05,
+      "loss": 0.5104,
+      "step": 607
+    },
+    {
+      "epoch": 2.4627848101265823,
+      "grad_norm": 0.29718273550703084,
+      "learning_rate": 2.3863527238004633e-05,
+      "loss": 0.5016,
+      "step": 608
+    },
+    {
+      "epoch": 2.466835443037975,
+      "grad_norm": 0.26409963268164044,
+      "learning_rate": 2.380782218329337e-05,
+      "loss": 0.5037,
+      "step": 609
+    },
+    {
+      "epoch": 2.470886075949367,
+      "grad_norm": 0.2959501649458629,
+      "learning_rate": 2.3752086460905725e-05,
+      "loss": 0.4995,
+      "step": 610
+    },
+    {
+      "epoch": 2.4749367088607594,
+      "grad_norm": 0.27553427378479395,
+      "learning_rate": 2.3696320519729544e-05,
+      "loss": 0.4859,
+      "step": 611
+    },
+    {
+      "epoch": 2.478987341772152,
+      "grad_norm": 0.2949135180976183,
+      "learning_rate": 2.3640524808896045e-05,
+      "loss": 0.4868,
+      "step": 612
+    },
+    {
+      "epoch": 2.4830379746835445,
+      "grad_norm": 0.2612370372073154,
+      "learning_rate": 2.3584699777776222e-05,
+      "loss": 0.4964,
+      "step": 613
+    },
+    {
+      "epoch": 2.4870886075949366,
+      "grad_norm": 0.3354666880249803,
+      "learning_rate": 2.3528845875977195e-05,
+      "loss": 0.5032,
+      "step": 614
+    },
+    {
+      "epoch": 2.491139240506329,
+      "grad_norm": 0.27514884289822517,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.4968,
+      "step": 615
+    },
+    {
+      "epoch": 2.4951898734177216,
+      "grad_norm": 0.27796319998884866,
+      "learning_rate": 2.341705325992901e-05,
+      "loss": 0.4917,
+      "step": 616
+    },
+    {
+      "epoch": 2.499240506329114,
+      "grad_norm": 0.2966415731216416,
+      "learning_rate": 2.336111544604222e-05,
+      "loss": 0.5006,
+      "step": 617
+    },
+    {
+      "epoch": 2.503291139240506,
+      "grad_norm": 0.33198184155237814,
+      "learning_rate": 2.33051505621937e-05,
+      "loss": 0.4962,
+      "step": 618
+    },
+    {
+      "epoch": 2.5073417721518987,
+      "grad_norm": 0.32037650148201513,
+      "learning_rate": 2.324915905911693e-05,
+      "loss": 0.5011,
+      "step": 619
+    },
+    {
+      "epoch": 2.511392405063291,
+      "grad_norm": 0.28922588892108747,
+      "learning_rate": 2.319314138775977e-05,
+      "loss": 0.4926,
+      "step": 620
+    },
+    {
+      "epoch": 2.5154430379746833,
+      "grad_norm": 0.26866023164828073,
+      "learning_rate": 2.3137097999280856e-05,
+      "loss": 0.4991,
+      "step": 621
+    },
+    {
+      "epoch": 2.519493670886076,
+      "grad_norm": 0.2827621378296447,
+      "learning_rate": 2.308102934504593e-05,
+      "loss": 0.4885,
+      "step": 622
+    },
+    {
+      "epoch": 2.5235443037974683,
+      "grad_norm": 0.3097043619358832,
+      "learning_rate": 2.3024935876624222e-05,
+      "loss": 0.4915,
+      "step": 623
+    },
+    {
+      "epoch": 2.527594936708861,
+      "grad_norm": 0.2875740411828757,
+      "learning_rate": 2.2968818045784813e-05,
+      "loss": 0.4846,
+      "step": 624
+    },
+    {
+      "epoch": 2.5316455696202533,
+      "grad_norm": 0.29068658002516107,
+      "learning_rate": 2.2912676304493006e-05,
+      "loss": 0.4972,
+      "step": 625
+    },
+    {
+      "epoch": 2.5356962025316454,
+      "grad_norm": 0.30112159308355235,
+      "learning_rate": 2.2856511104906668e-05,
+      "loss": 0.4879,
+      "step": 626
+    },
+    {
+      "epoch": 2.539746835443038,
+      "grad_norm": 0.2635491337442751,
+      "learning_rate": 2.2800322899372586e-05,
+      "loss": 0.5004,
+      "step": 627
+    },
+    {
+      "epoch": 2.5437974683544304,
+      "grad_norm": 0.31842707049819025,
+      "learning_rate": 2.2744112140422844e-05,
+      "loss": 0.4931,
+      "step": 628
+    },
+    {
+      "epoch": 2.547848101265823,
+      "grad_norm": 0.25857047000797534,
+      "learning_rate": 2.2687879280771177e-05,
+      "loss": 0.4877,
+      "step": 629
+    },
+    {
+      "epoch": 2.5518987341772155,
+      "grad_norm": 0.29850515851932796,
+      "learning_rate": 2.26316247733093e-05,
+      "loss": 0.4903,
+      "step": 630
+    },
+    {
+      "epoch": 2.5559493670886075,
+      "grad_norm": 0.27993176492378274,
+      "learning_rate": 2.257534907110328e-05,
+      "loss": 0.494,
+      "step": 631
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.36662098771491847,
+      "learning_rate": 2.2519052627389882e-05,
+      "loss": 0.5123,
+      "step": 632
+    },
+    {
+      "epoch": 2.5640506329113926,
+      "grad_norm": 0.28523291455581773,
+      "learning_rate": 2.246273589557294e-05,
+      "loss": 0.4856,
+      "step": 633
+    },
+    {
+      "epoch": 2.5681012658227846,
+      "grad_norm": 0.29974671367842526,
+      "learning_rate": 2.240639932921966e-05,
+      "loss": 0.4968,
+      "step": 634
+    },
+    {
+      "epoch": 2.572151898734177,
+      "grad_norm": 0.31383357163459236,
+      "learning_rate": 2.2350043382056995e-05,
+      "loss": 0.4973,
+      "step": 635
+    },
+    {
+      "epoch": 2.5762025316455697,
+      "grad_norm": 0.3042441068619628,
+      "learning_rate": 2.2293668507968015e-05,
+      "loss": 0.4966,
+      "step": 636
+    },
+    {
+      "epoch": 2.580253164556962,
+      "grad_norm": 0.2905209914024264,
+      "learning_rate": 2.2237275160988186e-05,
+      "loss": 0.5026,
+      "step": 637
+    },
+    {
+      "epoch": 2.5843037974683547,
+      "grad_norm": 0.2746713978226537,
+      "learning_rate": 2.2180863795301787e-05,
+      "loss": 0.4877,
+      "step": 638
+    },
+    {
+      "epoch": 2.5883544303797468,
+      "grad_norm": 0.30206459356592374,
+      "learning_rate": 2.212443486523819e-05,
+      "loss": 0.5028,
+      "step": 639
+    },
+    {
+      "epoch": 2.5924050632911393,
+      "grad_norm": 0.29619706104773535,
+      "learning_rate": 2.2067988825268243e-05,
+      "loss": 0.4941,
+      "step": 640
+    },
+    {
+      "epoch": 2.596455696202532,
+      "grad_norm": 0.27759459808262466,
+      "learning_rate": 2.2011526130000596e-05,
+      "loss": 0.5033,
+      "step": 641
+    },
+    {
+      "epoch": 2.600506329113924,
+      "grad_norm": 0.34207946939687345,
+      "learning_rate": 2.1955047234178038e-05,
+      "loss": 0.4905,
+      "step": 642
+    },
+    {
+      "epoch": 2.6045569620253164,
+      "grad_norm": 0.289541806516582,
+      "learning_rate": 2.1898552592673825e-05,
+      "loss": 0.5071,
+      "step": 643
+    },
+    {
+      "epoch": 2.608607594936709,
+      "grad_norm": 0.33897922877646514,
+      "learning_rate": 2.184204266048803e-05,
+      "loss": 0.5081,
+      "step": 644
+    },
+    {
+      "epoch": 2.6126582278481014,
+      "grad_norm": 0.2801030019538947,
+      "learning_rate": 2.1785517892743887e-05,
+      "loss": 0.5038,
+      "step": 645
+    },
+    {
+      "epoch": 2.616708860759494,
+      "grad_norm": 0.4123932984994133,
+      "learning_rate": 2.17289787446841e-05,
+      "loss": 0.5143,
+      "step": 646
+    },
+    {
+      "epoch": 2.620759493670886,
+      "grad_norm": 0.2771609500278878,
+      "learning_rate": 2.1672425671667198e-05,
+      "loss": 0.4964,
+      "step": 647
+    },
+    {
+      "epoch": 2.6248101265822785,
+      "grad_norm": 0.3298003987223228,
+      "learning_rate": 2.161585912916385e-05,
+      "loss": 0.4938,
+      "step": 648
+    },
+    {
+      "epoch": 2.628860759493671,
+      "grad_norm": 0.29930235006281114,
+      "learning_rate": 2.1559279572753214e-05,
+      "loss": 0.497,
+      "step": 649
+    },
+    {
+      "epoch": 2.632911392405063,
+      "grad_norm": 0.3273586847008571,
+      "learning_rate": 2.1502687458119268e-05,
+      "loss": 0.5006,
+      "step": 650
+    },
+    {
+      "epoch": 2.6369620253164556,
+      "grad_norm": 0.29693780415058896,
+      "learning_rate": 2.1446083241047116e-05,
+      "loss": 0.5028,
+      "step": 651
+    },
+    {
+      "epoch": 2.641012658227848,
+      "grad_norm": 0.39720266304779406,
+      "learning_rate": 2.1389467377419333e-05,
+      "loss": 0.4944,
+      "step": 652
+    },
+    {
+      "epoch": 2.6450632911392407,
+      "grad_norm": 0.3061413946600301,
+      "learning_rate": 2.133284032321232e-05,
+      "loss": 0.4962,
+      "step": 653
+    },
+    {
+      "epoch": 2.649113924050633,
+      "grad_norm": 0.39421822851395333,
+      "learning_rate": 2.1276202534492566e-05,
+      "loss": 0.5055,
+      "step": 654
+    },
+    {
+      "epoch": 2.6531645569620252,
+      "grad_norm": 0.2648940708939872,
+      "learning_rate": 2.121955446741306e-05,
+      "loss": 0.4976,
+      "step": 655
+    },
+    {
+      "epoch": 2.6572151898734178,
+      "grad_norm": 0.3273026481078858,
+      "learning_rate": 2.1162896578209517e-05,
+      "loss": 0.4967,
+      "step": 656
+    },
+    {
+      "epoch": 2.6612658227848103,
+      "grad_norm": 0.28913611708435133,
+      "learning_rate": 2.1106229323196813e-05,
+      "loss": 0.4874,
+      "step": 657
+    },
+    {
+      "epoch": 2.6653164556962023,
+      "grad_norm": 0.32782233628991164,
+      "learning_rate": 2.1049553158765214e-05,
+      "loss": 0.4877,
+      "step": 658
+    },
+    {
+      "epoch": 2.669367088607595,
+      "grad_norm": 0.31894426242960916,
+      "learning_rate": 2.0992868541376764e-05,
+      "loss": 0.5043,
+      "step": 659
+    },
+    {
+      "epoch": 2.6734177215189874,
+      "grad_norm": 0.3152201058568446,
+      "learning_rate": 2.093617592756158e-05,
+      "loss": 0.4965,
+      "step": 660
+    },
+    {
+      "epoch": 2.67746835443038,
+      "grad_norm": 0.28616806749112955,
+      "learning_rate": 2.0879475773914167e-05,
+      "loss": 0.4886,
+      "step": 661
+    },
+    {
+      "epoch": 2.6815189873417724,
+      "grad_norm": 0.30579787133746056,
+      "learning_rate": 2.082276853708978e-05,
+      "loss": 0.5055,
+      "step": 662
+    },
+    {
+      "epoch": 2.6855696202531645,
+      "grad_norm": 0.2797489009466575,
+      "learning_rate": 2.076605467380071e-05,
+      "loss": 0.5028,
+      "step": 663
+    },
+    {
+      "epoch": 2.689620253164557,
+      "grad_norm": 0.3080551790710324,
+      "learning_rate": 2.0709334640812613e-05,
+      "loss": 0.5107,
+      "step": 664
+    },
+    {
+      "epoch": 2.6936708860759495,
+      "grad_norm": 0.2743343343989471,
+      "learning_rate": 2.0652608894940824e-05,
+      "loss": 0.4975,
+      "step": 665
+    },
+    {
+      "epoch": 2.6977215189873416,
+      "grad_norm": 0.3105182440965287,
+      "learning_rate": 2.0595877893046722e-05,
+      "loss": 0.5032,
+      "step": 666
+    },
+    {
+      "epoch": 2.701772151898734,
+      "grad_norm": 0.2571231781361565,
+      "learning_rate": 2.0539142092033985e-05,
+      "loss": 0.4917,
+      "step": 667
+    },
+    {
+      "epoch": 2.7058227848101266,
+      "grad_norm": 0.28775542715297925,
+      "learning_rate": 2.048240194884496e-05,
+      "loss": 0.5399,
+      "step": 668
+    },
+    {
+      "epoch": 2.709873417721519,
+      "grad_norm": 0.2756267958286078,
+      "learning_rate": 2.042565792045695e-05,
+      "loss": 0.5076,
+      "step": 669
+    },
+    {
+      "epoch": 2.7139240506329116,
+      "grad_norm": 0.2842531703217749,
+      "learning_rate": 2.036891046387857e-05,
+      "loss": 0.4914,
+      "step": 670
+    },
+    {
+      "epoch": 2.7179746835443037,
+      "grad_norm": 0.2773480211098765,
+      "learning_rate": 2.0312160036146036e-05,
+      "loss": 0.493,
+      "step": 671
+    },
+    {
+      "epoch": 2.7220253164556962,
+      "grad_norm": 0.2826452470236453,
+      "learning_rate": 2.025540709431948e-05,
+      "loss": 0.4965,
+      "step": 672
+    },
+    {
+      "epoch": 2.7260759493670887,
+      "grad_norm": 0.26657365743113487,
+      "learning_rate": 2.0198652095479298e-05,
+      "loss": 0.4943,
+      "step": 673
+    },
+    {
+      "epoch": 2.730126582278481,
+      "grad_norm": 0.2915439556788406,
+      "learning_rate": 2.014189549672245e-05,
+      "loss": 0.5093,
+      "step": 674
+    },
+    {
+      "epoch": 2.7341772151898733,
+      "grad_norm": 0.2862066783149554,
+      "learning_rate": 2.0085137755158776e-05,
+      "loss": 0.4878,
+      "step": 675
+    },
+    {
+      "epoch": 2.738227848101266,
+      "grad_norm": 0.298160084251785,
+      "learning_rate": 2.0028379327907327e-05,
+      "loss": 0.504,
+      "step": 676
+    },
+    {
+      "epoch": 2.7422784810126584,
+      "grad_norm": 2.0376198456238543,
+      "learning_rate": 1.9971620672092676e-05,
+      "loss": 0.5262,
+      "step": 677
+    },
+    {
+      "epoch": 2.746329113924051,
+      "grad_norm": 0.35009625876342043,
+      "learning_rate": 1.991486224484123e-05,
+      "loss": 0.4948,
+      "step": 678
+    },
+    {
+      "epoch": 2.750379746835443,
+      "grad_norm": 0.2831380485829953,
+      "learning_rate": 1.985810450327756e-05,
+      "loss": 0.507,
+      "step": 679
+    },
+    {
+      "epoch": 2.7544303797468355,
+      "grad_norm": 0.300762205194341,
+      "learning_rate": 1.9801347904520706e-05,
+      "loss": 0.4845,
+      "step": 680
+    },
+    {
+      "epoch": 2.758481012658228,
+      "grad_norm": 0.2738009756081014,
+      "learning_rate": 1.974459290568053e-05,
+      "loss": 0.4841,
+      "step": 681
+    },
+    {
+      "epoch": 2.76253164556962,
+      "grad_norm": 0.2737402593350786,
+      "learning_rate": 1.968783996385397e-05,
+      "loss": 0.5028,
+      "step": 682
+    },
+    {
+      "epoch": 2.7665822784810126,
+      "grad_norm": 0.2812489312815858,
+      "learning_rate": 1.963108953612143e-05,
+      "loss": 0.4915,
+      "step": 683
+    },
+    {
+      "epoch": 2.770632911392405,
+      "grad_norm": 0.25802633431883,
+      "learning_rate": 1.9574342079543056e-05,
+      "loss": 0.5051,
+      "step": 684
+    },
+    {
+      "epoch": 2.7746835443037976,
+      "grad_norm": 0.27649359775041665,
+      "learning_rate": 1.9517598051155046e-05,
+      "loss": 0.5032,
+      "step": 685
+    },
+    {
+      "epoch": 2.77873417721519,
+      "grad_norm": 0.36820557424940736,
+      "learning_rate": 1.9460857907966025e-05,
+      "loss": 0.5013,
+      "step": 686
+    },
+    {
+      "epoch": 2.782784810126582,
+      "grad_norm": 0.29733612745412613,
+      "learning_rate": 1.9404122106953285e-05,
+      "loss": 0.4921,
+      "step": 687
+    },
+    {
+      "epoch": 2.7868354430379747,
+      "grad_norm": 0.25570784399063073,
+      "learning_rate": 1.9347391105059176e-05,
+      "loss": 0.5121,
+      "step": 688
+    },
+    {
+      "epoch": 2.790886075949367,
+      "grad_norm": 0.29028322365709885,
+      "learning_rate": 1.92906653591874e-05,
+      "loss": 0.4975,
+      "step": 689
+    },
+    {
+      "epoch": 2.7949367088607593,
+      "grad_norm": 0.2858158142533702,
+      "learning_rate": 1.9233945326199295e-05,
+      "loss": 0.4912,
+      "step": 690
+    },
+    {
+      "epoch": 2.798987341772152,
+      "grad_norm": 0.27038908339861406,
+      "learning_rate": 1.917723146291022e-05,
+      "loss": 0.4881,
+      "step": 691
+    },
+    {
+      "epoch": 2.8030379746835443,
+      "grad_norm": 0.3244278903539978,
+      "learning_rate": 1.912052422608584e-05,
+      "loss": 0.499,
+      "step": 692
+    },
+    {
+      "epoch": 2.807088607594937,
+      "grad_norm": 0.2668629031840216,
+      "learning_rate": 1.9063824072438428e-05,
+      "loss": 0.4891,
+      "step": 693
+    },
+    {
+      "epoch": 2.8111392405063294,
+      "grad_norm": 0.32876427256691837,
+      "learning_rate": 1.9007131458623246e-05,
+      "loss": 0.4952,
+      "step": 694
+    },
+    {
+      "epoch": 2.8151898734177214,
+      "grad_norm": 0.27395260441680314,
+      "learning_rate": 1.895044684123479e-05,
+      "loss": 0.4897,
+      "step": 695
+    },
+    {
+      "epoch": 2.819240506329114,
+      "grad_norm": 0.29684657219223215,
+      "learning_rate": 1.8893770676803194e-05,
+      "loss": 0.4947,
+      "step": 696
+    },
+    {
+      "epoch": 2.8232911392405065,
+      "grad_norm": 0.2524991107718365,
+      "learning_rate": 1.8837103421790486e-05,
+      "loss": 0.4995,
+      "step": 697
+    },
+    {
+      "epoch": 2.8273417721518985,
+      "grad_norm": 0.2835877207971906,
+      "learning_rate": 1.8780445532586952e-05,
+      "loss": 0.4829,
+      "step": 698
+    },
+    {
+      "epoch": 2.831392405063291,
+      "grad_norm": 0.2763473812502709,
+      "learning_rate": 1.872379746550743e-05,
+      "loss": 0.4923,
+      "step": 699
+    },
+    {
+      "epoch": 2.8354430379746836,
+      "grad_norm": 0.26485940763163784,
+      "learning_rate": 1.866715967678769e-05,
+      "loss": 0.5001,
+      "step": 700
+    },
+    {
+      "epoch": 2.839493670886076,
+      "grad_norm": 0.27639888937223445,
+      "learning_rate": 1.861053262258067e-05,
+      "loss": 0.4767,
+      "step": 701
+    },
+    {
+      "epoch": 2.8435443037974686,
+      "grad_norm": 0.2721870373012777,
+      "learning_rate": 1.8553916758952897e-05,
+      "loss": 0.4934,
+      "step": 702
+    },
+    {
+      "epoch": 2.8475949367088607,
+      "grad_norm": 0.2550266758670135,
+      "learning_rate": 1.8497312541880735e-05,
+      "loss": 0.4953,
+      "step": 703
+    },
+    {
+      "epoch": 2.851645569620253,
+      "grad_norm": 0.35796950669575617,
+      "learning_rate": 1.8440720427246786e-05,
+      "loss": 0.4947,
+      "step": 704
+    },
+    {
+      "epoch": 2.8556962025316457,
+      "grad_norm": 0.2614461247313244,
+      "learning_rate": 1.8384140870836157e-05,
+      "loss": 0.4896,
+      "step": 705
+    },
+    {
+      "epoch": 2.8597468354430378,
+      "grad_norm": 1.6133534115728645,
+      "learning_rate": 1.8327574328332806e-05,
+      "loss": 0.5013,
+      "step": 706
+    },
+    {
+      "epoch": 2.8637974683544303,
+      "grad_norm": 0.3066929857091205,
+      "learning_rate": 1.8271021255315906e-05,
+      "loss": 0.4877,
+      "step": 707
+    },
+    {
+      "epoch": 2.867848101265823,
+      "grad_norm": 0.25264249371691305,
+      "learning_rate": 1.8214482107256117e-05,
+      "loss": 0.5023,
+      "step": 708
+    },
+    {
+      "epoch": 2.8718987341772153,
+      "grad_norm": 0.3000244077050904,
+      "learning_rate": 1.8157957339511968e-05,
+      "loss": 0.4936,
+      "step": 709
+    },
+    {
+      "epoch": 2.875949367088608,
+      "grad_norm": 0.27713021321079545,
+      "learning_rate": 1.8101447407326182e-05,
+      "loss": 0.5005,
+      "step": 710
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.2737721318120769,
+      "learning_rate": 1.8044952765821966e-05,
+      "loss": 0.4817,
+      "step": 711
+    },
+    {
+      "epoch": 2.8840506329113924,
+      "grad_norm": 0.3121847023934415,
+      "learning_rate": 1.7988473869999407e-05,
+      "loss": 0.4806,
+      "step": 712
+    },
+    {
+      "epoch": 2.888101265822785,
+      "grad_norm": 0.2655240611091641,
+      "learning_rate": 1.7932011174731764e-05,
+      "loss": 0.5003,
+      "step": 713
+    },
+    {
+      "epoch": 2.892151898734177,
+      "grad_norm": 0.2957878806289894,
+      "learning_rate": 1.7875565134761817e-05,
+      "loss": 0.4952,
+      "step": 714
+    },
+    {
+      "epoch": 2.8962025316455695,
+      "grad_norm": 0.2942890537605514,
+      "learning_rate": 1.7819136204698226e-05,
+      "loss": 0.4967,
+      "step": 715
+    },
+    {
+      "epoch": 2.900253164556962,
+      "grad_norm": 0.2922875722976685,
+      "learning_rate": 1.776272483901182e-05,
+      "loss": 0.503,
+      "step": 716
+    },
+    {
+      "epoch": 2.9043037974683545,
+      "grad_norm": 0.26705258154835576,
+      "learning_rate": 1.7706331492031995e-05,
+      "loss": 0.5033,
+      "step": 717
+    },
+    {
+      "epoch": 2.908354430379747,
+      "grad_norm": 0.31071720739091907,
+      "learning_rate": 1.764995661794301e-05,
+      "loss": 0.4842,
+      "step": 718
+    },
+    {
+      "epoch": 2.912405063291139,
+      "grad_norm": 0.24276996734463183,
+      "learning_rate": 1.759360067078035e-05,
+      "loss": 0.5013,
+      "step": 719
+    },
+    {
+      "epoch": 2.9164556962025316,
+      "grad_norm": 0.23951810930285092,
+      "learning_rate": 1.7537264104427064e-05,
+      "loss": 0.4868,
+      "step": 720
+    },
+    {
+      "epoch": 2.920506329113924,
+      "grad_norm": 0.2683974126922137,
+      "learning_rate": 1.748094737261012e-05,
+      "loss": 0.5001,
+      "step": 721
+    },
+    {
+      "epoch": 2.9245569620253162,
+      "grad_norm": 0.251064338218705,
+      "learning_rate": 1.7424650928896726e-05,
+      "loss": 0.4966,
+      "step": 722
+    },
+    {
+      "epoch": 2.9286075949367087,
+      "grad_norm": 0.2552877190403974,
+      "learning_rate": 1.7368375226690712e-05,
+      "loss": 0.4967,
+      "step": 723
+    },
+    {
+      "epoch": 2.9326582278481013,
+      "grad_norm": 0.27055770247994326,
+      "learning_rate": 1.731212071922883e-05,
+      "loss": 0.5093,
+      "step": 724
+    },
+    {
+      "epoch": 2.9367088607594938,
+      "grad_norm": 0.26596501557101804,
+      "learning_rate": 1.7255887859577156e-05,
+      "loss": 0.4906,
+      "step": 725
+    },
+    {
+      "epoch": 2.9407594936708863,
+      "grad_norm": 0.2501337280468811,
+      "learning_rate": 1.7199677100627427e-05,
+      "loss": 0.4904,
+      "step": 726
+    },
+    {
+      "epoch": 2.9448101265822784,
+      "grad_norm": 0.2694848375164097,
+      "learning_rate": 1.7143488895093343e-05,
+      "loss": 0.4918,
+      "step": 727
+    },
+    {
+      "epoch": 2.948860759493671,
+      "grad_norm": 0.2634575949801865,
+      "learning_rate": 1.7087323695506994e-05,
+      "loss": 0.4974,
+      "step": 728
+    },
+    {
+      "epoch": 2.9529113924050634,
+      "grad_norm": 0.24835178319071471,
+      "learning_rate": 1.7031181954215194e-05,
+      "loss": 0.5002,
+      "step": 729
+    },
+    {
+      "epoch": 2.9569620253164555,
+      "grad_norm": 0.2955210966050652,
+      "learning_rate": 1.6975064123375788e-05,
+      "loss": 0.4909,
+      "step": 730
+    },
+    {
+      "epoch": 2.961012658227848,
+      "grad_norm": 0.300641133862044,
+      "learning_rate": 1.6918970654954084e-05,
+      "loss": 0.5052,
+      "step": 731
+    },
+    {
+      "epoch": 2.9650632911392405,
+      "grad_norm": 0.2617354775594947,
+      "learning_rate": 1.686290200071915e-05,
+      "loss": 0.5036,
+      "step": 732
+    },
+    {
+      "epoch": 2.969113924050633,
+      "grad_norm": 0.28691307269361793,
+      "learning_rate": 1.6806858612240234e-05,
+      "loss": 0.4971,
+      "step": 733
+    },
+    {
+      "epoch": 2.9731645569620255,
+      "grad_norm": 0.26335302115148607,
+      "learning_rate": 1.6750840940883078e-05,
+      "loss": 0.49,
+      "step": 734
+    },
+    {
+      "epoch": 2.9772151898734176,
+      "grad_norm": 3.4727442143314513,
+      "learning_rate": 1.6694849437806305e-05,
+      "loss": 0.506,
+      "step": 735
+    },
+    {
+      "epoch": 2.98126582278481,
+      "grad_norm": 0.3192207578793137,
+      "learning_rate": 1.663888455395778e-05,
+      "loss": 0.5011,
+      "step": 736
+    },
+    {
+      "epoch": 2.9853164556962026,
+      "grad_norm": 0.4396074404535879,
+      "learning_rate": 1.6582946740070995e-05,
+      "loss": 0.5028,
+      "step": 737
+    },
+    {
+      "epoch": 2.9893670886075947,
+      "grad_norm": 0.3821504980040192,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.5023,
+      "step": 738
+    },
+    {
+      "epoch": 2.993417721518987,
+      "grad_norm": 0.2670358593731776,
+      "learning_rate": 1.6471154124022818e-05,
+      "loss": 0.5054,
+      "step": 739
+    },
+    {
+      "epoch": 2.9974683544303797,
+      "grad_norm": 0.27969312303732413,
+      "learning_rate": 1.6415300222223788e-05,
+      "loss": 0.4913,
+      "step": 740
+    },
+    {
+      "epoch": 3.0015189873417722,
+      "grad_norm": 0.29171649301826386,
+      "learning_rate": 1.6359475191103958e-05,
+      "loss": 0.4657,
+      "step": 741
+    },
+    {
+      "epoch": 3.0055696202531648,
+      "grad_norm": 0.39972494156104044,
+      "learning_rate": 1.6303679480270466e-05,
+      "loss": 0.4452,
+      "step": 742
+    },
+    {
+      "epoch": 3.009620253164557,
+      "grad_norm": 0.6353984690276141,
+      "learning_rate": 1.624791353909428e-05,
+      "loss": 0.438,
+      "step": 743
+    },
+    {
+      "epoch": 3.0136708860759494,
+      "grad_norm": 0.3518505938242377,
+      "learning_rate": 1.619217781670663e-05,
+      "loss": 0.4348,
+      "step": 744
+    },
+    {
+      "epoch": 3.017721518987342,
+      "grad_norm": 0.41518844880795264,
+      "learning_rate": 1.6136472761995373e-05,
+      "loss": 0.4406,
+      "step": 745
+    },
+    {
+      "epoch": 3.0217721518987344,
+      "grad_norm": 0.3458547126104315,
+      "learning_rate": 1.608079882360134e-05,
+      "loss": 0.438,
+      "step": 746
+    },
+    {
+      "epoch": 3.0258227848101265,
+      "grad_norm": 0.3416722454862477,
+      "learning_rate": 1.60251564499148e-05,
+      "loss": 0.4471,
+      "step": 747
+    },
+    {
+      "epoch": 3.029873417721519,
+      "grad_norm": 0.34497444487730256,
+      "learning_rate": 1.596954608907176e-05,
+      "loss": 0.4441,
+      "step": 748
+    },
+    {
+      "epoch": 3.0339240506329115,
+      "grad_norm": 0.37570356781825276,
+      "learning_rate": 1.591396818895043e-05,
+      "loss": 0.4253,
+      "step": 749
+    },
+    {
+      "epoch": 3.037974683544304,
+      "grad_norm": 0.2996699480621195,
+      "learning_rate": 1.585842319716759e-05,
+      "loss": 0.4257,
+      "step": 750
+    },
+    {
+      "epoch": 3.042025316455696,
+      "grad_norm": 0.3613568369448993,
+      "learning_rate": 1.5802911561074944e-05,
+      "loss": 0.4324,
+      "step": 751
+    },
+    {
+      "epoch": 3.0460759493670886,
+      "grad_norm": 0.29818813482264406,
+      "learning_rate": 1.5747433727755595e-05,
+      "loss": 0.4224,
+      "step": 752
+    },
+    {
+      "epoch": 3.050126582278481,
+      "grad_norm": 0.32680691617629415,
+      "learning_rate": 1.5691990144020376e-05,
+      "loss": 0.4361,
+      "step": 753
+    },
+    {
+      "epoch": 3.0541772151898736,
+      "grad_norm": 0.31576869081803993,
+      "learning_rate": 1.5636581256404297e-05,
+      "loss": 0.4299,
+      "step": 754
+    },
+    {
+      "epoch": 3.0582278481012657,
+      "grad_norm": 0.2925381486116208,
+      "learning_rate": 1.558120751116291e-05,
+      "loss": 0.4395,
+      "step": 755
+    },
+    {
+      "epoch": 3.062278481012658,
+      "grad_norm": 0.34387959945691604,
+      "learning_rate": 1.552586935426876e-05,
+      "loss": 0.4302,
+      "step": 756
+    },
+    {
+      "epoch": 3.0663291139240507,
+      "grad_norm": 0.3301456618558225,
+      "learning_rate": 1.547056723140774e-05,
+      "loss": 0.4313,
+      "step": 757
+    },
+    {
+      "epoch": 3.0703797468354432,
+      "grad_norm": 0.28201007855114824,
+      "learning_rate": 1.5415301587975565e-05,
+      "loss": 0.431,
+      "step": 758
+    },
+    {
+      "epoch": 3.0744303797468353,
+      "grad_norm": 0.3210832257613439,
+      "learning_rate": 1.536007286907411e-05,
+      "loss": 0.4382,
+      "step": 759
+    },
+    {
+      "epoch": 3.078481012658228,
+      "grad_norm": 0.2845440775505645,
+      "learning_rate": 1.5304881519507896e-05,
+      "loss": 0.4377,
+      "step": 760
+    },
+    {
+      "epoch": 3.0825316455696203,
+      "grad_norm": 0.30850636036321694,
+      "learning_rate": 1.5249727983780453e-05,
+      "loss": 0.4225,
+      "step": 761
+    },
+    {
+      "epoch": 3.086582278481013,
+      "grad_norm": 0.25848900392492297,
+      "learning_rate": 1.5194612706090786e-05,
+      "loss": 0.4263,
+      "step": 762
+    },
+    {
+      "epoch": 3.090632911392405,
+      "grad_norm": 0.31678876258217614,
+      "learning_rate": 1.5139536130329771e-05,
+      "loss": 0.4286,
+      "step": 763
+    },
+    {
+      "epoch": 3.0946835443037974,
+      "grad_norm": 0.27720936345976654,
+      "learning_rate": 1.508449870007656e-05,
+      "loss": 0.4339,
+      "step": 764
+    },
+    {
+      "epoch": 3.09873417721519,
+      "grad_norm": 0.27476423070673495,
+      "learning_rate": 1.5029500858595056e-05,
+      "loss": 0.4255,
+      "step": 765
+    },
+    {
+      "epoch": 3.1027848101265825,
+      "grad_norm": 0.2716025347453592,
+      "learning_rate": 1.4974543048830328e-05,
+      "loss": 0.4344,
+      "step": 766
+    },
+    {
+      "epoch": 3.1068354430379745,
+      "grad_norm": 0.2707203814394801,
+      "learning_rate": 1.4919625713405e-05,
+      "loss": 0.4227,
+      "step": 767
+    },
+    {
+      "epoch": 3.110886075949367,
+      "grad_norm": 0.2976307456731226,
+      "learning_rate": 1.4864749294615756e-05,
+      "loss": 0.4242,
+      "step": 768
+    },
+    {
+      "epoch": 3.1149367088607596,
+      "grad_norm": 0.27377586251959085,
+      "learning_rate": 1.4809914234429716e-05,
+      "loss": 0.4334,
+      "step": 769
+    },
+    {
+      "epoch": 3.118987341772152,
+      "grad_norm": 0.27837528917951615,
+      "learning_rate": 1.4755120974480923e-05,
+      "loss": 0.4369,
+      "step": 770
+    },
+    {
+      "epoch": 3.123037974683544,
+      "grad_norm": 0.28244733712178194,
+      "learning_rate": 1.4700369956066771e-05,
+      "loss": 0.4344,
+      "step": 771
+    },
+    {
+      "epoch": 3.1270886075949367,
+      "grad_norm": 0.26282981493289764,
+      "learning_rate": 1.4645661620144413e-05,
+      "loss": 0.4381,
+      "step": 772
+    },
+    {
+      "epoch": 3.131139240506329,
+      "grad_norm": 0.24521993529106859,
+      "learning_rate": 1.4590996407327284e-05,
+      "loss": 0.426,
+      "step": 773
+    },
+    {
+      "epoch": 3.1351898734177217,
+      "grad_norm": 0.26001656512825644,
+      "learning_rate": 1.4536374757881487e-05,
+      "loss": 0.4325,
+      "step": 774
+    },
+    {
+      "epoch": 3.1392405063291138,
+      "grad_norm": 0.25739672426342586,
+      "learning_rate": 1.4481797111722271e-05,
+      "loss": 0.4292,
+      "step": 775
+    },
+    {
+      "epoch": 3.1432911392405063,
+      "grad_norm": 0.2558041238552263,
+      "learning_rate": 1.4427263908410507e-05,
+      "loss": 0.4418,
+      "step": 776
+    },
+    {
+      "epoch": 3.147341772151899,
+      "grad_norm": 0.23554103954948294,
+      "learning_rate": 1.4372775587149108e-05,
+      "loss": 0.4332,
+      "step": 777
+    },
+    {
+      "epoch": 3.1513924050632913,
+      "grad_norm": 0.32553550885698535,
+      "learning_rate": 1.4318332586779522e-05,
+      "loss": 0.4249,
+      "step": 778
+    },
+    {
+      "epoch": 3.1554430379746834,
+      "grad_norm": 0.2505559760668328,
+      "learning_rate": 1.4263935345778202e-05,
+      "loss": 0.4462,
+      "step": 779
+    },
+    {
+      "epoch": 3.159493670886076,
+      "grad_norm": 0.24605912956100648,
+      "learning_rate": 1.420958430225303e-05,
+      "loss": 0.4419,
+      "step": 780
+    },
+    {
+      "epoch": 3.1635443037974684,
+      "grad_norm": 0.26705625759158536,
+      "learning_rate": 1.415527989393985e-05,
+      "loss": 0.4232,
+      "step": 781
+    },
+    {
+      "epoch": 3.167594936708861,
+      "grad_norm": 0.3092296115471002,
+      "learning_rate": 1.410102255819891e-05,
+      "loss": 0.4244,
+      "step": 782
+    },
+    {
+      "epoch": 3.171645569620253,
+      "grad_norm": 0.2562421835534282,
+      "learning_rate": 1.404681273201131e-05,
+      "loss": 0.434,
+      "step": 783
+    },
+    {
+      "epoch": 3.1756962025316455,
+      "grad_norm": 0.30292328317796197,
+      "learning_rate": 1.399265085197556e-05,
+      "loss": 0.4242,
+      "step": 784
+    },
+    {
+      "epoch": 3.179746835443038,
+      "grad_norm": 0.23074618007151385,
+      "learning_rate": 1.393853735430398e-05,
+      "loss": 0.43,
+      "step": 785
+    },
+    {
+      "epoch": 3.1837974683544306,
+      "grad_norm": 0.28722928093934147,
+      "learning_rate": 1.3884472674819246e-05,
+      "loss": 0.4303,
+      "step": 786
+    },
+    {
+      "epoch": 3.1878481012658226,
+      "grad_norm": 0.26395234188326233,
+      "learning_rate": 1.3830457248950864e-05,
+      "loss": 0.4443,
+      "step": 787
+    },
+    {
+      "epoch": 3.191898734177215,
+      "grad_norm": 0.26450192496115144,
+      "learning_rate": 1.377649151173163e-05,
+      "loss": 0.4297,
+      "step": 788
+    },
+    {
+      "epoch": 3.1959493670886077,
+      "grad_norm": 0.26705864718413697,
+      "learning_rate": 1.3722575897794181e-05,
+      "loss": 0.4358,
+      "step": 789
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.27132374025593453,
+      "learning_rate": 1.3668710841367472e-05,
+      "loss": 0.4256,
+      "step": 790
+    },
+    {
+      "epoch": 3.2040506329113922,
+      "grad_norm": 0.246608083573907,
+      "learning_rate": 1.361489677627324e-05,
+      "loss": 0.4535,
+      "step": 791
+    },
+    {
+      "epoch": 3.2081012658227848,
+      "grad_norm": 0.25091600330232317,
+      "learning_rate": 1.3561134135922585e-05,
+      "loss": 0.4351,
+      "step": 792
+    },
+    {
+      "epoch": 3.2121518987341773,
+      "grad_norm": 0.24417463000316258,
+      "learning_rate": 1.350742335331241e-05,
+      "loss": 0.4316,
+      "step": 793
+    },
+    {
+      "epoch": 3.21620253164557,
+      "grad_norm": 0.24899623437466628,
+      "learning_rate": 1.345376486102198e-05,
+      "loss": 0.4371,
+      "step": 794
+    },
+    {
+      "epoch": 3.220253164556962,
+      "grad_norm": 0.2707806761613245,
+      "learning_rate": 1.3400159091209414e-05,
+      "loss": 0.4321,
+      "step": 795
+    },
+    {
+      "epoch": 3.2243037974683544,
+      "grad_norm": 0.24153668966758615,
+      "learning_rate": 1.3346606475608216e-05,
+      "loss": 0.4324,
+      "step": 796
+    },
+    {
+      "epoch": 3.228354430379747,
+      "grad_norm": 0.26914183299053124,
+      "learning_rate": 1.3293107445523781e-05,
+      "loss": 0.4361,
+      "step": 797
+    },
+    {
+      "epoch": 3.2324050632911394,
+      "grad_norm": 2.199706234677971,
+      "learning_rate": 1.3239662431829949e-05,
+      "loss": 0.446,
+      "step": 798
+    },
+    {
+      "epoch": 3.2364556962025315,
+      "grad_norm": 0.28314967225401405,
+      "learning_rate": 1.3186271864965509e-05,
+      "loss": 0.4338,
+      "step": 799
+    },
+    {
+      "epoch": 3.240506329113924,
+      "grad_norm": 0.2625511072655538,
+      "learning_rate": 1.3132936174930756e-05,
+      "loss": 0.4165,
+      "step": 800
+    },
+    {
+      "epoch": 3.2445569620253165,
+      "grad_norm": 0.30238883644151415,
+      "learning_rate": 1.3079655791283995e-05,
+      "loss": 0.4431,
+      "step": 801
+    },
+    {
+      "epoch": 3.248607594936709,
+      "grad_norm": 0.28853368064812934,
+      "learning_rate": 1.3026431143138108e-05,
+      "loss": 0.4352,
+      "step": 802
+    },
+    {
+      "epoch": 3.252658227848101,
+      "grad_norm": 0.3011655806884066,
+      "learning_rate": 1.2973262659157114e-05,
+      "loss": 0.4274,
+      "step": 803
+    },
+    {
+      "epoch": 3.2567088607594936,
+      "grad_norm": 0.2611284651006007,
+      "learning_rate": 1.2920150767552651e-05,
+      "loss": 0.432,
+      "step": 804
+    },
+    {
+      "epoch": 3.260759493670886,
+      "grad_norm": 0.2849311560884644,
+      "learning_rate": 1.2867095896080607e-05,
+      "loss": 0.4388,
+      "step": 805
+    },
+    {
+      "epoch": 3.2648101265822786,
+      "grad_norm": 0.23460173079276833,
+      "learning_rate": 1.2814098472037612e-05,
+      "loss": 0.4477,
+      "step": 806
+    },
+    {
+      "epoch": 3.2688607594936707,
+      "grad_norm": 0.27435842772560776,
+      "learning_rate": 1.276115892225764e-05,
+      "loss": 0.4273,
+      "step": 807
+    },
+    {
+      "epoch": 3.2729113924050632,
+      "grad_norm": 0.25563684189548697,
+      "learning_rate": 1.2708277673108555e-05,
+      "loss": 0.4294,
+      "step": 808
+    },
+    {
+      "epoch": 3.2769620253164558,
+      "grad_norm": 0.24289940179427122,
+      "learning_rate": 1.2655455150488649e-05,
+      "loss": 0.4395,
+      "step": 809
+    },
+    {
+      "epoch": 3.2810126582278483,
+      "grad_norm": 0.26292049591866495,
+      "learning_rate": 1.2602691779823272e-05,
+      "loss": 0.4275,
+      "step": 810
+    },
+    {
+      "epoch": 3.2850632911392403,
+      "grad_norm": 0.24541676480776042,
+      "learning_rate": 1.2549987986061355e-05,
+      "loss": 0.4244,
+      "step": 811
+    },
+    {
+      "epoch": 3.289113924050633,
+      "grad_norm": 0.24736775506489292,
+      "learning_rate": 1.2497344193672005e-05,
+      "loss": 0.4376,
+      "step": 812
+    },
+    {
+      "epoch": 3.2931645569620254,
+      "grad_norm": 0.25386803498782706,
+      "learning_rate": 1.2444760826641092e-05,
+      "loss": 0.4336,
+      "step": 813
+    },
+    {
+      "epoch": 3.297215189873418,
+      "grad_norm": 0.25192969441066226,
+      "learning_rate": 1.2392238308467817e-05,
+      "loss": 0.435,
+      "step": 814
+    },
+    {
+      "epoch": 3.30126582278481,
+      "grad_norm": 0.2618396734734109,
+      "learning_rate": 1.2339777062161326e-05,
+      "loss": 0.436,
+      "step": 815
+    },
+    {
+      "epoch": 3.3053164556962025,
+      "grad_norm": 0.25036945200016286,
+      "learning_rate": 1.2287377510237293e-05,
+      "loss": 0.4333,
+      "step": 816
+    },
+    {
+      "epoch": 3.309367088607595,
+      "grad_norm": 0.24628982363812588,
+      "learning_rate": 1.2235040074714488e-05,
+      "loss": 0.438,
+      "step": 817
+    },
+    {
+      "epoch": 3.3134177215189875,
+      "grad_norm": 0.25052559186170864,
+      "learning_rate": 1.2182765177111434e-05,
+      "loss": 0.4266,
+      "step": 818
+    },
+    {
+      "epoch": 3.3174683544303796,
+      "grad_norm": 0.8948975323516721,
+      "learning_rate": 1.213055323844297e-05,
+      "loss": 0.4447,
+      "step": 819
+    },
+    {
+      "epoch": 3.321518987341772,
+      "grad_norm": 0.2523461972388844,
+      "learning_rate": 1.2078404679216864e-05,
+      "loss": 0.4483,
+      "step": 820
+    },
+    {
+      "epoch": 3.3255696202531646,
+      "grad_norm": 0.277984973030442,
+      "learning_rate": 1.2026319919430458e-05,
+      "loss": 0.4296,
+      "step": 821
+    },
+    {
+      "epoch": 3.329620253164557,
+      "grad_norm": 0.25372345863118817,
+      "learning_rate": 1.1974299378567227e-05,
+      "loss": 0.4259,
+      "step": 822
+    },
+    {
+      "epoch": 3.333670886075949,
+      "grad_norm": 0.24132471456352442,
+      "learning_rate": 1.1922343475593462e-05,
+      "loss": 0.4359,
+      "step": 823
+    },
+    {
+      "epoch": 3.3377215189873417,
+      "grad_norm": 0.2680411758275496,
+      "learning_rate": 1.187045262895488e-05,
+      "loss": 0.4192,
+      "step": 824
+    },
+    {
+      "epoch": 3.3417721518987342,
+      "grad_norm": 0.7451730610341022,
+      "learning_rate": 1.1818627256573203e-05,
+      "loss": 0.4513,
+      "step": 825
+    },
+    {
+      "epoch": 3.3458227848101267,
+      "grad_norm": 0.24535078170741464,
+      "learning_rate": 1.1766867775842864e-05,
+      "loss": 0.4394,
+      "step": 826
+    },
+    {
+      "epoch": 3.349873417721519,
+      "grad_norm": 0.2491287344859943,
+      "learning_rate": 1.1715174603627615e-05,
+      "loss": 0.4306,
+      "step": 827
+    },
+    {
+      "epoch": 3.3539240506329113,
+      "grad_norm": 0.24734380333829714,
+      "learning_rate": 1.1663548156257147e-05,
+      "loss": 0.4413,
+      "step": 828
+    },
+    {
+      "epoch": 3.357974683544304,
+      "grad_norm": 0.24931659302656545,
+      "learning_rate": 1.161198884952377e-05,
+      "loss": 0.445,
+      "step": 829
+    },
+    {
+      "epoch": 3.3620253164556964,
+      "grad_norm": 0.2479727212618388,
+      "learning_rate": 1.1560497098679056e-05,
+      "loss": 0.4304,
+      "step": 830
+    },
+    {
+      "epoch": 3.3660759493670884,
+      "grad_norm": 0.23844520627030583,
+      "learning_rate": 1.1509073318430479e-05,
+      "loss": 0.4367,
+      "step": 831
+    },
+    {
+      "epoch": 3.370126582278481,
+      "grad_norm": 0.2512595824771267,
+      "learning_rate": 1.1457717922938116e-05,
+      "loss": 0.4404,
+      "step": 832
+    },
+    {
+      "epoch": 3.3741772151898735,
+      "grad_norm": 0.22613117974494454,
+      "learning_rate": 1.1406431325811233e-05,
+      "loss": 0.4401,
+      "step": 833
+    },
+    {
+      "epoch": 3.378227848101266,
+      "grad_norm": 0.22892908115885713,
+      "learning_rate": 1.135521394010506e-05,
+      "loss": 0.4368,
+      "step": 834
+    },
+    {
+      "epoch": 3.382278481012658,
+      "grad_norm": 0.3194146452421841,
+      "learning_rate": 1.1304066178317367e-05,
+      "loss": 0.4407,
+      "step": 835
+    },
+    {
+      "epoch": 3.3863291139240506,
+      "grad_norm": 0.2321210950427201,
+      "learning_rate": 1.1252988452385199e-05,
+      "loss": 0.4468,
+      "step": 836
+    },
+    {
+      "epoch": 3.390379746835443,
+      "grad_norm": 0.21793805256561724,
+      "learning_rate": 1.1201981173681536e-05,
+      "loss": 0.4339,
+      "step": 837
+    },
+    {
+      "epoch": 3.3944303797468356,
+      "grad_norm": 0.2244557876691824,
+      "learning_rate": 1.1151044753011991e-05,
+      "loss": 0.4349,
+      "step": 838
+    },
+    {
+      "epoch": 3.3984810126582277,
+      "grad_norm": 0.24006721033247913,
+      "learning_rate": 1.1100179600611491e-05,
+      "loss": 0.4312,
+      "step": 839
+    },
+    {
+      "epoch": 3.40253164556962,
+      "grad_norm": 0.2432592321228346,
+      "learning_rate": 1.1049386126140985e-05,
+      "loss": 0.42,
+      "step": 840
+    },
+    {
+      "epoch": 3.4065822784810127,
+      "grad_norm": 0.2480670745885521,
+      "learning_rate": 1.0998664738684128e-05,
+      "loss": 0.4263,
+      "step": 841
+    },
+    {
+      "epoch": 3.410632911392405,
+      "grad_norm": 0.25276207702728254,
+      "learning_rate": 1.0948015846744e-05,
+      "loss": 0.429,
+      "step": 842
+    },
+    {
+      "epoch": 3.4146835443037973,
+      "grad_norm": 0.22607075149549363,
+      "learning_rate": 1.0897439858239832e-05,
+      "loss": 0.4365,
+      "step": 843
+    },
+    {
+      "epoch": 3.41873417721519,
+      "grad_norm": 0.23778794662839584,
+      "learning_rate": 1.0846937180503652e-05,
+      "loss": 0.4446,
+      "step": 844
+    },
+    {
+      "epoch": 3.4227848101265823,
+      "grad_norm": 0.2371753886849899,
+      "learning_rate": 1.0796508220277117e-05,
+      "loss": 0.4439,
+      "step": 845
+    },
+    {
+      "epoch": 3.426835443037975,
+      "grad_norm": 0.22598441194013716,
+      "learning_rate": 1.0746153383708107e-05,
+      "loss": 0.4356,
+      "step": 846
+    },
+    {
+      "epoch": 3.430886075949367,
+      "grad_norm": 0.23463668639034727,
+      "learning_rate": 1.0695873076347579e-05,
+      "loss": 0.4433,
+      "step": 847
+    },
+    {
+      "epoch": 3.4349367088607594,
+      "grad_norm": 0.2529996286979705,
+      "learning_rate": 1.0645667703146205e-05,
+      "loss": 0.4339,
+      "step": 848
+    },
+    {
+      "epoch": 3.438987341772152,
+      "grad_norm": 0.23128490164593216,
+      "learning_rate": 1.0595537668451161e-05,
+      "loss": 0.428,
+      "step": 849
+    },
+    {
+      "epoch": 3.4430379746835444,
+      "grad_norm": 0.2339293328782258,
+      "learning_rate": 1.0545483376002854e-05,
+      "loss": 0.428,
+      "step": 850
+    },
+    {
+      "epoch": 3.4470886075949365,
+      "grad_norm": 0.23090762867222508,
+      "learning_rate": 1.0495505228931676e-05,
+      "loss": 0.4349,
+      "step": 851
+    },
+    {
+      "epoch": 3.451139240506329,
+      "grad_norm": 0.25926230534575706,
+      "learning_rate": 1.044560362975474e-05,
+      "loss": 0.4453,
+      "step": 852
+    },
+    {
+      "epoch": 3.4551898734177215,
+      "grad_norm": 0.2298099595292276,
+      "learning_rate": 1.0395778980372695e-05,
+      "loss": 0.4234,
+      "step": 853
+    },
+    {
+      "epoch": 3.459240506329114,
+      "grad_norm": 0.24643898795173722,
+      "learning_rate": 1.0346031682066381e-05,
+      "loss": 0.4237,
+      "step": 854
+    },
+    {
+      "epoch": 3.463291139240506,
+      "grad_norm": 0.2405481595681901,
+      "learning_rate": 1.0296362135493724e-05,
+      "loss": 0.4477,
+      "step": 855
+    },
+    {
+      "epoch": 3.4673417721518986,
+      "grad_norm": 0.247784524325911,
+      "learning_rate": 1.0246770740686422e-05,
+      "loss": 0.4522,
+      "step": 856
+    },
+    {
+      "epoch": 3.471392405063291,
+      "grad_norm": 0.23916340197085414,
+      "learning_rate": 1.0197257897046743e-05,
+      "loss": 0.4322,
+      "step": 857
+    },
+    {
+      "epoch": 3.4754430379746837,
+      "grad_norm": 0.22235566241066676,
+      "learning_rate": 1.014782400334433e-05,
+      "loss": 0.442,
+      "step": 858
+    },
+    {
+      "epoch": 3.479493670886076,
+      "grad_norm": 0.25471181692955375,
+      "learning_rate": 1.009846945771296e-05,
+      "loss": 0.4388,
+      "step": 859
+    },
+    {
+      "epoch": 3.4835443037974683,
+      "grad_norm": 0.23133450268973466,
+      "learning_rate": 1.0049194657647363e-05,
+      "loss": 0.424,
+      "step": 860
+    },
+    {
+      "epoch": 3.487594936708861,
+      "grad_norm": 0.26930543277787383,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.4283,
+      "step": 861
+    },
+    {
+      "epoch": 3.4916455696202533,
+      "grad_norm": 0.22659847458033638,
+      "learning_rate": 9.950885880977891e-06,
+      "loss": 0.4355,
+      "step": 862
+    },
+    {
+      "epoch": 3.4956962025316454,
+      "grad_norm": 0.23450962650308468,
+      "learning_rate": 9.901852696139382e-06,
+      "loss": 0.4397,
+      "step": 863
+    },
+    {
+      "epoch": 3.499746835443038,
+      "grad_norm": 0.25542496873503956,
+      "learning_rate": 9.852900840391027e-06,
+      "loss": 0.4218,
+      "step": 864
+    },
+    {
+      "epoch": 3.5037974683544304,
+      "grad_norm": 0.22317613352825574,
+      "learning_rate": 9.804030707984313e-06,
+      "loss": 0.4498,
+      "step": 865
+    },
+    {
+      "epoch": 3.507848101265823,
+      "grad_norm": 0.22908695834525852,
+      "learning_rate": 9.755242692512599e-06,
+      "loss": 0.4399,
+      "step": 866
+    },
+    {
+      "epoch": 3.5118987341772154,
+      "grad_norm": 0.22888339568377178,
+      "learning_rate": 9.70653718690782e-06,
+      "loss": 0.4317,
+      "step": 867
+    },
+    {
+      "epoch": 3.5159493670886075,
+      "grad_norm": 0.2625679989070159,
+      "learning_rate": 9.657914583437454e-06,
+      "loss": 0.455,
+      "step": 868
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.24444270062905643,
+      "learning_rate": 9.609375273701246e-06,
+      "loss": 0.427,
+      "step": 869
+    },
+    {
+      "epoch": 3.5240506329113925,
+      "grad_norm": 0.2276296573650595,
+      "learning_rate": 9.560919648628133e-06,
+      "loss": 0.4107,
+      "step": 870
+    },
+    {
+      "epoch": 3.5281012658227846,
+      "grad_norm": 0.23145900475495254,
+      "learning_rate": 9.512548098473047e-06,
+      "loss": 0.4618,
+      "step": 871
+    },
+    {
+      "epoch": 3.532151898734177,
+      "grad_norm": 0.2420372134330986,
+      "learning_rate": 9.464261012813825e-06,
+      "loss": 0.4275,
+      "step": 872
+    },
+    {
+      "epoch": 3.5362025316455696,
+      "grad_norm": 0.21692549075446732,
+      "learning_rate": 9.416058780547987e-06,
+      "loss": 0.4244,
+      "step": 873
+    },
+    {
+      "epoch": 3.540253164556962,
+      "grad_norm": 0.24392297180938743,
+      "learning_rate": 9.367941789889714e-06,
+      "loss": 0.4426,
+      "step": 874
+    },
+    {
+      "epoch": 3.5443037974683547,
+      "grad_norm": 0.331088991469201,
+      "learning_rate": 9.319910428366607e-06,
+      "loss": 0.4429,
+      "step": 875
+    },
+    {
+      "epoch": 3.5483544303797467,
+      "grad_norm": 0.23170361031134187,
+      "learning_rate": 9.271965082816667e-06,
+      "loss": 0.4415,
+      "step": 876
+    },
+    {
+      "epoch": 3.5524050632911393,
+      "grad_norm": 0.23142870046183459,
+      "learning_rate": 9.224106139385111e-06,
+      "loss": 0.4352,
+      "step": 877
+    },
+    {
+      "epoch": 3.5564556962025318,
+      "grad_norm": 0.25087776593403127,
+      "learning_rate": 9.176333983521291e-06,
+      "loss": 0.4309,
+      "step": 878
+    },
+    {
+      "epoch": 3.560506329113924,
+      "grad_norm": 0.24275482261801723,
+      "learning_rate": 9.12864899997558e-06,
+      "loss": 0.4329,
+      "step": 879
+    },
+    {
+      "epoch": 3.5645569620253164,
+      "grad_norm": 0.2319912948069167,
+      "learning_rate": 9.08105157279628e-06,
+      "loss": 0.4303,
+      "step": 880
+    },
+    {
+      "epoch": 3.568607594936709,
+      "grad_norm": 0.22094081707848126,
+      "learning_rate": 9.03354208532653e-06,
+      "loss": 0.4384,
+      "step": 881
+    },
+    {
+      "epoch": 3.5726582278481014,
+      "grad_norm": 0.2389739697547056,
+      "learning_rate": 8.986120920201205e-06,
+      "loss": 0.4532,
+      "step": 882
+    },
+    {
+      "epoch": 3.576708860759494,
+      "grad_norm": 0.22283704341002258,
+      "learning_rate": 8.938788459343852e-06,
+      "loss": 0.4229,
+      "step": 883
+    },
+    {
+      "epoch": 3.580759493670886,
+      "grad_norm": 0.2208936103655416,
+      "learning_rate": 8.8915450839636e-06,
+      "loss": 0.4333,
+      "step": 884
+    },
+    {
+      "epoch": 3.5848101265822785,
+      "grad_norm": 0.21901772588871224,
+      "learning_rate": 8.844391174552116e-06,
+      "loss": 0.4372,
+      "step": 885
+    },
+    {
+      "epoch": 3.588860759493671,
+      "grad_norm": 0.22152639821000533,
+      "learning_rate": 8.797327110880479e-06,
+      "loss": 0.4319,
+      "step": 886
+    },
+    {
+      "epoch": 3.592911392405063,
+      "grad_norm": 0.2378875252708083,
+      "learning_rate": 8.750353271996206e-06,
+      "loss": 0.4387,
+      "step": 887
+    },
+    {
+      "epoch": 3.5969620253164556,
+      "grad_norm": 0.23776250884930814,
+      "learning_rate": 8.703470036220132e-06,
+      "loss": 0.454,
+      "step": 888
+    },
+    {
+      "epoch": 3.601012658227848,
+      "grad_norm": 0.23590804573208565,
+      "learning_rate": 8.656677781143394e-06,
+      "loss": 0.4484,
+      "step": 889
+    },
+    {
+      "epoch": 3.6050632911392406,
+      "grad_norm": 0.2166370248259315,
+      "learning_rate": 8.609976883624377e-06,
+      "loss": 0.428,
+      "step": 890
+    },
+    {
+      "epoch": 3.609113924050633,
+      "grad_norm": 0.24406600502618456,
+      "learning_rate": 8.563367719785698e-06,
+      "loss": 0.4488,
+      "step": 891
+    },
+    {
+      "epoch": 3.613164556962025,
+      "grad_norm": 0.24115148559978083,
+      "learning_rate": 8.516850665011138e-06,
+      "loss": 0.4339,
+      "step": 892
+    },
+    {
+      "epoch": 3.6172151898734177,
+      "grad_norm": 0.2323811894314447,
+      "learning_rate": 8.47042609394269e-06,
+      "loss": 0.4512,
+      "step": 893
+    },
+    {
+      "epoch": 3.6212658227848102,
+      "grad_norm": 0.21327108748085902,
+      "learning_rate": 8.424094380477432e-06,
+      "loss": 0.4414,
+      "step": 894
+    },
+    {
+      "epoch": 3.6253164556962023,
+      "grad_norm": 0.23618841644853317,
+      "learning_rate": 8.37785589776465e-06,
+      "loss": 0.4456,
+      "step": 895
+    },
+    {
+      "epoch": 3.629367088607595,
+      "grad_norm": 0.22345917683168867,
+      "learning_rate": 8.331711018202694e-06,
+      "loss": 0.4422,
+      "step": 896
+    },
+    {
+      "epoch": 3.6334177215189873,
+      "grad_norm": 0.2141398783902164,
+      "learning_rate": 8.285660113436104e-06,
+      "loss": 0.4342,
+      "step": 897
+    },
+    {
+      "epoch": 3.63746835443038,
+      "grad_norm": 0.22010859077817935,
+      "learning_rate": 8.239703554352527e-06,
+      "loss": 0.4483,
+      "step": 898
+    },
+    {
+      "epoch": 3.6415189873417724,
+      "grad_norm": 0.2279122948943588,
+      "learning_rate": 8.193841711079775e-06,
+      "loss": 0.4287,
+      "step": 899
+    },
+    {
+      "epoch": 3.6455696202531644,
+      "grad_norm": 0.22711991489068956,
+      "learning_rate": 8.148074952982828e-06,
+      "loss": 0.4327,
+      "step": 900
+    },
+    {
+      "epoch": 3.649620253164557,
+      "grad_norm": 0.2149956786662109,
+      "learning_rate": 8.102403648660859e-06,
+      "loss": 0.4395,
+      "step": 901
+    },
+    {
+      "epoch": 3.6536708860759495,
+      "grad_norm": 0.23899319226769294,
+      "learning_rate": 8.056828165944282e-06,
+      "loss": 0.4225,
+      "step": 902
+    },
+    {
+      "epoch": 3.6577215189873415,
+      "grad_norm": 0.2154996550822818,
+      "learning_rate": 8.011348871891762e-06,
+      "loss": 0.4373,
+      "step": 903
+    },
+    {
+      "epoch": 3.661772151898734,
+      "grad_norm": 0.2217236469867085,
+      "learning_rate": 7.965966132787287e-06,
+      "loss": 0.4292,
+      "step": 904
+    },
+    {
+      "epoch": 3.6658227848101266,
+      "grad_norm": 0.2256959692845054,
+      "learning_rate": 7.920680314137189e-06,
+      "loss": 0.4383,
+      "step": 905
+    },
+    {
+      "epoch": 3.669873417721519,
+      "grad_norm": 0.22899639563586932,
+      "learning_rate": 7.875491780667246e-06,
+      "loss": 0.4363,
+      "step": 906
+    },
+    {
+      "epoch": 3.6739240506329116,
+      "grad_norm": 0.22272411309150816,
+      "learning_rate": 7.830400896319667e-06,
+      "loss": 0.4454,
+      "step": 907
+    },
+    {
+      "epoch": 3.6779746835443037,
+      "grad_norm": 0.23280240632818003,
+      "learning_rate": 7.785408024250259e-06,
+      "loss": 0.4313,
+      "step": 908
+    },
+    {
+      "epoch": 3.682025316455696,
+      "grad_norm": 0.230996205252856,
+      "learning_rate": 7.74051352682542e-06,
+      "loss": 0.4369,
+      "step": 909
+    },
+    {
+      "epoch": 3.6860759493670887,
+      "grad_norm": 0.22732652929456496,
+      "learning_rate": 7.695717765619257e-06,
+      "loss": 0.4277,
+      "step": 910
+    },
+    {
+      "epoch": 3.690126582278481,
+      "grad_norm": 0.2353599285391223,
+      "learning_rate": 7.651021101410673e-06,
+      "loss": 0.4078,
+      "step": 911
+    },
+    {
+      "epoch": 3.6941772151898733,
+      "grad_norm": 0.2278797586746193,
+      "learning_rate": 7.606423894180464e-06,
+      "loss": 0.4314,
+      "step": 912
+    },
+    {
+      "epoch": 3.698227848101266,
+      "grad_norm": 0.21609619167255956,
+      "learning_rate": 7.56192650310839e-06,
+      "loss": 0.438,
+      "step": 913
+    },
+    {
+      "epoch": 3.7022784810126583,
+      "grad_norm": 0.23370581384498834,
+      "learning_rate": 7.517529286570349e-06,
+      "loss": 0.4366,
+      "step": 914
+    },
+    {
+      "epoch": 3.706329113924051,
+      "grad_norm": 0.22391474750277374,
+      "learning_rate": 7.473232602135387e-06,
+      "loss": 0.4273,
+      "step": 915
+    },
+    {
+      "epoch": 3.710379746835443,
+      "grad_norm": 0.22368553920770398,
+      "learning_rate": 7.429036806562935e-06,
+      "loss": 0.4244,
+      "step": 916
+    },
+    {
+      "epoch": 3.7144303797468354,
+      "grad_norm": 0.22766670654059012,
+      "learning_rate": 7.3849422557998455e-06,
+      "loss": 0.4286,
+      "step": 917
+    },
+    {
+      "epoch": 3.718481012658228,
+      "grad_norm": 0.2333757012065019,
+      "learning_rate": 7.340949304977567e-06,
+      "loss": 0.4195,
+      "step": 918
+    },
+    {
+      "epoch": 3.72253164556962,
+      "grad_norm": 0.23590292026114595,
+      "learning_rate": 7.297058308409282e-06,
+      "loss": 0.4507,
+      "step": 919
+    },
+    {
+      "epoch": 3.7265822784810125,
+      "grad_norm": 0.23186479985601863,
+      "learning_rate": 7.25326961958704e-06,
+      "loss": 0.4234,
+      "step": 920
+    },
+    {
+      "epoch": 3.730632911392405,
+      "grad_norm": 0.22358519939204388,
+      "learning_rate": 7.209583591178921e-06,
+      "loss": 0.427,
+      "step": 921
+    },
+    {
+      "epoch": 3.7346835443037976,
+      "grad_norm": 0.22593832110298348,
+      "learning_rate": 7.1660005750261925e-06,
+      "loss": 0.4291,
+      "step": 922
+    },
+    {
+      "epoch": 3.73873417721519,
+      "grad_norm": 0.2406367931332586,
+      "learning_rate": 7.1225209221404765e-06,
+      "loss": 0.4398,
+      "step": 923
+    },
+    {
+      "epoch": 3.742784810126582,
+      "grad_norm": 0.21982472694430605,
+      "learning_rate": 7.079144982700909e-06,
+      "loss": 0.4476,
+      "step": 924
+    },
+    {
+      "epoch": 3.7468354430379747,
+      "grad_norm": 0.22440512612740252,
+      "learning_rate": 7.0358731060513695e-06,
+      "loss": 0.4211,
+      "step": 925
+    },
+    {
+      "epoch": 3.750886075949367,
+      "grad_norm": 0.22003753461372091,
+      "learning_rate": 6.99270564069757e-06,
+      "loss": 0.4377,
+      "step": 926
+    },
+    {
+      "epoch": 3.7549367088607593,
+      "grad_norm": 0.21622339771870716,
+      "learning_rate": 6.949642934304375e-06,
+      "loss": 0.4287,
+      "step": 927
+    },
+    {
+      "epoch": 3.7589873417721518,
+      "grad_norm": 0.2297566505445792,
+      "learning_rate": 6.906685333692871e-06,
+      "loss": 0.4415,
+      "step": 928
+    },
+    {
+      "epoch": 3.7630379746835443,
+      "grad_norm": 0.2283627078344015,
+      "learning_rate": 6.86383318483769e-06,
+      "loss": 0.4255,
+      "step": 929
+    },
+    {
+      "epoch": 3.767088607594937,
+      "grad_norm": 0.22470706694279788,
+      "learning_rate": 6.821086832864139e-06,
+      "loss": 0.4417,
+      "step": 930
+    },
+    {
+      "epoch": 3.7711392405063293,
+      "grad_norm": 0.24718472325677998,
+      "learning_rate": 6.77844662204546e-06,
+      "loss": 0.4169,
+      "step": 931
+    },
+    {
+      "epoch": 3.7751898734177214,
+      "grad_norm": 1.0756860147812457,
+      "learning_rate": 6.7359128958000455e-06,
+      "loss": 0.4574,
+      "step": 932
+    },
+    {
+      "epoch": 3.779240506329114,
+      "grad_norm": 0.22808890817961006,
+      "learning_rate": 6.693485996688695e-06,
+      "loss": 0.4346,
+      "step": 933
+    },
+    {
+      "epoch": 3.7832911392405064,
+      "grad_norm": 0.2190319878831937,
+      "learning_rate": 6.651166266411801e-06,
+      "loss": 0.4416,
+      "step": 934
+    },
+    {
+      "epoch": 3.7873417721518985,
+      "grad_norm": 0.22791698552106424,
+      "learning_rate": 6.6089540458066725e-06,
+      "loss": 0.4342,
+      "step": 935
+    },
+    {
+      "epoch": 3.791392405063291,
+      "grad_norm": 0.21132297780719464,
+      "learning_rate": 6.566849674844711e-06,
+      "loss": 0.4299,
+      "step": 936
+    },
+    {
+      "epoch": 3.7954430379746835,
+      "grad_norm": 0.22185241431382452,
+      "learning_rate": 6.524853492628747e-06,
+      "loss": 0.4391,
+      "step": 937
+    },
+    {
+      "epoch": 3.799493670886076,
+      "grad_norm": 0.2151262258198523,
+      "learning_rate": 6.4829658373902536e-06,
+      "loss": 0.4328,
+      "step": 938
+    },
+    {
+      "epoch": 3.8035443037974686,
+      "grad_norm": 0.22117323264053507,
+      "learning_rate": 6.441187046486648e-06,
+      "loss": 0.4389,
+      "step": 939
+    },
+    {
+      "epoch": 3.8075949367088606,
+      "grad_norm": 0.22880056252823155,
+      "learning_rate": 6.399517456398567e-06,
+      "loss": 0.4139,
+      "step": 940
+    },
+    {
+      "epoch": 3.811645569620253,
+      "grad_norm": 0.23318322810624187,
+      "learning_rate": 6.357957402727164e-06,
+      "loss": 0.4251,
+      "step": 941
+    },
+    {
+      "epoch": 3.8156962025316457,
+      "grad_norm": 0.2151789426617896,
+      "learning_rate": 6.316507220191395e-06,
+      "loss": 0.4407,
+      "step": 942
+    },
+    {
+      "epoch": 3.8197468354430377,
+      "grad_norm": 0.22694260898457233,
+      "learning_rate": 6.275167242625331e-06,
+      "loss": 0.4398,
+      "step": 943
+    },
+    {
+      "epoch": 3.8237974683544302,
+      "grad_norm": 0.20324129674921568,
+      "learning_rate": 6.233937802975471e-06,
+      "loss": 0.4311,
+      "step": 944
+    },
+    {
+      "epoch": 3.8278481012658228,
+      "grad_norm": 0.22288686771185795,
+      "learning_rate": 6.192819233298046e-06,
+      "loss": 0.435,
+      "step": 945
+    },
+    {
+      "epoch": 3.8318987341772153,
+      "grad_norm": 0.2076953992632956,
+      "learning_rate": 6.151811864756383e-06,
+      "loss": 0.4533,
+      "step": 946
+    },
+    {
+      "epoch": 3.835949367088608,
+      "grad_norm": 0.24310692736029557,
+      "learning_rate": 6.1109160276181655e-06,
+      "loss": 0.4297,
+      "step": 947
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.22514587563816796,
+      "learning_rate": 6.070132051252868e-06,
+      "loss": 0.4288,
+      "step": 948
+    },
+    {
+      "epoch": 3.8440506329113924,
+      "grad_norm": 0.24915213965466915,
+      "learning_rate": 6.0294602641290034e-06,
+      "loss": 0.4293,
+      "step": 949
+    },
+    {
+      "epoch": 3.848101265822785,
+      "grad_norm": 0.22810081798111068,
+      "learning_rate": 5.988900993811575e-06,
+      "loss": 0.4418,
+      "step": 950
+    },
+    {
+      "epoch": 3.852151898734177,
+      "grad_norm": 0.23177744607373482,
+      "learning_rate": 5.948454566959363e-06,
+      "loss": 0.4204,
+      "step": 951
+    },
+    {
+      "epoch": 3.8562025316455695,
+      "grad_norm": 0.6967527908486665,
+      "learning_rate": 5.908121309322328e-06,
+      "loss": 0.4336,
+      "step": 952
+    },
+    {
+      "epoch": 3.860253164556962,
+      "grad_norm": 0.20433726174108705,
+      "learning_rate": 5.867901545738976e-06,
+      "loss": 0.4409,
+      "step": 953
+    },
+    {
+      "epoch": 3.8643037974683545,
+      "grad_norm": 0.2169509489983157,
+      "learning_rate": 5.827795600133774e-06,
+      "loss": 0.4409,
+      "step": 954
+    },
+    {
+      "epoch": 3.868354430379747,
+      "grad_norm": 0.2377563600054345,
+      "learning_rate": 5.787803795514466e-06,
+      "loss": 0.4317,
+      "step": 955
+    },
+    {
+      "epoch": 3.872405063291139,
+      "grad_norm": 0.22427520141468954,
+      "learning_rate": 5.747926453969576e-06,
+      "loss": 0.4456,
+      "step": 956
+    },
+    {
+      "epoch": 3.8764556962025316,
+      "grad_norm": 0.22001973050749532,
+      "learning_rate": 5.708163896665708e-06,
+      "loss": 0.4159,
+      "step": 957
+    },
+    {
+      "epoch": 3.880506329113924,
+      "grad_norm": 0.23459941396598313,
+      "learning_rate": 5.668516443845047e-06,
+      "loss": 0.4438,
+      "step": 958
+    },
+    {
+      "epoch": 3.884556962025316,
+      "grad_norm": 0.22573734798488415,
+      "learning_rate": 5.6289844148227225e-06,
+      "loss": 0.4295,
+      "step": 959
+    },
+    {
+      "epoch": 3.8886075949367087,
+      "grad_norm": 0.23111106332736728,
+      "learning_rate": 5.5895681279842615e-06,
+      "loss": 0.4181,
+      "step": 960
+    },
+    {
+      "epoch": 3.8926582278481012,
+      "grad_norm": 0.21892411126493275,
+      "learning_rate": 5.550267900783019e-06,
+      "loss": 0.426,
+      "step": 961
+    },
+    {
+      "epoch": 3.8967088607594937,
+      "grad_norm": 0.24193606278913665,
+      "learning_rate": 5.511084049737623e-06,
+      "loss": 0.4357,
+      "step": 962
+    },
+    {
+      "epoch": 3.9007594936708863,
+      "grad_norm": 0.21840411443183994,
+      "learning_rate": 5.4720168904294215e-06,
+      "loss": 0.4345,
+      "step": 963
+    },
+    {
+      "epoch": 3.9048101265822783,
+      "grad_norm": 0.23188694797682186,
+      "learning_rate": 5.433066737499948e-06,
+      "loss": 0.4126,
+      "step": 964
+    },
+    {
+      "epoch": 3.908860759493671,
+      "grad_norm": 0.23234318834614573,
+      "learning_rate": 5.394233904648376e-06,
+      "loss": 0.4377,
+      "step": 965
+    },
+    {
+      "epoch": 3.9129113924050634,
+      "grad_norm": 0.21502346680863935,
+      "learning_rate": 5.355518704628997e-06,
+      "loss": 0.4403,
+      "step": 966
+    },
+    {
+      "epoch": 3.9169620253164554,
+      "grad_norm": 0.22333495235042333,
+      "learning_rate": 5.316921449248731e-06,
+      "loss": 0.4509,
+      "step": 967
+    },
+    {
+      "epoch": 3.921012658227848,
+      "grad_norm": 0.22728731631091265,
+      "learning_rate": 5.278442449364538e-06,
+      "loss": 0.4278,
+      "step": 968
+    },
+    {
+      "epoch": 3.9250632911392405,
+      "grad_norm": 0.22413707598349458,
+      "learning_rate": 5.240082014881016e-06,
+      "loss": 0.4218,
+      "step": 969
+    },
+    {
+      "epoch": 3.929113924050633,
+      "grad_norm": 0.22585958566748024,
+      "learning_rate": 5.201840454747822e-06,
+      "loss": 0.4326,
+      "step": 970
+    },
+    {
+      "epoch": 3.9331645569620255,
+      "grad_norm": 0.2386327872482809,
+      "learning_rate": 5.163718076957223e-06,
+      "loss": 0.4282,
+      "step": 971
+    },
+    {
+      "epoch": 3.9372151898734176,
+      "grad_norm": 0.22410683521458702,
+      "learning_rate": 5.125715188541609e-06,
+      "loss": 0.4414,
+      "step": 972
+    },
+    {
+      "epoch": 3.94126582278481,
+      "grad_norm": 0.22255415285578037,
+      "learning_rate": 5.087832095571021e-06,
+      "loss": 0.4432,
+      "step": 973
+    },
+    {
+      "epoch": 3.9453164556962026,
+      "grad_norm": 0.24223364809622028,
+      "learning_rate": 5.0500691031506766e-06,
+      "loss": 0.4442,
+      "step": 974
+    },
+    {
+      "epoch": 3.9493670886075947,
+      "grad_norm": 0.24821571792488745,
+      "learning_rate": 5.01242651541854e-06,
+      "loss": 0.427,
+      "step": 975
+    },
+    {
+      "epoch": 3.953417721518987,
+      "grad_norm": 0.21006415093023628,
+      "learning_rate": 4.974904635542815e-06,
+      "loss": 0.4312,
+      "step": 976
+    },
+    {
+      "epoch": 3.9574683544303797,
+      "grad_norm": 0.24396448821863756,
+      "learning_rate": 4.937503765719582e-06,
+      "loss": 0.4325,
+      "step": 977
+    },
+    {
+      "epoch": 3.961518987341772,
+      "grad_norm": 0.22346513916564936,
+      "learning_rate": 4.900224207170299e-06,
+      "loss": 0.4295,
+      "step": 978
+    },
+    {
+      "epoch": 3.9655696202531647,
+      "grad_norm": 0.20695230358951305,
+      "learning_rate": 4.8630662601394065e-06,
+      "loss": 0.4294,
+      "step": 979
+    },
+    {
+      "epoch": 3.969620253164557,
+      "grad_norm": 0.20694222623328004,
+      "learning_rate": 4.8260302238918995e-06,
+      "loss": 0.4406,
+      "step": 980
+    },
+    {
+      "epoch": 3.9736708860759493,
+      "grad_norm": 0.22409807511411772,
+      "learning_rate": 4.789116396710924e-06,
+      "loss": 0.4282,
+      "step": 981
+    },
+    {
+      "epoch": 3.977721518987342,
+      "grad_norm": 0.24313953095137128,
+      "learning_rate": 4.752325075895368e-06,
+      "loss": 0.445,
+      "step": 982
+    },
+    {
+      "epoch": 3.981772151898734,
+      "grad_norm": 0.2076609148410702,
+      "learning_rate": 4.715656557757473e-06,
+      "loss": 0.4283,
+      "step": 983
+    },
+    {
+      "epoch": 3.9858227848101264,
+      "grad_norm": 0.2448630108227433,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.4227,
+      "step": 984
+    },
+    {
+      "epoch": 3.989873417721519,
+      "grad_norm": 0.22088334230502513,
+      "learning_rate": 4.6426891098160585e-06,
+      "loss": 0.4359,
+      "step": 985
+    },
+    {
+      "epoch": 3.9939240506329114,
+      "grad_norm": 0.22248086247477292,
+      "learning_rate": 4.6063907676823474e-06,
+      "loss": 0.4345,
+      "step": 986
+    },
+    {
+      "epoch": 3.997974683544304,
+      "grad_norm": 0.21057772418774506,
+      "learning_rate": 4.570216403561141e-06,
+      "loss": 0.4449,
+      "step": 987
+    },
+    {
+      "epoch": 4.002025316455696,
+      "grad_norm": 0.27936470628305055,
+      "learning_rate": 4.534166308795815e-06,
+      "loss": 0.4148,
+      "step": 988
+    },
+    {
+      "epoch": 4.006075949367089,
+      "grad_norm": 0.34649012558800063,
+      "learning_rate": 4.498240773728859e-06,
+      "loss": 0.3914,
+      "step": 989
+    },
+    {
+      "epoch": 4.010126582278481,
+      "grad_norm": 0.2344059851013062,
+      "learning_rate": 4.462440087699609e-06,
+      "loss": 0.3878,
+      "step": 990
+    },
+    {
+      "epoch": 4.014177215189873,
+      "grad_norm": 0.3565465708165861,
+      "learning_rate": 4.426764539041861e-06,
+      "loss": 0.3955,
+      "step": 991
+    },
+    {
+      "epoch": 4.018227848101266,
+      "grad_norm": 0.41847414642809544,
+      "learning_rate": 4.391214415081582e-06,
+      "loss": 0.3741,
+      "step": 992
+    },
+    {
+      "epoch": 4.022278481012658,
+      "grad_norm": 0.2696360898616472,
+      "learning_rate": 4.355790002134579e-06,
+      "loss": 0.3803,
+      "step": 993
+    },
+    {
+      "epoch": 4.02632911392405,
+      "grad_norm": 0.3217687179389904,
+      "learning_rate": 4.320491585504207e-06,
+      "loss": 0.3969,
+      "step": 994
+    },
+    {
+      "epoch": 4.030379746835443,
+      "grad_norm": 0.3485491720428124,
+      "learning_rate": 4.2853194494790615e-06,
+      "loss": 0.4054,
+      "step": 995
+    },
+    {
+      "epoch": 4.034430379746835,
+      "grad_norm": 0.3112513627825323,
+      "learning_rate": 4.250273877330691e-06,
+      "loss": 0.3754,
+      "step": 996
+    },
+    {
+      "epoch": 4.038481012658228,
+      "grad_norm": 0.24126623650164705,
+      "learning_rate": 4.215355151311313e-06,
+      "loss": 0.4111,
+      "step": 997
+    },
+    {
+      "epoch": 4.04253164556962,
+      "grad_norm": 0.29193703751689226,
+      "learning_rate": 4.180563552651542e-06,
+      "loss": 0.3983,
+      "step": 998
+    },
+    {
+      "epoch": 4.046582278481012,
+      "grad_norm": 0.31819694867131804,
+      "learning_rate": 4.145899361558147e-06,
+      "loss": 0.3807,
+      "step": 999
+    },
+    {
+      "epoch": 4.050632911392405,
+      "grad_norm": 0.2606446834407702,
+      "learning_rate": 4.111362857211738e-06,
+      "loss": 0.3931,
+      "step": 1000
+    },
+    {
+      "epoch": 4.054683544303797,
+      "grad_norm": 0.23040430309731336,
+      "learning_rate": 4.076954317764592e-06,
+      "loss": 0.3791,
+      "step": 1001
+    },
+    {
+      "epoch": 4.0587341772151895,
+      "grad_norm": 0.26610501987071244,
+      "learning_rate": 4.042674020338335e-06,
+      "loss": 0.4032,
+      "step": 1002
+    },
+    {
+      "epoch": 4.062784810126582,
+      "grad_norm": 0.26733340481487705,
+      "learning_rate": 4.0085222410217835e-06,
+      "loss": 0.3936,
+      "step": 1003
+    },
+    {
+      "epoch": 4.0668354430379745,
+      "grad_norm": 0.25223693209182396,
+      "learning_rate": 3.974499254868674e-06,
+      "loss": 0.3989,
+      "step": 1004
+    },
+    {
+      "epoch": 4.0708860759493675,
+      "grad_norm": 0.235986077333984,
+      "learning_rate": 3.940605335895451e-06,
+      "loss": 0.4108,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0749367088607595,
+      "grad_norm": 0.24033551312796064,
+      "learning_rate": 3.90684075707908e-06,
+      "loss": 0.3826,
+      "step": 1006
+    },
+    {
+      "epoch": 4.078987341772152,
+      "grad_norm": 0.2436843342693253,
+      "learning_rate": 3.8732057903548505e-06,
+      "loss": 0.3919,
+      "step": 1007
+    },
+    {
+      "epoch": 4.083037974683545,
+      "grad_norm": 0.23479344901799487,
+      "learning_rate": 3.8397007066141375e-06,
+      "loss": 0.4037,
+      "step": 1008
+    },
+    {
+      "epoch": 4.087088607594937,
+      "grad_norm": 0.22333538672269826,
+      "learning_rate": 3.806325775702304e-06,
+      "loss": 0.3866,
+      "step": 1009
+    },
+    {
+      "epoch": 4.091139240506329,
+      "grad_norm": 0.2261182551243623,
+      "learning_rate": 3.773081266416434e-06,
+      "loss": 0.3867,
+      "step": 1010
+    },
+    {
+      "epoch": 4.095189873417722,
+      "grad_norm": 0.24357638011686214,
+      "learning_rate": 3.739967446503245e-06,
+      "loss": 0.4015,
+      "step": 1011
+    },
+    {
+      "epoch": 4.099240506329114,
+      "grad_norm": 0.21276282332532367,
+      "learning_rate": 3.706984582656894e-06,
+      "loss": 0.3932,
+      "step": 1012
+    },
+    {
+      "epoch": 4.103291139240507,
+      "grad_norm": 0.23299985020152772,
+      "learning_rate": 3.6741329405168237e-06,
+      "loss": 0.396,
+      "step": 1013
+    },
+    {
+      "epoch": 4.107341772151899,
+      "grad_norm": 0.21584885267971365,
+      "learning_rate": 3.641412784665648e-06,
+      "loss": 0.3938,
+      "step": 1014
+    },
+    {
+      "epoch": 4.111392405063291,
+      "grad_norm": 0.229827668855286,
+      "learning_rate": 3.608824378627005e-06,
+      "loss": 0.3928,
+      "step": 1015
+    },
+    {
+      "epoch": 4.115443037974684,
+      "grad_norm": 0.3062374328438704,
+      "learning_rate": 3.5763679848634337e-06,
+      "loss": 0.405,
+      "step": 1016
+    },
+    {
+      "epoch": 4.119493670886076,
+      "grad_norm": 0.21814313804331079,
+      "learning_rate": 3.544043864774269e-06,
+      "loss": 0.3825,
+      "step": 1017
+    },
+    {
+      "epoch": 4.123544303797468,
+      "grad_norm": 0.2036499543468261,
+      "learning_rate": 3.5118522786935282e-06,
+      "loss": 0.4087,
+      "step": 1018
+    },
+    {
+      "epoch": 4.127594936708861,
+      "grad_norm": 0.21832515185907797,
+      "learning_rate": 3.479793485887819e-06,
+      "loss": 0.4042,
+      "step": 1019
+    },
+    {
+      "epoch": 4.131645569620253,
+      "grad_norm": 0.22894487926924217,
+      "learning_rate": 3.4478677445542653e-06,
+      "loss": 0.4015,
+      "step": 1020
+    },
+    {
+      "epoch": 4.135696202531646,
+      "grad_norm": 0.213975361861999,
+      "learning_rate": 3.4160753118183767e-06,
+      "loss": 0.3858,
+      "step": 1021
+    },
+    {
+      "epoch": 4.139746835443038,
+      "grad_norm": 0.20356939354704023,
+      "learning_rate": 3.3844164437320527e-06,
+      "loss": 0.4045,
+      "step": 1022
+    },
+    {
+      "epoch": 4.14379746835443,
+      "grad_norm": 0.21929758052995293,
+      "learning_rate": 3.3528913952714558e-06,
+      "loss": 0.4052,
+      "step": 1023
+    },
+    {
+      "epoch": 4.147848101265823,
+      "grad_norm": 0.21807965459207312,
+      "learning_rate": 3.321500420335e-06,
+      "loss": 0.3939,
+      "step": 1024
+    },
+    {
+      "epoch": 4.151898734177215,
+      "grad_norm": 0.2022446929729402,
+      "learning_rate": 3.290243771741275e-06,
+      "loss": 0.4005,
+      "step": 1025
+    },
+    {
+      "epoch": 4.155949367088607,
+      "grad_norm": 0.2072637499708677,
+      "learning_rate": 3.2591217012270325e-06,
+      "loss": 0.4024,
+      "step": 1026
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.21640520663925777,
+      "learning_rate": 3.228134459445149e-06,
+      "loss": 0.3831,
+      "step": 1027
+    },
+    {
+      "epoch": 4.164050632911392,
+      "grad_norm": 0.21800501325377117,
+      "learning_rate": 3.1972822959626205e-06,
+      "loss": 0.3871,
+      "step": 1028
+    },
+    {
+      "epoch": 4.168101265822785,
+      "grad_norm": 0.2136254400478826,
+      "learning_rate": 3.166565459258513e-06,
+      "loss": 0.3961,
+      "step": 1029
+    },
+    {
+      "epoch": 4.172151898734177,
+      "grad_norm": 0.21245926459871498,
+      "learning_rate": 3.1359841967220193e-06,
+      "loss": 0.3835,
+      "step": 1030
+    },
+    {
+      "epoch": 4.176202531645569,
+      "grad_norm": 0.2035551700647929,
+      "learning_rate": 3.105538754650419e-06,
+      "loss": 0.3926,
+      "step": 1031
+    },
+    {
+      "epoch": 4.180253164556962,
+      "grad_norm": 0.3020278868582561,
+      "learning_rate": 3.07522937824712e-06,
+      "loss": 0.4013,
+      "step": 1032
+    },
+    {
+      "epoch": 4.184303797468354,
+      "grad_norm": 0.20755477167534905,
+      "learning_rate": 3.0450563116196697e-06,
+      "loss": 0.3936,
+      "step": 1033
+    },
+    {
+      "epoch": 4.188354430379746,
+      "grad_norm": 0.20816660361723296,
+      "learning_rate": 3.0150197977778008e-06,
+      "loss": 0.3903,
+      "step": 1034
+    },
+    {
+      "epoch": 4.192405063291139,
+      "grad_norm": 0.2156193783360792,
+      "learning_rate": 2.985120078631465e-06,
+      "loss": 0.3983,
+      "step": 1035
+    },
+    {
+      "epoch": 4.1964556962025314,
+      "grad_norm": 0.7984063563270593,
+      "learning_rate": 2.9553573949888893e-06,
+      "loss": 0.4019,
+      "step": 1036
+    },
+    {
+      "epoch": 4.200506329113924,
+      "grad_norm": 0.20348266025608333,
+      "learning_rate": 2.9257319865546384e-06,
+      "loss": 0.4159,
+      "step": 1037
+    },
+    {
+      "epoch": 4.2045569620253165,
+      "grad_norm": 0.21708227459072632,
+      "learning_rate": 2.896244091927678e-06,
+      "loss": 0.3809,
+      "step": 1038
+    },
+    {
+      "epoch": 4.2086075949367086,
+      "grad_norm": 0.20020203831651787,
+      "learning_rate": 2.8668939485994584e-06,
+      "loss": 0.4004,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2126582278481015,
+      "grad_norm": 0.20096373123766048,
+      "learning_rate": 2.837681792951994e-06,
+      "loss": 0.3982,
+      "step": 1040
+    },
+    {
+      "epoch": 4.216708860759494,
+      "grad_norm": 0.20906931847279614,
+      "learning_rate": 2.808607860255981e-06,
+      "loss": 0.3895,
+      "step": 1041
+    },
+    {
+      "epoch": 4.220759493670886,
+      "grad_norm": 0.20228533967126333,
+      "learning_rate": 2.7796723846688634e-06,
+      "loss": 0.3944,
+      "step": 1042
+    },
+    {
+      "epoch": 4.224810126582279,
+      "grad_norm": 0.2092331473061436,
+      "learning_rate": 2.7508755992329937e-06,
+      "loss": 0.3928,
+      "step": 1043
+    },
+    {
+      "epoch": 4.228860759493671,
+      "grad_norm": 0.2020075775547855,
+      "learning_rate": 2.722217735873718e-06,
+      "loss": 0.3983,
+      "step": 1044
+    },
+    {
+      "epoch": 4.232911392405064,
+      "grad_norm": 0.20652686225886843,
+      "learning_rate": 2.6936990253975315e-06,
+      "loss": 0.3843,
+      "step": 1045
+    },
+    {
+      "epoch": 4.236962025316456,
+      "grad_norm": 0.20412521885236318,
+      "learning_rate": 2.665319697490205e-06,
+      "loss": 0.3895,
+      "step": 1046
+    },
+    {
+      "epoch": 4.241012658227848,
+      "grad_norm": 0.20069776654969554,
+      "learning_rate": 2.637079980714945e-06,
+      "loss": 0.3789,
+      "step": 1047
+    },
+    {
+      "epoch": 4.245063291139241,
+      "grad_norm": 0.20634711112781629,
+      "learning_rate": 2.6089801025105453e-06,
+      "loss": 0.4086,
+      "step": 1048
+    },
+    {
+      "epoch": 4.249113924050633,
+      "grad_norm": 0.19616103227910603,
+      "learning_rate": 2.581020289189571e-06,
+      "loss": 0.3948,
+      "step": 1049
+    },
+    {
+      "epoch": 4.253164556962025,
+      "grad_norm": 0.2080166171912427,
+      "learning_rate": 2.553200765936501e-06,
+      "loss": 0.3693,
+      "step": 1050
+    },
+    {
+      "epoch": 4.257215189873418,
+      "grad_norm": 0.1983874296729655,
+      "learning_rate": 2.525521756805962e-06,
+      "loss": 0.4014,
+      "step": 1051
+    },
+    {
+      "epoch": 4.26126582278481,
+      "grad_norm": 0.20714797193401588,
+      "learning_rate": 2.497983484720885e-06,
+      "loss": 0.3733,
+      "step": 1052
+    },
+    {
+      "epoch": 4.265316455696203,
+      "grad_norm": 0.2213411475312191,
+      "learning_rate": 2.470586171470728e-06,
+      "loss": 0.3857,
+      "step": 1053
+    },
+    {
+      "epoch": 4.269367088607595,
+      "grad_norm": 0.19751128617723765,
+      "learning_rate": 2.4433300377096836e-06,
+      "loss": 0.3843,
+      "step": 1054
+    },
+    {
+      "epoch": 4.273417721518987,
+      "grad_norm": 0.20356248727947313,
+      "learning_rate": 2.4162153029549073e-06,
+      "loss": 0.3905,
+      "step": 1055
+    },
+    {
+      "epoch": 4.27746835443038,
+      "grad_norm": 0.19865300082413656,
+      "learning_rate": 2.3892421855847458e-06,
+      "loss": 0.402,
+      "step": 1056
+    },
+    {
+      "epoch": 4.281518987341772,
+      "grad_norm": 0.20752136879626104,
+      "learning_rate": 2.362410902836978e-06,
+      "loss": 0.3845,
+      "step": 1057
+    },
+    {
+      "epoch": 4.285569620253165,
+      "grad_norm": 0.20494914867673356,
+      "learning_rate": 2.3357216708070653e-06,
+      "loss": 0.3889,
+      "step": 1058
+    },
+    {
+      "epoch": 4.289620253164557,
+      "grad_norm": 0.20435042296465927,
+      "learning_rate": 2.309174704446411e-06,
+      "loss": 0.3901,
+      "step": 1059
+    },
+    {
+      "epoch": 4.293670886075949,
+      "grad_norm": 0.19871637519605256,
+      "learning_rate": 2.2827702175606437e-06,
+      "loss": 0.395,
+      "step": 1060
+    },
+    {
+      "epoch": 4.297721518987342,
+      "grad_norm": 0.1992147281422628,
+      "learning_rate": 2.256508422807855e-06,
+      "loss": 0.405,
+      "step": 1061
+    },
+    {
+      "epoch": 4.301772151898734,
+      "grad_norm": 0.19353631853136016,
+      "learning_rate": 2.230389531696946e-06,
+      "loss": 0.3928,
+      "step": 1062
+    },
+    {
+      "epoch": 4.305822784810126,
+      "grad_norm": 0.20335742104142693,
+      "learning_rate": 2.204413754585857e-06,
+      "loss": 0.4066,
+      "step": 1063
+    },
+    {
+      "epoch": 4.309873417721519,
+      "grad_norm": 0.19490759628672547,
+      "learning_rate": 2.1785813006799406e-06,
+      "loss": 0.3892,
+      "step": 1064
+    },
+    {
+      "epoch": 4.313924050632911,
+      "grad_norm": 0.19881799229591432,
+      "learning_rate": 2.1528923780302224e-06,
+      "loss": 0.3916,
+      "step": 1065
+    },
+    {
+      "epoch": 4.317974683544303,
+      "grad_norm": 0.20698600113995647,
+      "learning_rate": 2.127347193531757e-06,
+      "loss": 0.3678,
+      "step": 1066
+    },
+    {
+      "epoch": 4.322025316455696,
+      "grad_norm": 0.20594573160246835,
+      "learning_rate": 2.101945952921942e-06,
+      "loss": 0.405,
+      "step": 1067
+    },
+    {
+      "epoch": 4.326075949367088,
+      "grad_norm": 0.19919841598124216,
+      "learning_rate": 2.0766888607788906e-06,
+      "loss": 0.4008,
+      "step": 1068
+    },
+    {
+      "epoch": 4.330126582278481,
+      "grad_norm": 0.19391645081841222,
+      "learning_rate": 2.0515761205197337e-06,
+      "loss": 0.4039,
+      "step": 1069
+    },
+    {
+      "epoch": 4.334177215189873,
+      "grad_norm": 0.19939578840126507,
+      "learning_rate": 2.0266079343990453e-06,
+      "loss": 0.3984,
+      "step": 1070
+    },
+    {
+      "epoch": 4.3382278481012655,
+      "grad_norm": 0.20202148208661755,
+      "learning_rate": 2.0017845035071494e-06,
+      "loss": 0.3891,
+      "step": 1071
+    },
+    {
+      "epoch": 4.3422784810126585,
+      "grad_norm": 0.20435778522981732,
+      "learning_rate": 1.9771060277685537e-06,
+      "loss": 0.3877,
+      "step": 1072
+    },
+    {
+      "epoch": 4.3463291139240505,
+      "grad_norm": 0.19287795484330797,
+      "learning_rate": 1.95257270594031e-06,
+      "loss": 0.4069,
+      "step": 1073
+    },
+    {
+      "epoch": 4.3503797468354435,
+      "grad_norm": 0.19857387567321036,
+      "learning_rate": 1.9281847356104188e-06,
+      "loss": 0.3888,
+      "step": 1074
+    },
+    {
+      "epoch": 4.3544303797468356,
+      "grad_norm": 0.20007418094682786,
+      "learning_rate": 1.9039423131962365e-06,
+      "loss": 0.396,
+      "step": 1075
+    },
+    {
+      "epoch": 4.358481012658228,
+      "grad_norm": 0.21327998178830743,
+      "learning_rate": 1.8798456339429027e-06,
+      "loss": 0.3759,
+      "step": 1076
+    },
+    {
+      "epoch": 4.362531645569621,
+      "grad_norm": 0.20164572807050724,
+      "learning_rate": 1.8558948919217612e-06,
+      "loss": 0.3841,
+      "step": 1077
+    },
+    {
+      "epoch": 4.366582278481013,
+      "grad_norm": 0.19886855940612602,
+      "learning_rate": 1.8320902800287954e-06,
+      "loss": 0.3867,
+      "step": 1078
+    },
+    {
+      "epoch": 4.370632911392405,
+      "grad_norm": 0.191969394466545,
+      "learning_rate": 1.8084319899830726e-06,
+      "loss": 0.3843,
+      "step": 1079
+    },
+    {
+      "epoch": 4.374683544303798,
+      "grad_norm": 0.2040898307209935,
+      "learning_rate": 1.7849202123252097e-06,
+      "loss": 0.3948,
+      "step": 1080
+    },
+    {
+      "epoch": 4.37873417721519,
+      "grad_norm": 0.2026571461353361,
+      "learning_rate": 1.7615551364158401e-06,
+      "loss": 0.3879,
+      "step": 1081
+    },
+    {
+      "epoch": 4.382784810126582,
+      "grad_norm": 0.18595678340372856,
+      "learning_rate": 1.738336950434061e-06,
+      "loss": 0.395,
+      "step": 1082
+    },
+    {
+      "epoch": 4.386835443037975,
+      "grad_norm": 0.2036656206313868,
+      "learning_rate": 1.715265841375957e-06,
+      "loss": 0.3922,
+      "step": 1083
+    },
+    {
+      "epoch": 4.390886075949367,
+      "grad_norm": 0.2006621349442602,
+      "learning_rate": 1.6923419950530684e-06,
+      "loss": 0.3982,
+      "step": 1084
+    },
+    {
+      "epoch": 4.39493670886076,
+      "grad_norm": 0.19884690619626794,
+      "learning_rate": 1.6695655960909008e-06,
+      "loss": 0.387,
+      "step": 1085
+    },
+    {
+      "epoch": 4.398987341772152,
+      "grad_norm": 0.19517162445033356,
+      "learning_rate": 1.646936827927441e-06,
+      "loss": 0.3865,
+      "step": 1086
+    },
+    {
+      "epoch": 4.403037974683544,
+      "grad_norm": 0.20284322194549262,
+      "learning_rate": 1.6244558728116766e-06,
+      "loss": 0.3937,
+      "step": 1087
+    },
+    {
+      "epoch": 4.407088607594937,
+      "grad_norm": 0.2017227812037139,
+      "learning_rate": 1.6021229118021265e-06,
+      "loss": 0.3769,
+      "step": 1088
+    },
+    {
+      "epoch": 4.411139240506329,
+      "grad_norm": 0.19469517789716137,
+      "learning_rate": 1.5799381247653967e-06,
+      "loss": 0.3914,
+      "step": 1089
+    },
+    {
+      "epoch": 4.415189873417722,
+      "grad_norm": 0.1990601730581273,
+      "learning_rate": 1.5579016903747013e-06,
+      "loss": 0.3964,
+      "step": 1090
+    },
+    {
+      "epoch": 4.419240506329114,
+      "grad_norm": 0.20459361393342906,
+      "learning_rate": 1.5360137861084656e-06,
+      "loss": 0.3893,
+      "step": 1091
+    },
+    {
+      "epoch": 4.423291139240506,
+      "grad_norm": 0.20394260747206805,
+      "learning_rate": 1.5142745882488475e-06,
+      "loss": 0.3786,
+      "step": 1092
+    },
+    {
+      "epoch": 4.427341772151899,
+      "grad_norm": 0.20074690429751338,
+      "learning_rate": 1.4926842718803691e-06,
+      "loss": 0.3928,
+      "step": 1093
+    },
+    {
+      "epoch": 4.431392405063291,
+      "grad_norm": 0.469926681494236,
+      "learning_rate": 1.4712430108884657e-06,
+      "loss": 0.3803,
+      "step": 1094
+    },
+    {
+      "epoch": 4.435443037974683,
+      "grad_norm": 0.19742099205989388,
+      "learning_rate": 1.4499509779581078e-06,
+      "loss": 0.39,
+      "step": 1095
+    },
+    {
+      "epoch": 4.439493670886076,
+      "grad_norm": 0.19636580623834166,
+      "learning_rate": 1.4288083445723988e-06,
+      "loss": 0.3743,
+      "step": 1096
+    },
+    {
+      "epoch": 4.443544303797468,
+      "grad_norm": 0.20196779046126348,
+      "learning_rate": 1.4078152810112045e-06,
+      "loss": 0.3875,
+      "step": 1097
+    },
+    {
+      "epoch": 4.44759493670886,
+      "grad_norm": 0.18936408538911,
+      "learning_rate": 1.3869719563497697e-06,
+      "loss": 0.386,
+      "step": 1098
+    },
+    {
+      "epoch": 4.451645569620253,
+      "grad_norm": 0.20093875659125568,
+      "learning_rate": 1.3662785384573663e-06,
+      "loss": 0.3958,
+      "step": 1099
+    },
+    {
+      "epoch": 4.455696202531645,
+      "grad_norm": 0.19559264669193077,
+      "learning_rate": 1.3457351939959383e-06,
+      "loss": 0.3849,
+      "step": 1100
+    },
+    {
+      "epoch": 4.459746835443038,
+      "grad_norm": 0.20177420257827253,
+      "learning_rate": 1.3253420884187551e-06,
+      "loss": 0.3927,
+      "step": 1101
+    },
+    {
+      "epoch": 4.46379746835443,
+      "grad_norm": 0.19630302802473543,
+      "learning_rate": 1.3050993859690953e-06,
+      "loss": 0.3971,
+      "step": 1102
+    },
+    {
+      "epoch": 4.467848101265822,
+      "grad_norm": 0.19747588143590947,
+      "learning_rate": 1.2850072496788869e-06,
+      "loss": 0.392,
+      "step": 1103
+    },
+    {
+      "epoch": 4.471898734177215,
+      "grad_norm": 0.20103898727546066,
+      "learning_rate": 1.2650658413674434e-06,
+      "loss": 0.387,
+      "step": 1104
+    },
+    {
+      "epoch": 4.4759493670886075,
+      "grad_norm": 0.20801359259633947,
+      "learning_rate": 1.2452753216401226e-06,
+      "loss": 0.3832,
+      "step": 1105
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.2041047339182285,
+      "learning_rate": 1.2256358498870503e-06,
+      "loss": 0.3939,
+      "step": 1106
+    },
+    {
+      "epoch": 4.4840506329113925,
+      "grad_norm": 0.2092170375149407,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.3993,
+      "step": 1107
+    },
+    {
+      "epoch": 4.488101265822785,
+      "grad_norm": 0.20070658046673998,
+      "learning_rate": 1.1868106817802816e-06,
+      "loss": 0.3806,
+      "step": 1108
+    },
+    {
+      "epoch": 4.4921518987341775,
+      "grad_norm": 0.19571943575330897,
+      "learning_rate": 1.1676252981191482e-06,
+      "loss": 0.3952,
+      "step": 1109
+    },
+    {
+      "epoch": 4.49620253164557,
+      "grad_norm": 0.19748335568242872,
+      "learning_rate": 1.1485915878148823e-06,
+      "loss": 0.3855,
+      "step": 1110
+    },
+    {
+      "epoch": 4.500253164556962,
+      "grad_norm": 0.20314383687360948,
+      "learning_rate": 1.1297097041623584e-06,
+      "loss": 0.3945,
+      "step": 1111
+    },
+    {
+      "epoch": 4.504303797468355,
+      "grad_norm": 0.1937602251258363,
+      "learning_rate": 1.1109797992336847e-06,
+      "loss": 0.4016,
+      "step": 1112
+    },
+    {
+      "epoch": 4.508354430379747,
+      "grad_norm": 0.2017508371247786,
+      "learning_rate": 1.092402023876933e-06,
+      "loss": 0.3874,
+      "step": 1113
+    },
+    {
+      "epoch": 4.512405063291139,
+      "grad_norm": 0.20286983416430618,
+      "learning_rate": 1.0739765277149527e-06,
+      "loss": 0.3959,
+      "step": 1114
+    },
+    {
+      "epoch": 4.516455696202532,
+      "grad_norm": 0.19561473040426602,
+      "learning_rate": 1.0557034591441596e-06,
+      "loss": 0.4006,
+      "step": 1115
+    },
+    {
+      "epoch": 4.520506329113924,
+      "grad_norm": 0.19706381203360074,
+      "learning_rate": 1.0375829653333324e-06,
+      "loss": 0.3924,
+      "step": 1116
+    },
+    {
+      "epoch": 4.524556962025317,
+      "grad_norm": 0.19686981068567305,
+      "learning_rate": 1.0196151922224385e-06,
+      "loss": 0.4044,
+      "step": 1117
+    },
+    {
+      "epoch": 4.528607594936709,
+      "grad_norm": 0.20100623829748446,
+      "learning_rate": 1.0018002845214526e-06,
+      "loss": 0.3727,
+      "step": 1118
+    },
+    {
+      "epoch": 4.532658227848101,
+      "grad_norm": 0.20297365034209036,
+      "learning_rate": 9.841383857091947e-07,
+      "loss": 0.3937,
+      "step": 1119
+    },
+    {
+      "epoch": 4.536708860759494,
+      "grad_norm": 0.19714757854025036,
+      "learning_rate": 9.666296380321616e-07,
+      "loss": 0.4024,
+      "step": 1120
+    },
+    {
+      "epoch": 4.540759493670886,
+      "grad_norm": 0.2043240868148635,
+      "learning_rate": 9.492741825034124e-07,
+      "loss": 0.3985,
+      "step": 1121
+    },
+    {
+      "epoch": 4.544810126582279,
+      "grad_norm": 0.20401025192876052,
+      "learning_rate": 9.320721589013892e-07,
+      "loss": 0.4027,
+      "step": 1122
+    },
+    {
+      "epoch": 4.548860759493671,
+      "grad_norm": 0.19472211487885052,
+      "learning_rate": 9.150237057688339e-07,
+      "loss": 0.4009,
+      "step": 1123
+    },
+    {
+      "epoch": 4.552911392405063,
+      "grad_norm": 0.19440640290564473,
+      "learning_rate": 8.981289604116328e-07,
+      "loss": 0.3901,
+      "step": 1124
+    },
+    {
+      "epoch": 4.556962025316456,
+      "grad_norm": 0.19507082407922052,
+      "learning_rate": 8.813880588977542e-07,
+      "loss": 0.3875,
+      "step": 1125
+    },
+    {
+      "epoch": 4.561012658227848,
+      "grad_norm": 0.1977261408322341,
+      "learning_rate": 8.648011360561126e-07,
+      "loss": 0.398,
+      "step": 1126
+    },
+    {
+      "epoch": 4.56506329113924,
+      "grad_norm": 0.2059865547633789,
+      "learning_rate": 8.483683254755037e-07,
+      "loss": 0.4135,
+      "step": 1127
+    },
+    {
+      "epoch": 4.569113924050633,
+      "grad_norm": 0.1999555654646436,
+      "learning_rate": 8.320897595035227e-07,
+      "loss": 0.3914,
+      "step": 1128
+    },
+    {
+      "epoch": 4.573164556962025,
+      "grad_norm": 0.21643155267221456,
+      "learning_rate": 8.159655692455093e-07,
+      "loss": 0.3815,
+      "step": 1129
+    },
+    {
+      "epoch": 4.577215189873417,
+      "grad_norm": 0.19565653449279782,
+      "learning_rate": 7.999958845634648e-07,
+      "loss": 0.3985,
+      "step": 1130
+    },
+    {
+      "epoch": 4.58126582278481,
+      "grad_norm": 0.20454705699542008,
+      "learning_rate": 7.841808340750478e-07,
+      "loss": 0.4,
+      "step": 1131
+    },
+    {
+      "epoch": 4.585316455696202,
+      "grad_norm": 0.20239005409045824,
+      "learning_rate": 7.685205451524869e-07,
+      "loss": 0.3851,
+      "step": 1132
+    },
+    {
+      "epoch": 4.589367088607595,
+      "grad_norm": 0.19722401171372572,
+      "learning_rate": 7.530151439216027e-07,
+      "loss": 0.3949,
+      "step": 1133
+    },
+    {
+      "epoch": 4.593417721518987,
+      "grad_norm": 0.19942737898431534,
+      "learning_rate": 7.376647552607675e-07,
+      "loss": 0.3967,
+      "step": 1134
+    },
+    {
+      "epoch": 4.597468354430379,
+      "grad_norm": 0.19986712731106215,
+      "learning_rate": 7.224695027998963e-07,
+      "loss": 0.396,
+      "step": 1135
+    },
+    {
+      "epoch": 4.601518987341772,
+      "grad_norm": 0.1916704856654779,
+      "learning_rate": 7.07429508919466e-07,
+      "loss": 0.3918,
+      "step": 1136
+    },
+    {
+      "epoch": 4.605569620253164,
+      "grad_norm": 0.19861736273658848,
+      "learning_rate": 6.925448947495206e-07,
+      "loss": 0.3886,
+      "step": 1137
+    },
+    {
+      "epoch": 4.609620253164557,
+      "grad_norm": 0.19402214755751754,
+      "learning_rate": 6.778157801686936e-07,
+      "loss": 0.3933,
+      "step": 1138
+    },
+    {
+      "epoch": 4.613670886075949,
+      "grad_norm": 0.1967852382078503,
+      "learning_rate": 6.632422838032515e-07,
+      "loss": 0.3962,
+      "step": 1139
+    },
+    {
+      "epoch": 4.6177215189873415,
+      "grad_norm": 0.19646211798255817,
+      "learning_rate": 6.488245230261281e-07,
+      "loss": 0.3984,
+      "step": 1140
+    },
+    {
+      "epoch": 4.6217721518987345,
+      "grad_norm": 0.19138505281834128,
+      "learning_rate": 6.345626139559868e-07,
+      "loss": 0.4033,
+      "step": 1141
+    },
+    {
+      "epoch": 4.6258227848101265,
+      "grad_norm": 0.19631757759789842,
+      "learning_rate": 6.204566714562866e-07,
+      "loss": 0.3938,
+      "step": 1142
+    },
+    {
+      "epoch": 4.629873417721519,
+      "grad_norm": 0.20031566545877383,
+      "learning_rate": 6.06506809134344e-07,
+      "loss": 0.3799,
+      "step": 1143
+    },
+    {
+      "epoch": 4.633924050632912,
+      "grad_norm": 0.19320587541776837,
+      "learning_rate": 5.927131393404373e-07,
+      "loss": 0.4021,
+      "step": 1144
+    },
+    {
+      "epoch": 4.637974683544304,
+      "grad_norm": 0.19378799681004516,
+      "learning_rate": 5.790757731668817e-07,
+      "loss": 0.3906,
+      "step": 1145
+    },
+    {
+      "epoch": 4.642025316455696,
+      "grad_norm": 0.19031122489277594,
+      "learning_rate": 5.655948204471507e-07,
+      "loss": 0.3807,
+      "step": 1146
+    },
+    {
+      "epoch": 4.646075949367089,
+      "grad_norm": 0.19641100367428052,
+      "learning_rate": 5.522703897549875e-07,
+      "loss": 0.3957,
+      "step": 1147
+    },
+    {
+      "epoch": 4.650126582278481,
+      "grad_norm": 0.18907270444088964,
+      "learning_rate": 5.391025884035239e-07,
+      "loss": 0.3982,
+      "step": 1148
+    },
+    {
+      "epoch": 4.654177215189874,
+      "grad_norm": 0.19713075385120005,
+      "learning_rate": 5.260915224444207e-07,
+      "loss": 0.3812,
+      "step": 1149
+    },
+    {
+      "epoch": 4.658227848101266,
+      "grad_norm": 0.19568076037355173,
+      "learning_rate": 5.132372966670129e-07,
+      "loss": 0.3946,
+      "step": 1150
+    },
+    {
+      "epoch": 4.662278481012658,
+      "grad_norm": 0.19509476286485025,
+      "learning_rate": 5.005400145974704e-07,
+      "loss": 0.3934,
+      "step": 1151
+    },
+    {
+      "epoch": 4.666329113924051,
+      "grad_norm": 0.1956910133077776,
+      "learning_rate": 4.879997784979562e-07,
+      "loss": 0.377,
+      "step": 1152
+    },
+    {
+      "epoch": 4.670379746835443,
+      "grad_norm": 0.19586508213959866,
+      "learning_rate": 4.7561668936580984e-07,
+      "loss": 0.3855,
+      "step": 1153
+    },
+    {
+      "epoch": 4.674430379746836,
+      "grad_norm": 0.20163796534685277,
+      "learning_rate": 4.6339084693272306e-07,
+      "loss": 0.3908,
+      "step": 1154
+    },
+    {
+      "epoch": 4.678481012658228,
+      "grad_norm": 0.1954514948541024,
+      "learning_rate": 4.5132234966395847e-07,
+      "loss": 0.3967,
+      "step": 1155
+    },
+    {
+      "epoch": 4.68253164556962,
+      "grad_norm": 0.19117203052620105,
+      "learning_rate": 4.3941129475752795e-07,
+      "loss": 0.3979,
+      "step": 1156
+    },
+    {
+      "epoch": 4.686582278481013,
+      "grad_norm": 0.1909613925229115,
+      "learning_rate": 4.27657778143431e-07,
+      "loss": 0.394,
+      "step": 1157
+    },
+    {
+      "epoch": 4.690632911392405,
+      "grad_norm": 0.19279847219419927,
+      "learning_rate": 4.1606189448287757e-07,
+      "loss": 0.3837,
+      "step": 1158
+    },
+    {
+      "epoch": 4.694683544303797,
+      "grad_norm": 0.19098468708835686,
+      "learning_rate": 4.046237371675177e-07,
+      "loss": 0.3862,
+      "step": 1159
+    },
+    {
+      "epoch": 4.69873417721519,
+      "grad_norm": 0.19188460065705723,
+      "learning_rate": 3.9334339831869963e-07,
+      "loss": 0.4095,
+      "step": 1160
+    },
+    {
+      "epoch": 4.702784810126582,
+      "grad_norm": 0.19750897873699108,
+      "learning_rate": 3.8222096878671955e-07,
+      "loss": 0.3759,
+      "step": 1161
+    },
+    {
+      "epoch": 4.706835443037974,
+      "grad_norm": 0.19770702762229986,
+      "learning_rate": 3.7125653815009545e-07,
+      "loss": 0.3991,
+      "step": 1162
+    },
+    {
+      "epoch": 4.710886075949367,
+      "grad_norm": 0.19787814951171614,
+      "learning_rate": 3.6045019471484974e-07,
+      "loss": 0.3805,
+      "step": 1163
+    },
+    {
+      "epoch": 4.714936708860759,
+      "grad_norm": 0.19201846278164317,
+      "learning_rate": 3.498020255137813e-07,
+      "loss": 0.3876,
+      "step": 1164
+    },
+    {
+      "epoch": 4.718987341772152,
+      "grad_norm": 0.1900788379691705,
+      "learning_rate": 3.393121163057811e-07,
+      "loss": 0.3841,
+      "step": 1165
+    },
+    {
+      "epoch": 4.723037974683544,
+      "grad_norm": 0.19500325102802232,
+      "learning_rate": 3.289805515751399e-07,
+      "loss": 0.4016,
+      "step": 1166
+    },
+    {
+      "epoch": 4.727088607594936,
+      "grad_norm": 0.19288819801252693,
+      "learning_rate": 3.188074145308573e-07,
+      "loss": 0.3761,
+      "step": 1167
+    },
+    {
+      "epoch": 4.731139240506329,
+      "grad_norm": 0.19347389957958605,
+      "learning_rate": 3.087927871059804e-07,
+      "loss": 0.3937,
+      "step": 1168
+    },
+    {
+      "epoch": 4.735189873417721,
+      "grad_norm": 0.1905510586664015,
+      "learning_rate": 2.989367499569418e-07,
+      "loss": 0.3807,
+      "step": 1169
+    },
+    {
+      "epoch": 4.739240506329114,
+      "grad_norm": 0.7079874371824011,
+      "learning_rate": 2.8923938246290917e-07,
+      "loss": 0.398,
+      "step": 1170
+    },
+    {
+      "epoch": 4.743291139240506,
+      "grad_norm": 0.1914432085100947,
+      "learning_rate": 2.7970076272514804e-07,
+      "loss": 0.3938,
+      "step": 1171
+    },
+    {
+      "epoch": 4.7473417721518985,
+      "grad_norm": 0.1912945391885138,
+      "learning_rate": 2.703209675663887e-07,
+      "loss": 0.3938,
+      "step": 1172
+    },
+    {
+      "epoch": 4.751392405063291,
+      "grad_norm": 0.19051223869998854,
+      "learning_rate": 2.6110007253021374e-07,
+      "loss": 0.3865,
+      "step": 1173
+    },
+    {
+      "epoch": 4.7554430379746835,
+      "grad_norm": 0.1911391765051781,
+      "learning_rate": 2.520381518804471e-07,
+      "loss": 0.4017,
+      "step": 1174
+    },
+    {
+      "epoch": 4.759493670886076,
+      "grad_norm": 0.19462373454018025,
+      "learning_rate": 2.4313527860054585e-07,
+      "loss": 0.382,
+      "step": 1175
+    },
+    {
+      "epoch": 4.7635443037974685,
+      "grad_norm": 0.19162372714318643,
+      "learning_rate": 2.343915243930317e-07,
+      "loss": 0.3958,
+      "step": 1176
+    },
+    {
+      "epoch": 4.767594936708861,
+      "grad_norm": 0.19227342947908058,
+      "learning_rate": 2.2580695967889367e-07,
+      "loss": 0.3825,
+      "step": 1177
+    },
+    {
+      "epoch": 4.7716455696202535,
+      "grad_norm": 0.19703699602549501,
+      "learning_rate": 2.1738165359704189e-07,
+      "loss": 0.3703,
+      "step": 1178
+    },
+    {
+      "epoch": 4.775696202531646,
+      "grad_norm": 0.19060865118910209,
+      "learning_rate": 2.0911567400373257e-07,
+      "loss": 0.3985,
+      "step": 1179
+    },
+    {
+      "epoch": 4.779746835443038,
+      "grad_norm": 0.18789230109602112,
+      "learning_rate": 2.0100908747202607e-07,
+      "loss": 0.4028,
+      "step": 1180
+    },
+    {
+      "epoch": 4.783797468354431,
+      "grad_norm": 0.19143187842580545,
+      "learning_rate": 1.9306195929125638e-07,
+      "loss": 0.399,
+      "step": 1181
+    },
+    {
+      "epoch": 4.787848101265823,
+      "grad_norm": 0.19962257727256366,
+      "learning_rate": 1.8527435346650247e-07,
+      "loss": 0.3962,
+      "step": 1182
+    },
+    {
+      "epoch": 4.791898734177215,
+      "grad_norm": 0.20780652231425642,
+      "learning_rate": 1.7764633271807108e-07,
+      "loss": 0.3729,
+      "step": 1183
+    },
+    {
+      "epoch": 4.795949367088608,
+      "grad_norm": 0.18622496069404443,
+      "learning_rate": 1.7017795848099262e-07,
+      "loss": 0.4045,
+      "step": 1184
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.18972948623145364,
+      "learning_rate": 1.6286929090452596e-07,
+      "loss": 0.3925,
+      "step": 1185
+    },
+    {
+      "epoch": 4.804050632911393,
+      "grad_norm": 0.20220792377089913,
+      "learning_rate": 1.557203888516745e-07,
+      "loss": 0.3793,
+      "step": 1186
+    },
+    {
+      "epoch": 4.808101265822785,
+      "grad_norm": 0.19294112533165295,
+      "learning_rate": 1.487313098987131e-07,
+      "loss": 0.39,
+      "step": 1187
+    },
+    {
+      "epoch": 4.812151898734177,
+      "grad_norm": 0.19056382310004671,
+      "learning_rate": 1.4190211033472402e-07,
+      "loss": 0.3977,
+      "step": 1188
+    },
+    {
+      "epoch": 4.81620253164557,
+      "grad_norm": 0.20027617356260152,
+      "learning_rate": 1.3523284516113955e-07,
+      "loss": 0.3792,
+      "step": 1189
+    },
+    {
+      "epoch": 4.820253164556962,
+      "grad_norm": 0.19078984417122416,
+      "learning_rate": 1.2872356809130682e-07,
+      "loss": 0.3969,
+      "step": 1190
+    },
+    {
+      "epoch": 4.824303797468355,
+      "grad_norm": 0.1922676920188985,
+      "learning_rate": 1.2237433155004807e-07,
+      "loss": 0.3805,
+      "step": 1191
+    },
+    {
+      "epoch": 4.828354430379747,
+      "grad_norm": 0.19325675961158972,
+      "learning_rate": 1.1618518667323886e-07,
+      "loss": 0.3959,
+      "step": 1192
+    },
+    {
+      "epoch": 4.832405063291139,
+      "grad_norm": 0.19513633188312607,
+      "learning_rate": 1.1015618330740385e-07,
+      "loss": 0.3937,
+      "step": 1193
+    },
+    {
+      "epoch": 4.836455696202532,
+      "grad_norm": 0.1939643805798085,
+      "learning_rate": 1.042873700093061e-07,
+      "loss": 0.3941,
+      "step": 1194
+    },
+    {
+      "epoch": 4.840506329113924,
+      "grad_norm": 0.19804257781814888,
+      "learning_rate": 9.857879404556291e-08,
+      "loss": 0.3897,
+      "step": 1195
+    },
+    {
+      "epoch": 4.844556962025316,
+      "grad_norm": 0.19463968740800933,
+      "learning_rate": 9.303050139225722e-08,
+      "loss": 0.3912,
+      "step": 1196
+    },
+    {
+      "epoch": 4.848607594936709,
+      "grad_norm": 0.21030721443041006,
+      "learning_rate": 8.76425367345779e-08,
+      "loss": 0.4062,
+      "step": 1197
+    },
+    {
+      "epoch": 4.852658227848101,
+      "grad_norm": 0.19392857691039522,
+      "learning_rate": 8.241494346644897e-08,
+      "loss": 0.3818,
+      "step": 1198
+    },
+    {
+      "epoch": 4.856708860759493,
+      "grad_norm": 0.20193909146249722,
+      "learning_rate": 7.734776369019204e-08,
+      "loss": 0.3801,
+      "step": 1199
+    },
+    {
+      "epoch": 4.860759493670886,
+      "grad_norm": 0.1887162131680345,
+      "learning_rate": 7.244103821617332e-08,
+      "loss": 0.3788,
+      "step": 1200
+    },
+    {
+      "epoch": 4.864810126582278,
+      "grad_norm": 0.18839070984851097,
+      "learning_rate": 6.769480656248606e-08,
+      "loss": 0.4093,
+      "step": 1201
+    },
+    {
+      "epoch": 4.868860759493671,
+      "grad_norm": 0.19559914780126442,
+      "learning_rate": 6.310910695462635e-08,
+      "loss": 0.3862,
+      "step": 1202
+    },
+    {
+      "epoch": 4.872911392405063,
+      "grad_norm": 0.1948004965675549,
+      "learning_rate": 5.8683976325191185e-08,
+      "loss": 0.404,
+      "step": 1203
+    },
+    {
+      "epoch": 4.876962025316455,
+      "grad_norm": 0.2057835053139491,
+      "learning_rate": 5.4419450313571984e-08,
+      "loss": 0.3895,
+      "step": 1204
+    },
+    {
+      "epoch": 4.881012658227848,
+      "grad_norm": 0.188931555498433,
+      "learning_rate": 5.031556326567488e-08,
+      "loss": 0.4082,
+      "step": 1205
+    },
+    {
+      "epoch": 4.88506329113924,
+      "grad_norm": 0.19698214741188194,
+      "learning_rate": 4.637234823364312e-08,
+      "loss": 0.3943,
+      "step": 1206
+    },
+    {
+      "epoch": 4.889113924050633,
+      "grad_norm": 0.1949750296518769,
+      "learning_rate": 4.258983697558838e-08,
+      "loss": 0.3806,
+      "step": 1207
+    },
+    {
+      "epoch": 4.8931645569620255,
+      "grad_norm": 0.19646767602691984,
+      "learning_rate": 3.896805995533548e-08,
+      "loss": 0.3814,
+      "step": 1208
+    },
+    {
+      "epoch": 4.8972151898734175,
+      "grad_norm": 0.18718099424715917,
+      "learning_rate": 3.550704634218028e-08,
+      "loss": 0.3937,
+      "step": 1209
+    },
+    {
+      "epoch": 4.9012658227848105,
+      "grad_norm": 0.18399171354535948,
+      "learning_rate": 3.2206824010647676e-08,
+      "loss": 0.401,
+      "step": 1210
+    },
+    {
+      "epoch": 4.905316455696203,
+      "grad_norm": 0.19771206220496435,
+      "learning_rate": 2.9067419540278476e-08,
+      "loss": 0.3885,
+      "step": 1211
+    },
+    {
+      "epoch": 4.909367088607595,
+      "grad_norm": 0.19184266987655796,
+      "learning_rate": 2.6088858215400638e-08,
+      "loss": 0.3888,
+      "step": 1212
+    },
+    {
+      "epoch": 4.913417721518988,
+      "grad_norm": 0.1965671611955227,
+      "learning_rate": 2.3271164024940564e-08,
+      "loss": 0.3908,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91746835443038,
+      "grad_norm": 0.1863954908357033,
+      "learning_rate": 2.061435966221881e-08,
+      "loss": 0.3978,
+      "step": 1214
+    },
+    {
+      "epoch": 4.921518987341772,
+      "grad_norm": 0.19503787791990446,
+      "learning_rate": 1.811846652477245e-08,
+      "loss": 0.3937,
+      "step": 1215
+    },
+    {
+      "epoch": 4.925569620253165,
+      "grad_norm": 0.2004221146101876,
+      "learning_rate": 1.5783504714184106e-08,
+      "loss": 0.394,
+      "step": 1216
+    },
+    {
+      "epoch": 4.929620253164557,
+      "grad_norm": 0.19398840735398226,
+      "learning_rate": 1.360949303591097e-08,
+      "loss": 0.3938,
+      "step": 1217
+    },
+    {
+      "epoch": 4.93367088607595,
+      "grad_norm": 0.18559761620764764,
+      "learning_rate": 1.1596448999144916e-08,
+      "loss": 0.409,
+      "step": 1218
+    },
+    {
+      "epoch": 4.937721518987342,
+      "grad_norm": 0.19099188415000423,
+      "learning_rate": 9.744388816668172e-09,
+      "loss": 0.3892,
+      "step": 1219
+    },
+    {
+      "epoch": 4.941772151898734,
+      "grad_norm": 0.9329567766965216,
+      "learning_rate": 8.05332740472009e-09,
+      "loss": 0.3884,
+      "step": 1220
+    },
+    {
+      "epoch": 4.945822784810127,
+      "grad_norm": 0.23225051267991426,
+      "learning_rate": 6.523278382872811e-09,
+      "loss": 0.3894,
+      "step": 1221
+    },
+    {
+      "epoch": 4.949873417721519,
+      "grad_norm": 0.1912805901187407,
+      "learning_rate": 5.15425407393133e-09,
+      "loss": 0.3878,
+      "step": 1222
+    },
+    {
+      "epoch": 4.953924050632912,
+      "grad_norm": 0.19260081143437563,
+      "learning_rate": 3.94626550383137e-09,
+      "loss": 0.4003,
+      "step": 1223
+    },
+    {
+      "epoch": 4.957974683544304,
+      "grad_norm": 0.18503508912949895,
+      "learning_rate": 2.899322401546112e-09,
+      "loss": 0.4014,
+      "step": 1224
+    },
+    {
+      "epoch": 4.962025316455696,
+      "grad_norm": 0.1886210237976882,
+      "learning_rate": 2.013433199010706e-09,
+      "loss": 0.3949,
+      "step": 1225
+    },
+    {
+      "epoch": 4.966075949367089,
+      "grad_norm": 0.2103717982454105,
+      "learning_rate": 1.2886050310556563e-09,
+      "loss": 0.3869,
+      "step": 1226
+    },
+    {
+      "epoch": 4.970126582278481,
+      "grad_norm": 0.2006587357943449,
+      "learning_rate": 7.248437353468695e-10,
+      "loss": 0.3904,
+      "step": 1227
+    },
+    {
+      "epoch": 4.974177215189873,
+      "grad_norm": 0.18924215253608026,
+      "learning_rate": 3.221538523412449e-10,
+      "loss": 0.396,
+      "step": 1228
+    },
+    {
+      "epoch": 4.978227848101266,
+      "grad_norm": 0.18824782737184623,
+      "learning_rate": 8.053862524670663e-11,
+      "loss": 0.4009,
+      "step": 1229
+    },
+    {
+      "epoch": 4.982278481012658,
+      "grad_norm": 0.19515885607361277,
+      "learning_rate": 0.0,
+      "loss": 0.3905,
+      "step": 1230
+    },
+    {
+      "epoch": 4.982278481012658,
+      "step": 1230,
+      "total_flos": 6.045069884550152e+18,
+      "train_loss": 0.5157663706356918,
+      "train_runtime": 55867.7041,
+      "train_samples_per_second": 2.828,
+      "train_steps_per_second": 0.022
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1230,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.045069884550152e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}