diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
deleted file mode 100644--- "a/last-checkpoint/trainer_state.json"
+++ /dev/null
@@ -1,11812 +0,0 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 14.0,
-  "global_step": 972622,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.01,
-      "learning_rate": 5.140331037318804e-06,
-      "loss": 9.66,
-      "step": 500
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.0280662074637608e-05,
-      "loss": 7.9578,
-      "step": 1000
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 1.542099311195641e-05,
-      "loss": 6.7356,
-      "step": 1500
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 2.0561324149275216e-05,
-      "loss": 6.3535,
-      "step": 2000
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 2.570165518659402e-05,
-      "loss": 6.0716,
-      "step": 2500
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 3.084198622391282e-05,
-      "loss": 5.7806,
-      "step": 3000
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 3.5982317261231625e-05,
-      "loss": 5.3296,
-      "step": 3500
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 4.112264829855043e-05,
-      "loss": 4.8984,
-      "step": 4000
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 4.6262979335869235e-05,
-      "loss": 4.5987,
-      "step": 4500
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 5.140331037318804e-05,
-      "loss": 4.3705,
-      "step": 5000
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 5.6543641410506844e-05,
-      "loss": 4.1875,
-      "step": 5500
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 6.168397244782565e-05,
-      "loss": 4.0215,
-      "step": 6000
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 6.682430348514444e-05,
-      "loss": 3.8753,
-      "step": 6500
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 7.196463452246325e-05,
-      "loss": 3.7559,
-      "step": 7000
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 7.710496555978204e-05,
-      "loss": 3.6464,
-      "step": 7500
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 8.224529659710087e-05,
-      "loss": 3.5461,
-      "step": 8000
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 8.738562763441966e-05,
-      "loss": 3.4592,
-      "step": 8500
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.252595867173847e-05,
-      "loss": 3.3816,
-      "step": 9000
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.766628970905726e-05,
-      "loss": 3.3121,
-      "step": 9500
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.997164799900302e-05,
-      "loss": 3.2467,
-      "step": 10000
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.991972125725027e-05,
-      "loss": 3.1821,
-      "step": 10500
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.986779451549754e-05,
-      "loss": 3.1286,
-      "step": 11000
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.981586777374481e-05,
-      "loss": 3.0801,
-      "step": 11500
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.976394103199208e-05,
-      "loss": 3.0316,
-      "step": 12000
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.971211814372284e-05,
-      "loss": 2.9902,
-      "step": 12500
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.966019140197011e-05,
-      "loss": 2.9502,
-      "step": 13000
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.960826466021737e-05,
-      "loss": 2.9131,
-      "step": 13500
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.955633791846463e-05,
-      "loss": 2.8816,
-      "step": 14000
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.95045150301954e-05,
-      "loss": 2.8481,
-      "step": 14500
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.945258828844267e-05,
-      "loss": 2.8154,
-      "step": 15000
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.940066154668994e-05,
-      "loss": 2.7873,
-      "step": 15500
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.93487348049372e-05,
-      "loss": 2.7658,
-      "step": 16000
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.929691191666797e-05,
-      "loss": 2.7377,
-      "step": 16500
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.924498517491524e-05,
-      "loss": 2.7145,
-      "step": 17000
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.91930584331625e-05,
-      "loss": 2.6943,
-      "step": 17500
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.914113169140976e-05,
-      "loss": 2.6711,
-      "step": 18000
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.908930880314053e-05,
-      "loss": 2.6557,
-      "step": 18500
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.90373820613878e-05,
-      "loss": 2.6369,
-      "step": 19000
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 9.898545531963507e-05,
-      "loss": 2.6161,
-      "step": 19500
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 9.893352857788233e-05,
-      "loss": 2.6013,
-      "step": 20000
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 9.88817056896131e-05,
-      "loss": 2.5856,
-      "step": 20500
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 9.882977894786037e-05,
-      "loss": 2.5747,
-      "step": 21000
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 9.877785220610762e-05,
-      "loss": 2.5553,
-      "step": 21500
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 9.872592546435489e-05,
-      "loss": 2.5415,
-      "step": 22000
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 9.867410257608566e-05,
-      "loss": 2.5352,
-      "step": 22500
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 9.862217583433293e-05,
-      "loss": 2.5192,
-      "step": 23000
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 9.85702490925802e-05,
-      "loss": 2.509,
-      "step": 23500
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 9.851832235082745e-05,
-      "loss": 2.4963,
-      "step": 24000
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 9.846649946255823e-05,
-      "loss": 2.4866,
-      "step": 24500
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 9.841457272080548e-05,
-      "loss": 2.4738,
-      "step": 25000
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 9.836264597905277e-05,
-      "loss": 2.4613,
-      "step": 25500
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 9.831071923730002e-05,
-      "loss": 2.4566,
-      "step": 26000
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 9.82588963490308e-05,
-      "loss": 2.4492,
-      "step": 26500
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 9.820696960727805e-05,
-      "loss": 2.4387,
-      "step": 27000
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 9.815504286552532e-05,
-      "loss": 2.431,
-      "step": 27500
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 9.810311612377259e-05,
-      "loss": 2.4203,
-      "step": 28000
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 9.805129323550336e-05,
-      "loss": 2.4134,
-      "step": 28500
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.799936649375063e-05,
-      "loss": 2.4058,
-      "step": 29000
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 9.794743975199788e-05,
-      "loss": 2.3948,
-      "step": 29500
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 9.789551301024515e-05,
-      "loss": 2.3866,
-      "step": 30000
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 9.784358626849242e-05,
-      "loss": 2.3806,
-      "step": 30500
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.779165952673967e-05,
-      "loss": 2.3738,
-      "step": 31000
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 9.773973278498695e-05,
-      "loss": 2.3675,
-      "step": 31500
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 9.76879098967177e-05,
-      "loss": 2.3633,
-      "step": 32000
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 9.763598315496499e-05,
-      "loss": 2.356,
-      "step": 32500
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.758405641321224e-05,
-      "loss": 2.3454,
-      "step": 33000
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 9.753212967145951e-05,
-      "loss": 2.3464,
-      "step": 33500
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 9.748030678319028e-05,
-      "loss": 2.3384,
-      "step": 34000
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 9.742838004143755e-05,
-      "loss": 2.3314,
-      "step": 34500
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 9.737645329968481e-05,
-      "loss": 2.3223,
-      "step": 35000
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 9.732452655793207e-05,
-      "loss": 2.3201,
-      "step": 35500
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 9.727270366966285e-05,
-      "loss": 2.3198,
-      "step": 36000
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.72207769279101e-05,
-      "loss": 2.3051,
-      "step": 36500
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 9.716885018615738e-05,
-      "loss": 2.3041,
-      "step": 37000
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 9.711692344440464e-05,
-      "loss": 2.3034,
-      "step": 37500
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.70649967026519e-05,
-      "loss": 2.2964,
-      "step": 38000
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 9.701317381438267e-05,
-      "loss": 2.2894,
-      "step": 38500
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 9.696124707262993e-05,
-      "loss": 2.2769,
-      "step": 39000
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 9.690932033087721e-05,
-      "loss": 2.2809,
-      "step": 39500
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.685739358912447e-05,
-      "loss": 2.2743,
-      "step": 40000
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 9.680557070085525e-05,
-      "loss": 2.2699,
-      "step": 40500
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 9.67536439591025e-05,
-      "loss": 2.2668,
-      "step": 41000
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.670171721734977e-05,
-      "loss": 2.2635,
-      "step": 41500
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 9.664979047559704e-05,
-      "loss": 2.2637,
-      "step": 42000
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 9.65979675873278e-05,
-      "loss": 2.2584,
-      "step": 42500
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 9.654604084557507e-05,
-      "loss": 2.2535,
-      "step": 43000
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 9.649411410382233e-05,
-      "loss": 2.2462,
-      "step": 43500
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 9.64421873620696e-05,
-      "loss": 2.2427,
-      "step": 44000
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 9.639036447380036e-05,
-      "loss": 2.242,
-      "step": 44500
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 9.633843773204764e-05,
-      "loss": 2.2341,
-      "step": 45000
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 9.62865109902949e-05,
-      "loss": 2.2321,
-      "step": 45500
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 9.623458424854217e-05,
-      "loss": 2.2311,
-      "step": 46000
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 9.618265750678943e-05,
-      "loss": 2.2242,
-      "step": 46500
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 9.61308346185202e-05,
-      "loss": 2.2206,
-      "step": 47000
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 9.607890787676747e-05,
-      "loss": 2.2164,
-      "step": 47500
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 9.602698113501472e-05,
-      "loss": 2.217,
-      "step": 48000
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 9.597505439326199e-05,
-      "loss": 2.2144,
-      "step": 48500
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 9.592323150499276e-05,
-      "loss": 2.211,
-      "step": 49000
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 9.587130476324003e-05,
-      "loss": 2.2056,
-      "step": 49500
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 9.58193780214873e-05,
-      "loss": 2.2013,
-      "step": 50000
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 9.576745127973455e-05,
-      "loss": 2.2074,
-      "step": 50500
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 9.571562839146533e-05,
-      "loss": 2.1932,
-      "step": 51000
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 9.566370164971258e-05,
-      "loss": 2.1911,
-      "step": 51500
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 9.561177490795985e-05,
-      "loss": 2.1942,
-      "step": 52000
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 9.555984816620712e-05,
-      "loss": 2.1868,
-      "step": 52500
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 9.550792142445439e-05,
-      "loss": 2.1795,
-      "step": 53000
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 9.545609853618515e-05,
-      "loss": 2.187,
-      "step": 53500
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.540417179443242e-05,
-      "loss": 2.1752,
-      "step": 54000
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 9.535224505267969e-05,
-      "loss": 2.1789,
-      "step": 54500
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 9.530031831092695e-05,
-      "loss": 2.1755,
-      "step": 55000
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 9.524849542265773e-05,
-      "loss": 2.1723,
-      "step": 55500
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.519656868090498e-05,
-      "loss": 2.1722,
-      "step": 56000
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 9.514464193915225e-05,
-      "loss": 2.166,
-      "step": 56500
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 9.509271519739952e-05,
-      "loss": 2.1679,
-      "step": 57000
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 9.504089230913028e-05,
-      "loss": 2.1619,
-      "step": 57500
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 9.498896556737755e-05,
-      "loss": 2.1595,
-      "step": 58000
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 9.493703882562482e-05,
-      "loss": 2.1557,
-      "step": 58500
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 9.488511208387207e-05,
-      "loss": 2.1567,
-      "step": 59000
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.483328919560285e-05,
-      "loss": 2.1504,
-      "step": 59500
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.478136245385011e-05,
-      "loss": 2.1459,
-      "step": 60000
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 9.472943571209738e-05,
-      "loss": 2.1499,
-      "step": 60500
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 9.467750897034465e-05,
-      "loss": 2.146,
-      "step": 61000
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 9.462568608207541e-05,
-      "loss": 2.1437,
-      "step": 61500
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 9.457375934032268e-05,
-      "loss": 2.1437,
-      "step": 62000
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 9.452183259856993e-05,
-      "loss": 2.1415,
-      "step": 62500
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 9.44699058568172e-05,
-      "loss": 2.1406,
-      "step": 63000
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 9.441808296854797e-05,
-      "loss": 2.1334,
-      "step": 63500
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 9.436615622679524e-05,
-      "loss": 2.1327,
-      "step": 64000
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 9.43142294850425e-05,
-      "loss": 2.1297,
-      "step": 64500
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 9.426230274328977e-05,
-      "loss": 2.1275,
-      "step": 65000
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 9.421037600153704e-05,
-      "loss": 2.1317,
-      "step": 65500
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 9.415855311326781e-05,
-      "loss": 2.1305,
-      "step": 66000
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.410662637151508e-05,
-      "loss": 2.1229,
-      "step": 66500
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 9.405469962976233e-05,
-      "loss": 2.1212,
-      "step": 67000
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 9.40027728880096e-05,
-      "loss": 2.1172,
-      "step": 67500
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 9.395094999974037e-05,
-      "loss": 2.1139,
-      "step": 68000
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 9.389902325798763e-05,
-      "loss": 2.1156,
-      "step": 68500
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 9.38470965162349e-05,
-      "loss": 2.111,
-      "step": 69000
-    },
-    {
-      "epoch": 1.0,
-      "eval_accuracy": 0.6055819008431793,
-      "eval_loss": 1.98311185836792,
-      "eval_runtime": 1276.0691,
-      "eval_samples_per_second": 422.342,
-      "eval_steps_per_second": 26.397,
-      "step": 69473
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 9.379516977448216e-05,
-      "loss": 2.1118,
-      "step": 69500
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 9.374334688621294e-05,
-      "loss": 2.1085,
-      "step": 70000
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 9.369142014446019e-05,
-      "loss": 2.1029,
-      "step": 70500
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 9.363949340270747e-05,
-      "loss": 2.1039,
-      "step": 71000
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 9.358756666095473e-05,
-      "loss": 2.1002,
-      "step": 71500
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 9.353574377268551e-05,
-      "loss": 2.1002,
-      "step": 72000
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 9.348381703093276e-05,
-      "loss": 2.0995,
-      "step": 72500
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 9.343189028918002e-05,
-      "loss": 2.1002,
-      "step": 73000
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 9.33799635474273e-05,
-      "loss": 2.0938,
-      "step": 73500
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 9.332814065915807e-05,
-      "loss": 2.0927,
-      "step": 74000
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 9.327621391740533e-05,
-      "loss": 2.0887,
-      "step": 74500
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 9.322428717565259e-05,
-      "loss": 2.0899,
-      "step": 75000
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 9.317236043389986e-05,
-      "loss": 2.0898,
-      "step": 75500
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 9.312053754563062e-05,
-      "loss": 2.0864,
-      "step": 76000
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 9.306861080387789e-05,
-      "loss": 2.0843,
-      "step": 76500
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 9.301668406212516e-05,
-      "loss": 2.0826,
-      "step": 77000
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 9.296475732037241e-05,
-      "loss": 2.0859,
-      "step": 77500
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 9.29129344321032e-05,
-      "loss": 2.0809,
-      "step": 78000
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 9.286100769035045e-05,
-      "loss": 2.0811,
-      "step": 78500
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 9.280908094859773e-05,
-      "loss": 2.079,
-      "step": 79000
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 9.275715420684499e-05,
-      "loss": 2.0733,
-      "step": 79500
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 9.270533131857577e-05,
-      "loss": 2.0784,
-      "step": 80000
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 9.265340457682302e-05,
-      "loss": 2.0768,
-      "step": 80500
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 9.260147783507029e-05,
-      "loss": 2.0749,
-      "step": 81000
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 9.254955109331756e-05,
-      "loss": 2.0716,
-      "step": 81500
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 9.249772820504832e-05,
-      "loss": 2.069,
-      "step": 82000
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 9.244580146329559e-05,
-      "loss": 2.0639,
-      "step": 82500
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 9.239387472154285e-05,
-      "loss": 2.069,
-      "step": 83000
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 9.234194797979011e-05,
-      "loss": 2.0648,
-      "step": 83500
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 9.229012509152088e-05,
-      "loss": 2.0674,
-      "step": 84000
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 9.223819834976816e-05,
-      "loss": 2.0609,
-      "step": 84500
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 9.218627160801542e-05,
-      "loss": 2.0592,
-      "step": 85000
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 9.213434486626267e-05,
-      "loss": 2.058,
-      "step": 85500
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 9.208252197799345e-05,
-      "loss": 2.0603,
-      "step": 86000
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 9.20305952362407e-05,
-      "loss": 2.0562,
-      "step": 86500
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 9.197866849448799e-05,
-      "loss": 2.0566,
-      "step": 87000
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 9.192674175273524e-05,
-      "loss": 2.0546,
-      "step": 87500
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 9.187491886446602e-05,
-      "loss": 2.0552,
-      "step": 88000
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 9.182299212271328e-05,
-      "loss": 2.0493,
-      "step": 88500
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 9.177106538096055e-05,
-      "loss": 2.0545,
-      "step": 89000
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 9.171913863920781e-05,
-      "loss": 2.0494,
-      "step": 89500
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 9.166721189745507e-05,
-      "loss": 2.0496,
-      "step": 90000
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 9.161538900918585e-05,
-      "loss": 2.0467,
-      "step": 90500
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 9.15634622674331e-05,
-      "loss": 2.047,
-      "step": 91000
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 9.151153552568037e-05,
-      "loss": 2.048,
-      "step": 91500
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 9.145960878392764e-05,
-      "loss": 2.0433,
-      "step": 92000
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 9.14077858956584e-05,
-      "loss": 2.0442,
-      "step": 92500
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 9.135585915390567e-05,
-      "loss": 2.0474,
-      "step": 93000
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 9.130393241215294e-05,
-      "loss": 2.0364,
-      "step": 93500
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 9.125200567040021e-05,
-      "loss": 2.0393,
-      "step": 94000
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 9.120018278213098e-05,
-      "loss": 2.0371,
-      "step": 94500
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 9.114825604037825e-05,
-      "loss": 2.0354,
-      "step": 95000
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 9.10963292986255e-05,
-      "loss": 2.0355,
-      "step": 95500
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 9.104440255687277e-05,
-      "loss": 2.0366,
-      "step": 96000
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 9.099257966860354e-05,
-      "loss": 2.0339,
-      "step": 96500
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 9.09406529268508e-05,
-      "loss": 2.0352,
-      "step": 97000
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 9.088872618509807e-05,
-      "loss": 2.0305,
-      "step": 97500
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 9.083679944334533e-05,
-      "loss": 2.0276,
-      "step": 98000
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 9.07849765550761e-05,
-      "loss": 2.0291,
-      "step": 98500
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 9.073304981332336e-05,
-      "loss": 2.0318,
-      "step": 99000
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 9.068112307157063e-05,
-      "loss": 2.022,
-      "step": 99500
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 9.06291963298179e-05,
-      "loss": 2.0272,
-      "step": 100000
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 9.057737344154866e-05,
-      "loss": 2.0284,
-      "step": 100500
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 9.052544669979593e-05,
-      "loss": 2.0258,
-      "step": 101000
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 9.04735199580432e-05,
-      "loss": 2.0218,
-      "step": 101500
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 9.042159321629046e-05,
-      "loss": 2.0221,
-      "step": 102000
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 9.036977032802124e-05,
-      "loss": 2.0181,
-      "step": 102500
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 9.031784358626849e-05,
-      "loss": 2.0207,
-      "step": 103000
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 9.026591684451576e-05,
-      "loss": 2.022,
-      "step": 103500
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 9.021399010276303e-05,
-      "loss": 2.0179,
-      "step": 104000
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 9.016216721449379e-05,
-      "loss": 2.0189,
-      "step": 104500
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 9.011024047274106e-05,
-      "loss": 2.0171,
-      "step": 105000
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 9.005831373098833e-05,
-      "loss": 2.0176,
-      "step": 105500
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 9.00063869892356e-05,
-      "loss": 2.0137,
-      "step": 106000
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 8.995456410096636e-05,
-      "loss": 2.0115,
-      "step": 106500
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 8.990263735921363e-05,
-      "loss": 2.0139,
-      "step": 107000
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 8.985071061746089e-05,
-      "loss": 2.011,
-      "step": 107500
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 8.979878387570816e-05,
-      "loss": 2.0085,
-      "step": 108000
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 8.974696098743892e-05,
-      "loss": 2.0117,
-      "step": 108500
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 8.969503424568619e-05,
-      "loss": 2.0095,
-      "step": 109000
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 8.964310750393346e-05,
-      "loss": 2.0053,
-      "step": 109500
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 8.959118076218071e-05,
-      "loss": 2.006,
-      "step": 110000
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 8.953935787391149e-05,
-      "loss": 2.0108,
-      "step": 110500
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 8.948743113215875e-05,
-      "loss": 2.0034,
-      "step": 111000
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 8.943550439040602e-05,
-      "loss": 2.0072,
-      "step": 111500
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 8.938357764865328e-05,
-      "loss": 2.005,
-      "step": 112000
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 8.933175476038405e-05,
-      "loss": 2.0017,
-      "step": 112500
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 8.927982801863132e-05,
-      "loss": 1.9997,
-      "step": 113000
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 8.922790127687859e-05,
-      "loss": 1.9991,
-      "step": 113500
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 8.917597453512585e-05,
-      "loss": 2.0034,
-      "step": 114000
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 8.912415164685662e-05,
-      "loss": 1.9974,
-      "step": 114500
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 8.907222490510389e-05,
-      "loss": 1.9995,
-      "step": 115000
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 8.902029816335114e-05,
-      "loss": 1.9954,
-      "step": 115500
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 8.896837142159841e-05,
-      "loss": 1.9951,
-      "step": 116000
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 8.891654853332918e-05,
-      "loss": 1.9955,
-      "step": 116500
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 8.886462179157645e-05,
-      "loss": 1.9954,
-      "step": 117000
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 8.881269504982372e-05,
-      "loss": 1.9907,
-      "step": 117500
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 8.876076830807097e-05,
-      "loss": 1.9932,
-      "step": 118000
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 8.870894541980175e-05,
-      "loss": 1.9925,
-      "step": 118500
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 8.8657018678049e-05,
-      "loss": 1.9931,
-      "step": 119000
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 8.860509193629629e-05,
-      "loss": 1.9931,
-      "step": 119500
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 8.855316519454354e-05,
-      "loss": 1.9863,
-      "step": 120000
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 8.850134230627432e-05,
-      "loss": 1.9922,
-      "step": 120500
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 8.844941556452158e-05,
-      "loss": 1.985,
-      "step": 121000
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 8.839748882276883e-05,
-      "loss": 1.9867,
-      "step": 121500
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 8.834556208101611e-05,
-      "loss": 1.988,
-      "step": 122000
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 8.829373919274686e-05,
-      "loss": 1.9858,
-      "step": 122500
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 8.824181245099415e-05,
-      "loss": 1.9871,
-      "step": 123000
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 8.81898857092414e-05,
-      "loss": 1.9836,
-      "step": 123500
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 8.813795896748867e-05,
-      "loss": 1.9807,
-      "step": 124000
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 8.808613607921944e-05,
-      "loss": 1.9758,
-      "step": 124500
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 8.80342093374667e-05,
-      "loss": 1.9839,
-      "step": 125000
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 8.798228259571397e-05,
-      "loss": 1.9819,
-      "step": 125500
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 8.793035585396123e-05,
-      "loss": 1.9805,
-      "step": 126000
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 8.787853296569201e-05,
-      "loss": 1.9809,
-      "step": 126500
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 8.782660622393926e-05,
-      "loss": 1.9795,
-      "step": 127000
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.777467948218654e-05,
-      "loss": 1.9802,
-      "step": 127500
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.77227527404338e-05,
-      "loss": 1.9773,
-      "step": 128000
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 8.767082599868107e-05,
-      "loss": 1.9793,
-      "step": 128500
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 8.761900311041183e-05,
-      "loss": 1.9731,
-      "step": 129000
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 8.75670763686591e-05,
-      "loss": 1.973,
-      "step": 129500
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 8.751514962690637e-05,
-      "loss": 1.9748,
-      "step": 130000
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 8.746322288515362e-05,
-      "loss": 1.9715,
-      "step": 130500
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 8.74113999968844e-05,
-      "loss": 1.9731,
-      "step": 131000
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 8.735947325513166e-05,
-      "loss": 1.9728,
-      "step": 131500
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 8.730754651337893e-05,
-      "loss": 1.9709,
-      "step": 132000
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 8.72556197716262e-05,
-      "loss": 1.9718,
-      "step": 132500
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 8.720379688335698e-05,
-      "loss": 1.9669,
-      "step": 133000
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 8.715187014160423e-05,
-      "loss": 1.9707,
-      "step": 133500
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 8.709994339985148e-05,
-      "loss": 1.9744,
-      "step": 134000
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 8.704801665809877e-05,
-      "loss": 1.9699,
-      "step": 134500
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 8.699619376982952e-05,
-      "loss": 1.9719,
-      "step": 135000
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 8.69442670280768e-05,
-      "loss": 1.9733,
-      "step": 135500
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 8.689234028632406e-05,
-      "loss": 1.9681,
-      "step": 136000
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 8.684041354457132e-05,
-      "loss": 1.9698,
-      "step": 136500
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 8.678859065630209e-05,
-      "loss": 1.9708,
-      "step": 137000
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 8.673666391454936e-05,
-      "loss": 1.9622,
-      "step": 137500
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 8.668473717279663e-05,
-      "loss": 1.9627,
-      "step": 138000
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 8.663281043104388e-05,
-      "loss": 1.9667,
-      "step": 138500
-    },
-    {
-      "epoch": 2.0,
-      "eval_accuracy": 0.6277449496523109,
-      "eval_loss": 1.833423137664795,
-      "eval_runtime": 1270.2848,
-      "eval_samples_per_second": 424.265,
-      "eval_steps_per_second": 26.517,
-      "step": 138946
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 8.658098754277466e-05,
-      "loss": 1.9637,
-      "step": 139000
-    },
-    {
-      "epoch": 2.01,
-      "learning_rate": 8.652906080102192e-05,
-      "loss": 1.9594,
-      "step": 139500
-    },
-    {
-      "epoch": 2.02,
-      "learning_rate": 8.647713405926918e-05,
-      "loss": 1.9558,
-      "step": 140000
-    },
-    {
-      "epoch": 2.02,
-      "learning_rate": 8.642520731751645e-05,
-      "loss": 1.959,
-      "step": 140500
-    },
-    {
-      "epoch": 2.03,
-      "learning_rate": 8.637338442924722e-05,
-      "loss": 1.9601,
-      "step": 141000
-    },
-    {
-      "epoch": 2.04,
-      "learning_rate": 8.632145768749449e-05,
-      "loss": 1.9579,
-      "step": 141500
-    },
-    {
-      "epoch": 2.04,
-      "learning_rate": 8.626953094574176e-05,
-      "loss": 1.9583,
-      "step": 142000
-    },
-    {
-      "epoch": 2.05,
-      "learning_rate": 8.621760420398902e-05,
-      "loss": 1.9565,
-      "step": 142500
-    },
-    {
-      "epoch": 2.06,
-      "learning_rate": 8.616578131571979e-05,
-      "loss": 1.9591,
-      "step": 143000
-    },
-    {
-      "epoch": 2.07,
-      "learning_rate": 8.611385457396706e-05,
-      "loss": 1.9529,
-      "step": 143500
-    },
-    {
-      "epoch": 2.07,
-      "learning_rate": 8.606192783221431e-05,
-      "loss": 1.9517,
-      "step": 144000
-    },
-    {
-      "epoch": 2.08,
-      "learning_rate": 8.601000109046158e-05,
-      "loss": 1.9522,
-      "step": 144500
-    },
-    {
-      "epoch": 2.09,
-      "learning_rate": 8.595817820219235e-05,
-      "loss": 1.9543,
-      "step": 145000
-    },
-    {
-      "epoch": 2.09,
-      "learning_rate": 8.590625146043962e-05,
-      "loss": 1.9546,
-      "step": 145500
-    },
-    {
-      "epoch": 2.1,
-      "learning_rate": 8.585432471868688e-05,
-      "loss": 1.9529,
-      "step": 146000
-    },
-    {
-      "epoch": 2.11,
-      "learning_rate": 8.580239797693414e-05,
-      "loss": 1.951,
-      "step": 146500
-    },
-    {
-      "epoch": 2.12,
-      "learning_rate": 8.575057508866492e-05,
-      "loss": 1.9542,
-      "step": 147000
-    },
-    {
-      "epoch": 2.12,
-      "learning_rate": 8.569864834691219e-05,
-      "loss": 1.9513,
-      "step": 147500
-    },
-    {
-      "epoch": 2.13,
-      "learning_rate": 8.564672160515944e-05,
-      "loss": 1.9507,
-      "step": 148000
-    },
-    {
-      "epoch": 2.14,
-      "learning_rate": 8.559479486340671e-05,
-      "loss": 1.948,
-      "step": 148500
-    },
-    {
-      "epoch": 2.14,
-      "learning_rate": 8.554297197513748e-05,
-      "loss": 1.9506,
-      "step": 149000
-    },
-    {
-      "epoch": 2.15,
-      "learning_rate": 8.549104523338474e-05,
-      "loss": 1.9474,
-      "step": 149500
-    },
-    {
-      "epoch": 2.16,
-      "learning_rate": 8.543911849163201e-05,
-      "loss": 1.9465,
-      "step": 150000
-    },
-    {
-      "epoch": 2.17,
-      "learning_rate": 8.538719174987927e-05,
-      "loss": 1.9501,
-      "step": 150500
-    },
-    {
-      "epoch": 2.17,
-      "learning_rate": 8.533536886161005e-05,
-      "loss": 1.9468,
-      "step": 151000
-    },
-    {
-      "epoch": 2.18,
-      "learning_rate": 8.52834421198573e-05,
-      "loss": 1.9448,
-      "step": 151500
-    },
-    {
-      "epoch": 2.19,
-      "learning_rate": 8.523151537810457e-05,
-      "loss": 1.947,
-      "step": 152000
-    },
-    {
-      "epoch": 2.2,
-      "learning_rate": 8.517958863635184e-05,
-      "loss": 1.9451,
-      "step": 152500
-    },
-    {
-      "epoch": 2.2,
-      "learning_rate": 8.51277657480826e-05,
-      "loss": 1.9453,
-      "step": 153000
-    },
-    {
-      "epoch": 2.21,
-      "learning_rate": 8.507583900632987e-05,
-      "loss": 1.9474,
-      "step": 153500
-    },
-    {
-      "epoch": 2.22,
-      "learning_rate": 8.502391226457714e-05,
-      "loss": 1.9443,
-      "step": 154000
-    },
-    {
-      "epoch": 2.22,
-      "learning_rate": 8.497198552282441e-05,
-      "loss": 1.9418,
-      "step": 154500
-    },
-    {
-      "epoch": 2.23,
-      "learning_rate": 8.492016263455518e-05,
-      "loss": 1.9412,
-      "step": 155000
-    },
-    {
-      "epoch": 2.24,
-      "learning_rate": 8.486823589280244e-05,
-      "loss": 1.9425,
-      "step": 155500
-    },
-    {
-      "epoch": 2.25,
-      "learning_rate": 8.48163091510497e-05,
-      "loss": 1.9403,
-      "step": 156000
-    },
-    {
-      "epoch": 2.25,
-      "learning_rate": 8.476438240929697e-05,
-      "loss": 1.9469,
-      "step": 156500
-    },
-    {
-      "epoch": 2.26,
-      "learning_rate": 8.471255952102773e-05,
-      "loss": 1.9344,
-      "step": 157000
-    },
-    {
-      "epoch": 2.27,
-      "learning_rate": 8.4660632779275e-05,
-      "loss": 1.9409,
-      "step": 157500
-    },
-    {
-      "epoch": 2.27,
-      "learning_rate": 8.460870603752227e-05,
-      "loss": 1.9353,
-      "step": 158000
-    },
-    {
-      "epoch": 2.28,
-      "learning_rate": 8.455677929576953e-05,
-      "loss": 1.94,
-      "step": 158500
-    },
-    {
-      "epoch": 2.29,
-      "learning_rate": 8.45049564075003e-05,
-      "loss": 1.9376,
-      "step": 159000
-    },
-    {
-      "epoch": 2.3,
-      "learning_rate": 8.445302966574756e-05,
-      "loss": 1.9371,
-      "step": 159500
-    },
-    {
-      "epoch": 2.3,
-      "learning_rate": 8.440110292399484e-05,
-      "loss": 1.938,
-      "step": 160000
-    },
-    {
-      "epoch": 2.31,
-      "learning_rate": 8.43491761822421e-05,
-      "loss": 1.9371,
-      "step": 160500
-    },
-    {
-      "epoch": 2.32,
-      "learning_rate": 8.429735329397288e-05,
-      "loss": 1.9377,
-      "step": 161000
-    },
-    {
-      "epoch": 2.32,
-      "learning_rate": 8.424542655222013e-05,
-      "loss": 1.9352,
-      "step": 161500
-    },
-    {
-      "epoch": 2.33,
-      "learning_rate": 8.419349981046739e-05,
-      "loss": 1.9325,
-      "step": 162000
-    },
-    {
-      "epoch": 2.34,
-      "learning_rate": 8.414157306871467e-05,
-      "loss": 1.9353,
-      "step": 162500
-    },
-    {
-      "epoch": 2.35,
-      "learning_rate": 8.408975018044543e-05,
-      "loss": 1.9306,
-      "step": 163000
-    },
-    {
-      "epoch": 2.35,
-      "learning_rate": 8.40378234386927e-05,
-      "loss": 1.9349,
-      "step": 163500
-    },
-    {
-      "epoch": 2.36,
-      "learning_rate": 8.398589669693996e-05,
-      "loss": 1.9311,
-      "step": 164000
-    },
-    {
-      "epoch": 2.37,
-      "learning_rate": 8.393396995518722e-05,
-      "loss": 1.931,
-      "step": 164500
-    },
-    {
-      "epoch": 2.38,
-      "learning_rate": 8.388214706691799e-05,
-      "loss": 1.9308,
-      "step": 165000
-    },
-    {
-      "epoch": 2.38,
-      "learning_rate": 8.383022032516526e-05,
-      "loss": 1.931,
-      "step": 165500
-    },
-    {
-      "epoch": 2.39,
-      "learning_rate": 8.377829358341253e-05,
-      "loss": 1.9317,
-      "step": 166000
-    },
-    {
-      "epoch": 2.4,
-      "learning_rate": 8.372636684165978e-05,
-      "loss": 1.931,
-      "step": 166500
-    },
-    {
-      "epoch": 2.4,
-      "learning_rate": 8.367454395339056e-05,
-      "loss": 1.9337,
-      "step": 167000
-    },
-    {
-      "epoch": 2.41,
-      "learning_rate": 8.362261721163782e-05,
-      "loss": 1.9331,
-      "step": 167500
-    },
-    {
-      "epoch": 2.42,
-      "learning_rate": 8.35706904698851e-05,
-      "loss": 1.9228,
-      "step": 168000
-    },
-    {
-      "epoch": 2.43,
-      "learning_rate": 8.351876372813235e-05,
-      "loss": 1.9276,
-      "step": 168500
-    },
-    {
-      "epoch": 2.43,
-      "learning_rate": 8.346694083986313e-05,
-      "loss": 1.9263,
-      "step": 169000
-    },
-    {
-      "epoch": 2.44,
-      "learning_rate": 8.341501409811039e-05,
-      "loss": 1.928,
-      "step": 169500
-    },
-    {
-      "epoch": 2.45,
-      "learning_rate": 8.336308735635766e-05,
-      "loss": 1.9274,
-      "step": 170000
-    },
-    {
-      "epoch": 2.45,
-      "learning_rate": 8.331116061460492e-05,
-      "loss": 1.9253,
-      "step": 170500
-    },
-    {
-      "epoch": 2.46,
-      "learning_rate": 8.325933772633569e-05,
-      "loss": 1.9253,
-      "step": 171000
-    },
-    {
-      "epoch": 2.47,
-      "learning_rate": 8.320741098458296e-05,
-      "loss": 1.9242,
-      "step": 171500
-    },
-    {
-      "epoch": 2.48,
-      "learning_rate": 8.315548424283021e-05,
-      "loss": 1.922,
-      "step": 172000
-    },
-    {
-      "epoch": 2.48,
-      "learning_rate": 8.31035575010775e-05,
-      "loss": 1.9241,
-      "step": 172500
-    },
-    {
-      "epoch": 2.49,
-      "learning_rate": 8.305173461280825e-05,
-      "loss": 1.9222,
-      "step": 173000
-    },
-    {
-      "epoch": 2.5,
-      "learning_rate": 8.299980787105553e-05,
-      "loss": 1.9222,
-      "step": 173500
-    },
-    {
-      "epoch": 2.5,
-      "learning_rate": 8.294788112930279e-05,
-      "loss": 1.921,
-      "step": 174000
-    },
-    {
-      "epoch": 2.51,
-      "learning_rate": 8.289595438755004e-05,
-      "loss": 1.925,
-      "step": 174500
-    },
-    {
-      "epoch": 2.52,
-      "learning_rate": 8.284413149928082e-05,
-      "loss": 1.9215,
-      "step": 175000
-    },
-    {
-      "epoch": 2.53,
-      "learning_rate": 8.279220475752807e-05,
-      "loss": 1.9219,
-      "step": 175500
-    },
-    {
-      "epoch": 2.53,
-      "learning_rate": 8.274027801577536e-05,
-      "loss": 1.9272,
-      "step": 176000
-    },
-    {
-      "epoch": 2.54,
-      "learning_rate": 8.268835127402261e-05,
-      "loss": 1.9202,
-      "step": 176500
-    },
-    {
-      "epoch": 2.55,
-      "learning_rate": 8.263642453226988e-05,
-      "loss": 1.9235,
-      "step": 177000
-    },
-    {
-      "epoch": 2.55,
-      "learning_rate": 8.258460164400065e-05,
-      "loss": 1.92,
-      "step": 177500
-    },
-    {
-      "epoch": 2.56,
-      "learning_rate": 8.253267490224791e-05,
-      "loss": 1.915,
-      "step": 178000
-    },
-    {
-      "epoch": 2.57,
-      "learning_rate": 8.248074816049518e-05,
-      "loss": 1.9147,
-      "step": 178500
-    },
-    {
-      "epoch": 2.58,
-      "learning_rate": 8.242882141874244e-05,
-      "loss": 1.9213,
-      "step": 179000
-    },
-    {
-      "epoch": 2.58,
-      "learning_rate": 8.237699853047322e-05,
-      "loss": 1.9187,
-      "step": 179500
-    },
-    {
-      "epoch": 2.59,
-      "learning_rate": 8.232507178872047e-05,
-      "loss": 1.9176,
-      "step": 180000
-    },
-    {
-      "epoch": 2.6,
-      "learning_rate": 8.227314504696774e-05,
-      "loss": 1.9149,
-      "step": 180500
-    },
-    {
-      "epoch": 2.61,
-      "learning_rate": 8.222121830521501e-05,
-      "loss": 1.9225,
-      "step": 181000
-    },
-    {
-      "epoch": 2.61,
-      "learning_rate": 8.216939541694577e-05,
-      "loss": 1.9187,
-      "step": 181500
-    },
-    {
-      "epoch": 2.62,
-      "learning_rate": 8.211746867519304e-05,
-      "loss": 1.9136,
-      "step": 182000
-    },
-    {
-      "epoch": 2.63,
-      "learning_rate": 8.206554193344031e-05,
-      "loss": 1.9119,
-      "step": 182500
-    },
-    {
-      "epoch": 2.63,
-      "learning_rate": 8.201361519168758e-05,
-      "loss": 1.9129,
-      "step": 183000
-    },
-    {
-      "epoch": 2.64,
-      "learning_rate": 8.196179230341835e-05,
-      "loss": 1.9192,
-      "step": 183500
-    },
-    {
-      "epoch": 2.65,
-      "learning_rate": 8.190986556166561e-05,
-      "loss": 1.9172,
-      "step": 184000
-    },
-    {
-      "epoch": 2.66,
-      "learning_rate": 8.185793881991287e-05,
-      "loss": 1.9185,
-      "step": 184500
-    },
-    {
-      "epoch": 2.66,
-      "learning_rate": 8.180601207816014e-05,
-      "loss": 1.9126,
-      "step": 185000
-    },
-    {
-      "epoch": 2.67,
-      "learning_rate": 8.17541891898909e-05,
-      "loss": 1.9156,
-      "step": 185500
-    },
-    {
-      "epoch": 2.68,
-      "learning_rate": 8.170226244813817e-05,
-      "loss": 1.9131,
-      "step": 186000
-    },
-    {
-      "epoch": 2.68,
-      "learning_rate": 8.165033570638544e-05,
-      "loss": 1.9099,
-      "step": 186500
-    },
-    {
-      "epoch": 2.69,
-      "learning_rate": 8.15984089646327e-05,
-      "loss": 1.9111,
-      "step": 187000
-    },
-    {
-      "epoch": 2.7,
-      "learning_rate": 8.154658607636347e-05,
-      "loss": 1.9124,
-      "step": 187500
-    },
-    {
-      "epoch": 2.71,
-      "learning_rate": 8.149465933461073e-05,
-      "loss": 1.9158,
-      "step": 188000
-    },
-    {
-      "epoch": 2.71,
-      "learning_rate": 8.1442732592858e-05,
-      "loss": 1.9092,
-      "step": 188500
-    },
-    {
-      "epoch": 2.72,
-      "learning_rate": 8.139080585110527e-05,
-      "loss": 1.9116,
-      "step": 189000
-    },
-    {
-      "epoch": 2.73,
-      "learning_rate": 8.133898296283603e-05,
-      "loss": 1.9074,
-      "step": 189500
-    },
-    {
-      "epoch": 2.73,
-      "learning_rate": 8.12870562210833e-05,
-      "loss": 1.9119,
-      "step": 190000
-    },
-    {
-      "epoch": 2.74,
-      "learning_rate": 8.123512947933057e-05,
-      "loss": 1.9099,
-      "step": 190500
-    },
-    {
-      "epoch": 2.75,
-      "learning_rate": 8.118320273757782e-05,
-      "loss": 1.907,
-      "step": 191000
-    },
-    {
-      "epoch": 2.76,
-      "learning_rate": 8.11313798493086e-05,
-      "loss": 1.9111,
-      "step": 191500
-    },
-    {
-      "epoch": 2.76,
-      "learning_rate": 8.107945310755587e-05,
-      "loss": 1.9092,
-      "step": 192000
-    },
-    {
-      "epoch": 2.77,
-      "learning_rate": 8.102752636580313e-05,
-      "loss": 1.9073,
-      "step": 192500
-    },
-    {
-      "epoch": 2.78,
-      "learning_rate": 8.09755996240504e-05,
-      "loss": 1.9052,
-      "step": 193000
-    },
-    {
-      "epoch": 2.79,
-      "learning_rate": 8.092377673578116e-05,
-      "loss": 1.9036,
-      "step": 193500
-    },
-    {
-      "epoch": 2.79,
-      "learning_rate": 8.087184999402843e-05,
-      "loss": 1.907,
-      "step": 194000
-    },
-    {
-      "epoch": 2.8,
-      "learning_rate": 8.08199232522757e-05,
-      "loss": 1.9082,
-      "step": 194500
-    },
-    {
-      "epoch": 2.81,
-      "learning_rate": 8.076799651052297e-05,
-      "loss": 1.9103,
-      "step": 195000
-    },
-    {
-      "epoch": 2.81,
-      "learning_rate": 8.071617362225373e-05,
-      "loss": 1.9028,
-      "step": 195500
-    },
-    {
-      "epoch": 2.82,
-      "learning_rate": 8.0664246880501e-05,
-      "loss": 1.9093,
-      "step": 196000
-    },
-    {
-      "epoch": 2.83,
-      "learning_rate": 8.061232013874825e-05,
-      "loss": 1.9014,
-      "step": 196500
-    },
-    {
-      "epoch": 2.84,
-      "learning_rate": 8.056039339699552e-05,
-      "loss": 1.9039,
-      "step": 197000
-    },
-    {
-      "epoch": 2.84,
-      "learning_rate": 8.050857050872629e-05,
-      "loss": 1.9032,
-      "step": 197500
-    },
-    {
-      "epoch": 2.85,
-      "learning_rate": 8.045664376697356e-05,
-      "loss": 1.9024,
-      "step": 198000
-    },
-    {
-      "epoch": 2.86,
-      "learning_rate": 8.040471702522083e-05,
-      "loss": 1.901,
-      "step": 198500
-    },
-    {
-      "epoch": 2.86,
-      "learning_rate": 8.035279028346808e-05,
-      "loss": 1.9027,
-      "step": 199000
-    },
-    {
-      "epoch": 2.87,
-      "learning_rate": 8.030096739519886e-05,
-      "loss": 1.9047,
-      "step": 199500
-    },
-    {
-      "epoch": 2.88,
-      "learning_rate": 8.024904065344611e-05,
-      "loss": 1.9004,
-      "step": 200000
-    },
-    {
-      "epoch": 2.89,
-      "learning_rate": 8.019711391169338e-05,
-      "loss": 1.9018,
-      "step": 200500
-    },
-    {
-      "epoch": 2.89,
-      "learning_rate": 8.014518716994065e-05,
-      "loss": 1.9016,
-      "step": 201000
-    },
-    {
-      "epoch": 2.9,
-      "learning_rate": 8.009336428167142e-05,
-      "loss": 1.8981,
-      "step": 201500
-    },
-    {
-      "epoch": 2.91,
-      "learning_rate": 8.004143753991869e-05,
-      "loss": 1.8975,
-      "step": 202000
-    },
-    {
-      "epoch": 2.91,
-      "learning_rate": 7.998951079816595e-05,
-      "loss": 1.8993,
-      "step": 202500
-    },
-    {
-      "epoch": 2.92,
-      "learning_rate": 7.993758405641322e-05,
-      "loss": 1.8978,
-      "step": 203000
-    },
-    {
-      "epoch": 2.93,
-      "learning_rate": 7.988576116814399e-05,
-      "loss": 1.8996,
-      "step": 203500
-    },
-    {
-      "epoch": 2.94,
-      "learning_rate": 7.983383442639126e-05,
-      "loss": 1.898,
-      "step": 204000
-    },
-    {
-      "epoch": 2.94,
-      "learning_rate": 7.978190768463851e-05,
-      "loss": 1.8992,
-      "step": 204500
-    },
-    {
-      "epoch": 2.95,
-      "learning_rate": 7.972998094288578e-05,
-      "loss": 1.8987,
-      "step": 205000
-    },
-    {
-      "epoch": 2.96,
-      "learning_rate": 7.967815805461655e-05,
-      "loss": 1.8968,
-      "step": 205500
-    },
-    {
-      "epoch": 2.97,
-      "learning_rate": 7.962623131286381e-05,
-      "loss": 1.8949,
-      "step": 206000
-    },
-    {
-      "epoch": 2.97,
-      "learning_rate": 7.957430457111108e-05,
-      "loss": 1.8961,
-      "step": 206500
-    },
-    {
-      "epoch": 2.98,
-      "learning_rate": 7.952237782935834e-05,
-      "loss": 1.8963,
-      "step": 207000
-    },
-    {
-      "epoch": 2.99,
-      "learning_rate": 7.947055494108912e-05,
-      "loss": 1.8949,
-      "step": 207500
-    },
-    {
-      "epoch": 2.99,
-      "learning_rate": 7.941862819933637e-05,
-      "loss": 1.8918,
-      "step": 208000
-    },
-    {
-      "epoch": 3.0,
-      "eval_accuracy": 0.6376255333966128,
-      "eval_loss": 1.765644907951355,
-      "eval_runtime": 1267.2057,
-      "eval_samples_per_second": 425.296,
-      "eval_steps_per_second": 26.581,
-      "step": 208419
-    },
-    {
-      "epoch": 3.0,
-      "learning_rate": 7.936670145758365e-05,
-      "loss": 1.896,
-      "step": 208500
-    },
-    {
-      "epoch": 3.01,
-      "learning_rate": 7.931477471583091e-05,
-      "loss": 1.889,
-      "step": 209000
-    },
-    {
-      "epoch": 3.02,
-      "learning_rate": 7.926295182756169e-05,
-      "loss": 1.888,
-      "step": 209500
-    },
-    {
-      "epoch": 3.02,
-      "learning_rate": 7.921102508580894e-05,
-      "loss": 1.8899,
-      "step": 210000
-    },
-    {
-      "epoch": 3.03,
-      "learning_rate": 7.91590983440562e-05,
-      "loss": 1.8929,
-      "step": 210500
-    },
-    {
-      "epoch": 3.04,
-      "learning_rate": 7.910717160230348e-05,
-      "loss": 1.8867,
-      "step": 211000
-    },
-    {
-      "epoch": 3.04,
-      "learning_rate": 7.905524486055073e-05,
-      "loss": 1.8923,
-      "step": 211500
-    },
-    {
-      "epoch": 3.05,
-      "learning_rate": 7.900342197228151e-05,
-      "loss": 1.8881,
-      "step": 212000
-    },
-    {
-      "epoch": 3.06,
-      "learning_rate": 7.895149523052877e-05,
-      "loss": 1.8883,
-      "step": 212500
-    },
-    {
-      "epoch": 3.07,
-      "learning_rate": 7.889956848877604e-05,
-      "loss": 1.8874,
-      "step": 213000
-    },
-    {
-      "epoch": 3.07,
-      "learning_rate": 7.88476417470233e-05,
-      "loss": 1.8859,
-      "step": 213500
-    },
-    {
-      "epoch": 3.08,
-      "learning_rate": 7.879581885875407e-05,
-      "loss": 1.8889,
-      "step": 214000
-    },
-    {
-      "epoch": 3.09,
-      "learning_rate": 7.874389211700134e-05,
-      "loss": 1.8864,
-      "step": 214500
-    },
-    {
-      "epoch": 3.09,
-      "learning_rate": 7.86919653752486e-05,
-      "loss": 1.8842,
-      "step": 215000
-    },
-    {
-      "epoch": 3.1,
-      "learning_rate": 7.864003863349588e-05,
-      "loss": 1.8901,
-      "step": 215500
-    },
-    {
-      "epoch": 3.11,
-      "learning_rate": 7.858821574522663e-05,
-      "loss": 1.888,
-      "step": 216000
-    },
-    {
-      "epoch": 3.12,
-      "learning_rate": 7.853628900347391e-05,
-      "loss": 1.8866,
-      "step": 216500
-    },
-    {
-      "epoch": 3.12,
-      "learning_rate": 7.848436226172117e-05,
-      "loss": 1.8815,
-      "step": 217000
-    },
-    {
-      "epoch": 3.13,
-      "learning_rate": 7.843243551996843e-05,
-      "loss": 1.887,
-      "step": 217500
-    },
-    {
-      "epoch": 3.14,
-      "learning_rate": 7.83806126316992e-05,
-      "loss": 1.8878,
-      "step": 218000
-    },
-    {
-      "epoch": 3.15,
-      "learning_rate": 7.832868588994647e-05,
-      "loss": 1.8863,
-      "step": 218500
-    },
-    {
-      "epoch": 3.15,
-      "learning_rate": 7.827675914819374e-05,
-      "loss": 1.8873,
-      "step": 219000
-    },
-    {
-      "epoch": 3.16,
-      "learning_rate": 7.822483240644099e-05,
-      "loss": 1.8812,
-      "step": 219500
-    },
-    {
-      "epoch": 3.17,
-      "learning_rate": 7.817300951817177e-05,
-      "loss": 1.8862,
-      "step": 220000
-    },
-    {
-      "epoch": 3.17,
-      "learning_rate": 7.812108277641903e-05,
-      "loss": 1.8852,
-      "step": 220500
-    },
-    {
-      "epoch": 3.18,
-      "learning_rate": 7.806915603466631e-05,
-      "loss": 1.886,
-      "step": 221000
-    },
-    {
-      "epoch": 3.19,
-      "learning_rate": 7.801722929291356e-05,
-      "loss": 1.8846,
-      "step": 221500
-    },
-    {
-      "epoch": 3.2,
-      "learning_rate": 7.796540640464434e-05,
-      "loss": 1.8843,
-      "step": 222000
-    },
-    {
-      "epoch": 3.2,
-      "learning_rate": 7.79134796628916e-05,
-      "loss": 1.8819,
-      "step": 222500
-    },
-    {
-      "epoch": 3.21,
-      "learning_rate": 7.786155292113885e-05,
-      "loss": 1.8774,
-      "step": 223000
-    },
-    {
-      "epoch": 3.22,
-      "learning_rate": 7.780962617938613e-05,
-      "loss": 1.8817,
-      "step": 223500
-    },
-    {
-      "epoch": 3.22,
-      "learning_rate": 7.775780329111689e-05,
-      "loss": 1.8838,
-      "step": 224000
-    },
-    {
-      "epoch": 3.23,
-      "learning_rate": 7.770587654936417e-05,
-      "loss": 1.8812,
-      "step": 224500
-    },
-    {
-      "epoch": 3.24,
-      "learning_rate": 7.765394980761142e-05,
-      "loss": 1.8824,
-      "step": 225000
-    },
-    {
-      "epoch": 3.25,
-      "learning_rate": 7.760202306585869e-05,
-      "loss": 1.8842,
-      "step": 225500
-    },
-    {
-      "epoch": 3.25,
-      "learning_rate": 7.755020017758946e-05,
-      "loss": 1.8833,
-      "step": 226000
-    },
-    {
-      "epoch": 3.26,
-      "learning_rate": 7.749827343583673e-05,
-      "loss": 1.8815,
-      "step": 226500
-    },
-    {
-      "epoch": 3.27,
-      "learning_rate": 7.7446346694084e-05,
-      "loss": 1.879,
-      "step": 227000
-    },
-    {
-      "epoch": 3.27,
-      "learning_rate": 7.739441995233125e-05,
-      "loss": 1.8779,
-      "step": 227500
-    },
-    {
-      "epoch": 3.28,
-      "learning_rate": 7.734259706406203e-05,
-      "loss": 1.8777,
-      "step": 228000
-    },
-    {
-      "epoch": 3.29,
-      "learning_rate": 7.729067032230928e-05,
-      "loss": 1.8768,
-      "step": 228500
-    },
-    {
-      "epoch": 3.3,
-      "learning_rate": 7.723874358055655e-05,
-      "loss": 1.8804,
-      "step": 229000
-    },
-    {
-      "epoch": 3.3,
-      "learning_rate": 7.718681683880382e-05,
-      "loss": 1.8789,
-      "step": 229500
-    },
-    {
-      "epoch": 3.31,
-      "learning_rate": 7.713499395053459e-05,
-      "loss": 1.8769,
-      "step": 230000
-    },
-    {
-      "epoch": 3.32,
-      "learning_rate": 7.708306720878185e-05,
-      "loss": 1.882,
-      "step": 230500
-    },
-    {
-      "epoch": 3.33,
-      "learning_rate": 7.703114046702912e-05,
-      "loss": 1.878,
-      "step": 231000
-    },
-    {
-      "epoch": 3.33,
-      "learning_rate": 7.697921372527639e-05,
-      "loss": 1.8774,
-      "step": 231500
-    },
-    {
-      "epoch": 3.34,
-      "learning_rate": 7.692739083700716e-05,
-      "loss": 1.8789,
-      "step": 232000
-    },
-    {
-      "epoch": 3.35,
-      "learning_rate": 7.687546409525443e-05,
-      "loss": 1.877,
-      "step": 232500
-    },
-    {
-      "epoch": 3.35,
-      "learning_rate": 7.682353735350168e-05,
-      "loss": 1.8732,
-      "step": 233000
-    },
-    {
-      "epoch": 3.36,
-      "learning_rate": 7.677161061174895e-05,
-      "loss": 1.8777,
-      "step": 233500
-    },
-    {
-      "epoch": 3.37,
-      "learning_rate": 7.671978772347972e-05,
-      "loss": 1.8753,
-      "step": 234000
-    },
-    {
-      "epoch": 3.38,
-      "learning_rate": 7.666786098172698e-05,
-      "loss": 1.8741,
-      "step": 234500
-    },
-    {
-      "epoch": 3.38,
-      "learning_rate": 7.661593423997425e-05,
-      "loss": 1.8763,
-      "step": 235000
-    },
-    {
-      "epoch": 3.39,
-      "learning_rate": 7.65640074982215e-05,
-      "loss": 1.8773,
-      "step": 235500
-    },
-    {
-      "epoch": 3.4,
-      "learning_rate": 7.651218460995229e-05,
-      "loss": 1.8731,
-      "step": 236000
-    },
-    {
-      "epoch": 3.4,
-      "learning_rate": 7.646025786819954e-05,
-      "loss": 1.8696,
-      "step": 236500
-    },
-    {
-      "epoch": 3.41,
-      "learning_rate": 7.640833112644681e-05,
-      "loss": 1.8737,
-      "step": 237000
-    },
-    {
-      "epoch": 3.42,
-      "learning_rate": 7.635640438469408e-05,
-      "loss": 1.8765,
-      "step": 237500
-    },
-    {
-      "epoch": 3.43,
-      "learning_rate": 7.630458149642484e-05,
-      "loss": 1.8751,
-      "step": 238000
-    },
-    {
-      "epoch": 3.43,
-      "learning_rate": 7.625265475467211e-05,
-      "loss": 1.8759,
-      "step": 238500
-    },
-    {
-      "epoch": 3.44,
-      "learning_rate": 7.620072801291938e-05,
-      "loss": 1.8716,
-      "step": 239000
-    },
-    {
-      "epoch": 3.45,
-      "learning_rate": 7.614880127116664e-05,
-      "loss": 1.8744,
-      "step": 239500
-    },
-    {
-      "epoch": 3.45,
-      "learning_rate": 7.609697838289742e-05,
-      "loss": 1.8746,
-      "step": 240000
-    },
-    {
-      "epoch": 3.46,
-      "learning_rate": 7.604505164114468e-05,
-      "loss": 1.8747,
-      "step": 240500
-    },
-    {
-      "epoch": 3.47,
-      "learning_rate": 7.599312489939194e-05,
-      "loss": 1.8738,
-      "step": 241000
-    },
-    {
-      "epoch": 3.48,
-      "learning_rate": 7.59411981576392e-05,
-      "loss": 1.8706,
-      "step": 241500
-    },
-    {
-      "epoch": 3.48,
-      "learning_rate": 7.588937526936997e-05,
-      "loss": 1.8679,
-      "step": 242000
-    },
-    {
-      "epoch": 3.49,
-      "learning_rate": 7.583744852761724e-05,
-      "loss": 1.8713,
-      "step": 242500
-    },
-    {
-      "epoch": 3.5,
-      "learning_rate": 7.578552178586451e-05,
-      "loss": 1.8706,
-      "step": 243000
-    },
-    {
-      "epoch": 3.5,
-      "learning_rate": 7.573359504411178e-05,
-      "loss": 1.8719,
-      "step": 243500
-    },
-    {
-      "epoch": 3.51,
-      "learning_rate": 7.568177215584254e-05,
-      "loss": 1.8685,
-      "step": 244000
-    },
-    {
-      "epoch": 3.52,
-      "learning_rate": 7.562984541408981e-05,
-      "loss": 1.8701,
-      "step": 244500
-    },
-    {
-      "epoch": 3.53,
-      "learning_rate": 7.557791867233707e-05,
-      "loss": 1.8715,
-      "step": 245000
-    },
-    {
-      "epoch": 3.53,
-      "learning_rate": 7.552599193058434e-05,
-      "loss": 1.8671,
-      "step": 245500
-    },
-    {
-      "epoch": 3.54,
-      "learning_rate": 7.54741690423151e-05,
-      "loss": 1.871,
-      "step": 246000
-    },
-    {
-      "epoch": 3.55,
-      "learning_rate": 7.542224230056237e-05,
-      "loss": 1.8719,
-      "step": 246500
-    },
-    {
-      "epoch": 3.56,
-      "learning_rate": 7.537031555880964e-05,
-      "loss": 1.8658,
-      "step": 247000
-    },
-    {
-      "epoch": 3.56,
-      "learning_rate": 7.531838881705689e-05,
-      "loss": 1.8702,
-      "step": 247500
-    },
-    {
-      "epoch": 3.57,
-      "learning_rate": 7.526656592878767e-05,
-      "loss": 1.8717,
-      "step": 248000
-    },
-    {
-      "epoch": 3.58,
-      "learning_rate": 7.521463918703493e-05,
-      "loss": 1.8633,
-      "step": 248500
-    },
-    {
-      "epoch": 3.58,
-      "learning_rate": 7.51627124452822e-05,
-      "loss": 1.8689,
-      "step": 249000
-    },
-    {
-      "epoch": 3.59,
-      "learning_rate": 7.511078570352946e-05,
-      "loss": 1.8647,
-      "step": 249500
-    },
-    {
-      "epoch": 3.6,
-      "learning_rate": 7.505896281526023e-05,
-      "loss": 1.868,
-      "step": 250000
-    },
-    {
-      "epoch": 3.61,
-      "learning_rate": 7.50070360735075e-05,
-      "loss": 1.8682,
-      "step": 250500
-    },
-    {
-      "epoch": 3.61,
-      "learning_rate": 7.495510933175477e-05,
-      "loss": 1.8668,
-      "step": 251000
-    },
-    {
-      "epoch": 3.62,
-      "learning_rate": 7.490318259000203e-05,
-      "loss": 1.8652,
-      "step": 251500
-    },
-    {
-      "epoch": 3.63,
-      "learning_rate": 7.48513597017328e-05,
-      "loss": 1.8655,
-      "step": 252000
-    },
-    {
-      "epoch": 3.63,
-      "learning_rate": 7.479943295998007e-05,
-      "loss": 1.8658,
-      "step": 252500
-    },
-    {
-      "epoch": 3.64,
-      "learning_rate": 7.474750621822732e-05,
-      "loss": 1.8606,
-      "step": 253000
-    },
-    {
-      "epoch": 3.65,
-      "learning_rate": 7.469557947647459e-05,
-      "loss": 1.8669,
-      "step": 253500
-    },
-    {
-      "epoch": 3.66,
-      "learning_rate": 7.464375658820536e-05,
-      "loss": 1.8661,
-      "step": 254000
-    },
-    {
-      "epoch": 3.66,
-      "learning_rate": 7.459182984645263e-05,
-      "loss": 1.8636,
-      "step": 254500
-    },
-    {
-      "epoch": 3.67,
-      "learning_rate": 7.45399031046999e-05,
-      "loss": 1.8663,
-      "step": 255000
-    },
-    {
-      "epoch": 3.68,
-      "learning_rate": 7.448797636294715e-05,
-      "loss": 1.8656,
-      "step": 255500
-    },
-    {
-      "epoch": 3.68,
-      "learning_rate": 7.443615347467793e-05,
-      "loss": 1.8647,
-      "step": 256000
-    },
-    {
-      "epoch": 3.69,
-      "learning_rate": 7.438422673292518e-05,
-      "loss": 1.8668,
-      "step": 256500
-    },
-    {
-      "epoch": 3.7,
-      "learning_rate": 7.433229999117247e-05,
-      "loss": 1.8636,
-      "step": 257000
-    },
-    {
-      "epoch": 3.71,
-      "learning_rate": 7.428037324941972e-05,
-      "loss": 1.8654,
-      "step": 257500
-    },
-    {
-      "epoch": 3.71,
-      "learning_rate": 7.42285503611505e-05,
-      "loss": 1.8595,
-      "step": 258000
-    },
-    {
-      "epoch": 3.72,
-      "learning_rate": 7.417662361939776e-05,
-      "loss": 1.866,
-      "step": 258500
-    },
-    {
-      "epoch": 3.73,
-      "learning_rate": 7.412469687764501e-05,
-      "loss": 1.8632,
-      "step": 259000
-    },
-    {
-      "epoch": 3.74,
-      "learning_rate": 7.407277013589229e-05,
-      "loss": 1.8598,
-      "step": 259500
-    },
-    {
-      "epoch": 3.74,
-      "learning_rate": 7.402094724762304e-05,
-      "loss": 1.8656,
-      "step": 260000
-    },
-    {
-      "epoch": 3.75,
-      "learning_rate": 7.396902050587033e-05,
-      "loss": 1.8598,
-      "step": 260500
-    },
-    {
-      "epoch": 3.76,
-      "learning_rate": 7.391709376411758e-05,
-      "loss": 1.862,
-      "step": 261000
-    },
-    {
-      "epoch": 3.76,
-      "learning_rate": 7.386516702236485e-05,
-      "loss": 1.8642,
-      "step": 261500
-    },
-    {
-      "epoch": 3.77,
-      "learning_rate": 7.381334413409562e-05,
-      "loss": 1.8629,
-      "step": 262000
-    },
-    {
-      "epoch": 3.78,
-      "learning_rate": 7.376141739234288e-05,
-      "loss": 1.8634,
-      "step": 262500
-    },
-    {
-      "epoch": 3.79,
-      "learning_rate": 7.370949065059015e-05,
-      "loss": 1.8613,
-      "step": 263000
-    },
-    {
-      "epoch": 3.79,
-      "learning_rate": 7.365756390883741e-05,
-      "loss": 1.8609,
-      "step": 263500
-    },
-    {
-      "epoch": 3.8,
-      "learning_rate": 7.360574102056819e-05,
-      "loss": 1.8618,
-      "step": 264000
-    },
-    {
-      "epoch": 3.81,
-      "learning_rate": 7.355381427881544e-05,
-      "loss": 1.8611,
-      "step": 264500
-    },
-    {
-      "epoch": 3.81,
-      "learning_rate": 7.350188753706272e-05,
-      "loss": 1.8628,
-      "step": 265000
-    },
-    {
-      "epoch": 3.82,
-      "learning_rate": 7.344996079530998e-05,
-      "loss": 1.8608,
-      "step": 265500
-    },
-    {
-      "epoch": 3.83,
-      "learning_rate": 7.339813790704076e-05,
-      "loss": 1.8583,
-      "step": 266000
-    },
-    {
-      "epoch": 3.84,
-      "learning_rate": 7.334621116528801e-05,
-      "loss": 1.8558,
-      "step": 266500
-    },
-    {
-      "epoch": 3.84,
-      "learning_rate": 7.329428442353528e-05,
-      "loss": 1.8582,
-      "step": 267000
-    },
-    {
-      "epoch": 3.85,
-      "learning_rate": 7.324235768178255e-05,
-      "loss": 1.8574,
-      "step": 267500
-    },
-    {
-      "epoch": 3.86,
-      "learning_rate": 7.319053479351332e-05,
-      "loss": 1.8574,
-      "step": 268000
-    },
-    {
-      "epoch": 3.86,
-      "learning_rate": 7.313860805176058e-05,
-      "loss": 1.8562,
-      "step": 268500
-    },
-    {
-      "epoch": 3.87,
-      "learning_rate": 7.308668131000784e-05,
-      "loss": 1.8591,
-      "step": 269000
-    },
-    {
-      "epoch": 3.88,
-      "learning_rate": 7.303475456825511e-05,
-      "loss": 1.8568,
-      "step": 269500
-    },
-    {
-      "epoch": 3.89,
-      "learning_rate": 7.298293167998587e-05,
-      "loss": 1.856,
-      "step": 270000
-    },
-    {
-      "epoch": 3.89,
-      "learning_rate": 7.293100493823316e-05,
-      "loss": 1.8577,
-      "step": 270500
-    },
-    {
-      "epoch": 3.9,
-      "learning_rate": 7.287907819648041e-05,
-      "loss": 1.8551,
-      "step": 271000
-    },
-    {
-      "epoch": 3.91,
-      "learning_rate": 7.282715145472766e-05,
-      "loss": 1.8555,
-      "step": 271500
-    },
-    {
-      "epoch": 3.92,
-      "learning_rate": 7.277522471297495e-05,
-      "loss": 1.8565,
-      "step": 272000
-    },
-    {
-      "epoch": 3.92,
-      "learning_rate": 7.27234018247057e-05,
-      "loss": 1.8574,
-      "step": 272500
-    },
-    {
-      "epoch": 3.93,
-      "learning_rate": 7.267147508295298e-05,
-      "loss": 1.8586,
-      "step": 273000
-    },
-    {
-      "epoch": 3.94,
-      "learning_rate": 7.261954834120024e-05,
-      "loss": 1.8537,
-      "step": 273500
-    },
-    {
-      "epoch": 3.94,
-      "learning_rate": 7.25676215994475e-05,
-      "loss": 1.8567,
-      "step": 274000
-    },
-    {
-      "epoch": 3.95,
-      "learning_rate": 7.251579871117827e-05,
-      "loss": 1.8585,
-      "step": 274500
-    },
-    {
-      "epoch": 3.96,
-      "learning_rate": 7.246387196942554e-05,
-      "loss": 1.854,
-      "step": 275000
-    },
-    {
-      "epoch": 3.97,
-      "learning_rate": 7.241194522767281e-05,
-      "loss": 1.8553,
-      "step": 275500
-    },
-    {
-      "epoch": 3.97,
-      "learning_rate": 7.236001848592006e-05,
-      "loss": 1.8532,
-      "step": 276000
-    },
-    {
-      "epoch": 3.98,
-      "learning_rate": 7.230819559765084e-05,
-      "loss": 1.8559,
-      "step": 276500
-    },
-    {
-      "epoch": 3.99,
-      "learning_rate": 7.22562688558981e-05,
-      "loss": 1.8567,
-      "step": 277000
-    },
-    {
-      "epoch": 3.99,
-      "learning_rate": 7.220434211414536e-05,
-      "loss": 1.8518,
-      "step": 277500
-    },
-    {
-      "epoch": 4.0,
-      "eval_accuracy": 0.6444414608959261,
-      "eval_loss": 1.721920132637024,
-      "eval_runtime": 1267.8764,
-      "eval_samples_per_second": 425.071,
-      "eval_steps_per_second": 26.567,
-      "step": 277892
-    },
-    {
-      "epoch": 4.0,
-      "learning_rate": 7.215241537239263e-05,
-      "loss": 1.8506,
-      "step": 278000
-    },
-    {
-      "epoch": 4.01,
-      "learning_rate": 7.21005924841234e-05,
-      "loss": 1.8528,
-      "step": 278500
-    },
-    {
-      "epoch": 4.02,
-      "learning_rate": 7.204866574237067e-05,
-      "loss": 1.8486,
-      "step": 279000
-    },
-    {
-      "epoch": 4.02,
-      "learning_rate": 7.199673900061794e-05,
-      "loss": 1.8504,
-      "step": 279500
-    },
-    {
-      "epoch": 4.03,
-      "learning_rate": 7.19448122588652e-05,
-      "loss": 1.8496,
-      "step": 280000
-    },
-    {
-      "epoch": 4.04,
-      "learning_rate": 7.189298937059597e-05,
-      "loss": 1.8481,
-      "step": 280500
-    },
-    {
-      "epoch": 4.04,
-      "learning_rate": 7.184106262884324e-05,
-      "loss": 1.8505,
-      "step": 281000
-    },
-    {
-      "epoch": 4.05,
-      "learning_rate": 7.178913588709049e-05,
-      "loss": 1.843,
-      "step": 281500
-    },
-    {
-      "epoch": 4.06,
-      "learning_rate": 7.173720914533776e-05,
-      "loss": 1.852,
-      "step": 282000
-    },
-    {
-      "epoch": 4.07,
-      "learning_rate": 7.168538625706853e-05,
-      "loss": 1.847,
-      "step": 282500
-    },
-    {
-      "epoch": 4.07,
-      "learning_rate": 7.16334595153158e-05,
-      "loss": 1.8492,
-      "step": 283000
-    },
-    {
-      "epoch": 4.08,
-      "learning_rate": 7.158153277356306e-05,
-      "loss": 1.8476,
-      "step": 283500
-    },
-    {
-      "epoch": 4.09,
-      "learning_rate": 7.152960603181032e-05,
-      "loss": 1.8524,
-      "step": 284000
-    },
-    {
-      "epoch": 4.1,
-      "learning_rate": 7.14777831435411e-05,
-      "loss": 1.8478,
-      "step": 284500
-    },
-    {
-      "epoch": 4.1,
-      "learning_rate": 7.142585640178835e-05,
-      "loss": 1.8498,
-      "step": 285000
-    },
-    {
-      "epoch": 4.11,
-      "learning_rate": 7.137392966003562e-05,
-      "loss": 1.8488,
-      "step": 285500
-    },
-    {
-      "epoch": 4.12,
-      "learning_rate": 7.132200291828289e-05,
-      "loss": 1.8522,
-      "step": 286000
-    },
-    {
-      "epoch": 4.12,
-      "learning_rate": 7.127018003001366e-05,
-      "loss": 1.8446,
-      "step": 286500
-    },
-    {
-      "epoch": 4.13,
-      "learning_rate": 7.121825328826092e-05,
-      "loss": 1.8471,
-      "step": 287000
-    },
-    {
-      "epoch": 4.14,
-      "learning_rate": 7.116632654650819e-05,
-      "loss": 1.8453,
-      "step": 287500
-    },
-    {
-      "epoch": 4.15,
-      "learning_rate": 7.111439980475545e-05,
-      "loss": 1.8443,
-      "step": 288000
-    },
-    {
-      "epoch": 4.15,
-      "learning_rate": 7.106257691648623e-05,
-      "loss": 1.8477,
-      "step": 288500
-    },
-    {
-      "epoch": 4.16,
-      "learning_rate": 7.101065017473348e-05,
-      "loss": 1.8482,
-      "step": 289000
-    },
-    {
-      "epoch": 4.17,
-      "learning_rate": 7.095872343298075e-05,
-      "loss": 1.8482,
-      "step": 289500
-    },
-    {
-      "epoch": 4.17,
-      "learning_rate": 7.090679669122802e-05,
-      "loss": 1.8437,
-      "step": 290000
-    },
-    {
-      "epoch": 4.18,
-      "learning_rate": 7.085497380295879e-05,
-      "loss": 1.8444,
-      "step": 290500
-    },
-    {
-      "epoch": 4.19,
-      "learning_rate": 7.080304706120605e-05,
-      "loss": 1.8431,
-      "step": 291000
-    },
-    {
-      "epoch": 4.2,
-      "learning_rate": 7.075112031945332e-05,
-      "loss": 1.8497,
-      "step": 291500
-    },
-    {
-      "epoch": 4.2,
-      "learning_rate": 7.069919357770059e-05,
-      "loss": 1.8444,
-      "step": 292000
-    },
-    {
-      "epoch": 4.21,
-      "learning_rate": 7.064737068943136e-05,
-      "loss": 1.8459,
-      "step": 292500
-    },
-    {
-      "epoch": 4.22,
-      "learning_rate": 7.059544394767862e-05,
-      "loss": 1.8435,
-      "step": 293000
-    },
-    {
-      "epoch": 4.22,
-      "learning_rate": 7.054351720592588e-05,
-      "loss": 1.8439,
-      "step": 293500
-    },
-    {
-      "epoch": 4.23,
-      "learning_rate": 7.049159046417315e-05,
-      "loss": 1.8423,
-      "step": 294000
-    },
-    {
-      "epoch": 4.24,
-      "learning_rate": 7.043976757590391e-05,
-      "loss": 1.8427,
-      "step": 294500
-    },
-    {
-      "epoch": 4.25,
-      "learning_rate": 7.038784083415118e-05,
-      "loss": 1.844,
-      "step": 295000
-    },
-    {
-      "epoch": 4.25,
-      "learning_rate": 7.033591409239845e-05,
-      "loss": 1.842,
-      "step": 295500
-    },
-    {
-      "epoch": 4.26,
-      "learning_rate": 7.02839873506457e-05,
-      "loss": 1.8401,
-      "step": 296000
-    },
-    {
-      "epoch": 4.27,
-      "learning_rate": 7.023216446237648e-05,
-      "loss": 1.8422,
-      "step": 296500
-    },
-    {
-      "epoch": 4.28,
-      "learning_rate": 7.018023772062374e-05,
-      "loss": 1.8425,
-      "step": 297000
-    },
-    {
-      "epoch": 4.28,
-      "learning_rate": 7.012831097887101e-05,
-      "loss": 1.8423,
-      "step": 297500
-    },
-    {
-      "epoch": 4.29,
-      "learning_rate": 7.007638423711828e-05,
-      "loss": 1.8378,
-      "step": 298000
-    },
-    {
-      "epoch": 4.3,
-      "learning_rate": 7.002456134884906e-05,
-      "loss": 1.8424,
-      "step": 298500
-    },
-    {
-      "epoch": 4.3,
-      "learning_rate": 6.997263460709631e-05,
-      "loss": 1.8459,
-      "step": 299000
-    },
-    {
-      "epoch": 4.31,
-      "learning_rate": 6.992070786534358e-05,
-      "loss": 1.8393,
-      "step": 299500
-    },
-    {
-      "epoch": 4.32,
-      "learning_rate": 6.986878112359085e-05,
-      "loss": 1.8382,
-      "step": 300000
-    },
-    {
-      "epoch": 4.33,
-      "learning_rate": 6.981695823532161e-05,
-      "loss": 1.8398,
-      "step": 300500
-    },
-    {
-      "epoch": 4.33,
-      "learning_rate": 6.976503149356888e-05,
-      "loss": 1.8415,
-      "step": 301000
-    },
-    {
-      "epoch": 4.34,
-      "learning_rate": 6.971310475181614e-05,
-      "loss": 1.836,
-      "step": 301500
-    },
-    {
-      "epoch": 4.35,
-      "learning_rate": 6.96611780100634e-05,
-      "loss": 1.8427,
-      "step": 302000
-    },
-    {
-      "epoch": 4.35,
-      "learning_rate": 6.960935512179417e-05,
-      "loss": 1.8376,
-      "step": 302500
-    },
-    {
-      "epoch": 4.36,
-      "learning_rate": 6.955742838004144e-05,
-      "loss": 1.8433,
-      "step": 303000
-    },
-    {
-      "epoch": 4.37,
-      "learning_rate": 6.950550163828871e-05,
-      "loss": 1.8437,
-      "step": 303500
-    },
-    {
-      "epoch": 4.38,
-      "learning_rate": 6.945357489653596e-05,
-      "loss": 1.842,
-      "step": 304000
-    },
-    {
-      "epoch": 4.38,
-      "learning_rate": 6.940175200826674e-05,
-      "loss": 1.8364,
-      "step": 304500
-    },
-    {
-      "epoch": 4.39,
-      "learning_rate": 6.9349825266514e-05,
-      "loss": 1.8412,
-      "step": 305000
-    },
-    {
-      "epoch": 4.4,
-      "learning_rate": 6.929789852476128e-05,
-      "loss": 1.8378,
-      "step": 305500
-    },
-    {
-      "epoch": 4.4,
-      "learning_rate": 6.924597178300853e-05,
-      "loss": 1.8409,
-      "step": 306000
-    },
-    {
-      "epoch": 4.41,
-      "learning_rate": 6.919414889473931e-05,
-      "loss": 1.8416,
-      "step": 306500
-    },
-    {
-      "epoch": 4.42,
-      "learning_rate": 6.914222215298657e-05,
-      "loss": 1.8389,
-      "step": 307000
-    },
-    {
-      "epoch": 4.43,
-      "learning_rate": 6.909029541123384e-05,
-      "loss": 1.8346,
-      "step": 307500
-    },
-    {
-      "epoch": 4.43,
-      "learning_rate": 6.90383686694811e-05,
-      "loss": 1.8392,
-      "step": 308000
-    },
-    {
-      "epoch": 4.44,
-      "learning_rate": 6.898654578121187e-05,
-      "loss": 1.8339,
-      "step": 308500
-    },
-    {
-      "epoch": 4.45,
-      "learning_rate": 6.893461903945914e-05,
-      "loss": 1.8365,
-      "step": 309000
-    },
-    {
-      "epoch": 4.45,
-      "learning_rate": 6.88826922977064e-05,
-      "loss": 1.8375,
-      "step": 309500
-    },
-    {
-      "epoch": 4.46,
-      "learning_rate": 6.883076555595368e-05,
-      "loss": 1.8386,
-      "step": 310000
-    },
-    {
-      "epoch": 4.47,
-      "learning_rate": 6.877894266768443e-05,
-      "loss": 1.8402,
-      "step": 310500
-    },
-    {
-      "epoch": 4.48,
-      "learning_rate": 6.872701592593171e-05,
-      "loss": 1.8372,
-      "step": 311000
-    },
-    {
-      "epoch": 4.48,
-      "learning_rate": 6.867508918417897e-05,
-      "loss": 1.8329,
-      "step": 311500
-    },
-    {
-      "epoch": 4.49,
-      "learning_rate": 6.862316244242622e-05,
-      "loss": 1.836,
-      "step": 312000
-    },
-    {
-      "epoch": 4.5,
-      "learning_rate": 6.8571339554157e-05,
-      "loss": 1.8352,
-      "step": 312500
-    },
-    {
-      "epoch": 4.51,
-      "learning_rate": 6.851941281240425e-05,
-      "loss": 1.8343,
-      "step": 313000
-    },
-    {
-      "epoch": 4.51,
-      "learning_rate": 6.846748607065154e-05,
-      "loss": 1.84,
-      "step": 313500
-    },
-    {
-      "epoch": 4.52,
-      "learning_rate": 6.841555932889879e-05,
-      "loss": 1.8387,
-      "step": 314000
-    },
-    {
-      "epoch": 4.53,
-      "learning_rate": 6.836373644062957e-05,
-      "loss": 1.8325,
-      "step": 314500
-    },
-    {
-      "epoch": 4.53,
-      "learning_rate": 6.831180969887683e-05,
-      "loss": 1.8345,
-      "step": 315000
-    },
-    {
-      "epoch": 4.54,
-      "learning_rate": 6.82598829571241e-05,
-      "loss": 1.835,
-      "step": 315500
-    },
-    {
-      "epoch": 4.55,
-      "learning_rate": 6.820795621537136e-05,
-      "loss": 1.8368,
-      "step": 316000
-    },
-    {
-      "epoch": 4.56,
-      "learning_rate": 6.815613332710213e-05,
-      "loss": 1.8377,
-      "step": 316500
-    },
-    {
-      "epoch": 4.56,
-      "learning_rate": 6.81042065853494e-05,
-      "loss": 1.8386,
-      "step": 317000
-    },
-    {
-      "epoch": 4.57,
-      "learning_rate": 6.805227984359665e-05,
-      "loss": 1.8343,
-      "step": 317500
-    },
-    {
-      "epoch": 4.58,
-      "learning_rate": 6.800035310184392e-05,
-      "loss": 1.8332,
-      "step": 318000
-    },
-    {
-      "epoch": 4.58,
-      "learning_rate": 6.794853021357469e-05,
-      "loss": 1.8349,
-      "step": 318500
-    },
-    {
-      "epoch": 4.59,
-      "learning_rate": 6.789660347182197e-05,
-      "loss": 1.8317,
-      "step": 319000
-    },
-    {
-      "epoch": 4.6,
-      "learning_rate": 6.784467673006922e-05,
-      "loss": 1.8323,
-      "step": 319500
-    },
-    {
-      "epoch": 4.61,
-      "learning_rate": 6.779274998831649e-05,
-      "loss": 1.8338,
-      "step": 320000
-    },
-    {
-      "epoch": 4.61,
-      "learning_rate": 6.774092710004726e-05,
-      "loss": 1.8322,
-      "step": 320500
-    },
-    {
-      "epoch": 4.62,
-      "learning_rate": 6.768900035829453e-05,
-      "loss": 1.8344,
-      "step": 321000
-    },
-    {
-      "epoch": 4.63,
-      "learning_rate": 6.76370736165418e-05,
-      "loss": 1.8345,
-      "step": 321500
-    },
-    {
-      "epoch": 4.63,
-      "learning_rate": 6.758514687478905e-05,
-      "loss": 1.8343,
-      "step": 322000
-    },
-    {
-      "epoch": 4.64,
-      "learning_rate": 6.753332398651983e-05,
-      "loss": 1.8333,
-      "step": 322500
-    },
-    {
-      "epoch": 4.65,
-      "learning_rate": 6.748139724476708e-05,
-      "loss": 1.8374,
-      "step": 323000
-    },
-    {
-      "epoch": 4.66,
-      "learning_rate": 6.742947050301435e-05,
-      "loss": 1.8339,
-      "step": 323500
-    },
-    {
-      "epoch": 4.66,
-      "learning_rate": 6.737754376126162e-05,
-      "loss": 1.8346,
-      "step": 324000
-    },
-    {
-      "epoch": 4.67,
-      "learning_rate": 6.732572087299239e-05,
-      "loss": 1.8318,
-      "step": 324500
-    },
-    {
-      "epoch": 4.68,
-      "learning_rate": 6.727379413123965e-05,
-      "loss": 1.8313,
-      "step": 325000
-    },
-    {
-      "epoch": 4.69,
-      "learning_rate": 6.722186738948691e-05,
-      "loss": 1.8287,
-      "step": 325500
-    },
-    {
-      "epoch": 4.69,
-      "learning_rate": 6.716994064773418e-05,
-      "loss": 1.8282,
-      "step": 326000
-    },
-    {
-      "epoch": 4.7,
-      "learning_rate": 6.711811775946494e-05,
-      "loss": 1.8341,
-      "step": 326500
-    },
-    {
-      "epoch": 4.71,
-      "learning_rate": 6.706619101771221e-05,
-      "loss": 1.8319,
-      "step": 327000
-    },
-    {
-      "epoch": 4.71,
-      "learning_rate": 6.701426427595948e-05,
-      "loss": 1.8299,
-      "step": 327500
-    },
-    {
-      "epoch": 4.72,
-      "learning_rate": 6.696233753420675e-05,
-      "loss": 1.8299,
-      "step": 328000
-    },
-    {
-      "epoch": 4.73,
-      "learning_rate": 6.691051464593751e-05,
-      "loss": 1.8296,
-      "step": 328500
-    },
-    {
-      "epoch": 4.74,
-      "learning_rate": 6.685858790418478e-05,
-      "loss": 1.8306,
-      "step": 329000
-    },
-    {
-      "epoch": 4.74,
-      "learning_rate": 6.680666116243205e-05,
-      "loss": 1.8353,
-      "step": 329500
-    },
-    {
-      "epoch": 4.75,
-      "learning_rate": 6.67547344206793e-05,
-      "loss": 1.8262,
-      "step": 330000
-    },
-    {
-      "epoch": 4.76,
-      "learning_rate": 6.670291153241009e-05,
-      "loss": 1.8285,
-      "step": 330500
-    },
-    {
-      "epoch": 4.76,
-      "learning_rate": 6.665098479065734e-05,
-      "loss": 1.8275,
-      "step": 331000
-    },
-    {
-      "epoch": 4.77,
-      "learning_rate": 6.659905804890461e-05,
-      "loss": 1.8286,
-      "step": 331500
-    },
-    {
-      "epoch": 4.78,
-      "learning_rate": 6.654713130715188e-05,
-      "loss": 1.8289,
-      "step": 332000
-    },
-    {
-      "epoch": 4.79,
-      "learning_rate": 6.649530841888264e-05,
-      "loss": 1.8296,
-      "step": 332500
-    },
-    {
-      "epoch": 4.79,
-      "learning_rate": 6.644338167712991e-05,
-      "loss": 1.8276,
-      "step": 333000
-    },
-    {
-      "epoch": 4.8,
-      "learning_rate": 6.639145493537718e-05,
-      "loss": 1.8309,
-      "step": 333500
-    },
-    {
-      "epoch": 4.81,
-      "learning_rate": 6.633952819362443e-05,
-      "loss": 1.8291,
-      "step": 334000
-    },
-    {
-      "epoch": 4.81,
-      "learning_rate": 6.628770530535521e-05,
-      "loss": 1.8289,
-      "step": 334500
-    },
-    {
-      "epoch": 4.82,
-      "learning_rate": 6.623577856360247e-05,
-      "loss": 1.8281,
-      "step": 335000
-    },
-    {
-      "epoch": 4.83,
-      "learning_rate": 6.618385182184974e-05,
-      "loss": 1.8289,
-      "step": 335500
-    },
-    {
-      "epoch": 4.84,
-      "learning_rate": 6.6131925080097e-05,
-      "loss": 1.8285,
-      "step": 336000
-    },
-    {
-      "epoch": 4.84,
-      "learning_rate": 6.608010219182777e-05,
-      "loss": 1.8256,
-      "step": 336500
-    },
-    {
-      "epoch": 4.85,
-      "learning_rate": 6.602817545007504e-05,
-      "loss": 1.8247,
-      "step": 337000
-    },
-    {
-      "epoch": 4.86,
-      "learning_rate": 6.59762487083223e-05,
-      "loss": 1.8225,
-      "step": 337500
-    },
-    {
-      "epoch": 4.87,
-      "learning_rate": 6.592432196656956e-05,
-      "loss": 1.8277,
-      "step": 338000
-    },
-    {
-      "epoch": 4.87,
-      "learning_rate": 6.587249907830033e-05,
-      "loss": 1.8311,
-      "step": 338500
-    },
-    {
-      "epoch": 4.88,
-      "learning_rate": 6.58205723365476e-05,
-      "loss": 1.827,
-      "step": 339000
-    },
-    {
-      "epoch": 4.89,
-      "learning_rate": 6.576864559479487e-05,
-      "loss": 1.8207,
-      "step": 339500
-    },
-    {
-      "epoch": 4.89,
-      "learning_rate": 6.571671885304213e-05,
-      "loss": 1.8263,
-      "step": 340000
-    },
-    {
-      "epoch": 4.9,
-      "learning_rate": 6.56648959647729e-05,
-      "loss": 1.8259,
-      "step": 340500
-    },
-    {
-      "epoch": 4.91,
-      "learning_rate": 6.561296922302017e-05,
-      "loss": 1.8265,
-      "step": 341000
-    },
-    {
-      "epoch": 4.92,
-      "learning_rate": 6.556104248126744e-05,
-      "loss": 1.8268,
-      "step": 341500
-    },
-    {
-      "epoch": 4.92,
-      "learning_rate": 6.550911573951469e-05,
-      "loss": 1.8251,
-      "step": 342000
-    },
-    {
-      "epoch": 4.93,
-      "learning_rate": 6.545729285124547e-05,
-      "loss": 1.8209,
-      "step": 342500
-    },
-    {
-      "epoch": 4.94,
-      "learning_rate": 6.540536610949273e-05,
-      "loss": 1.8248,
-      "step": 343000
-    },
-    {
-      "epoch": 4.94,
-      "learning_rate": 6.535343936774e-05,
-      "loss": 1.8261,
-      "step": 343500
-    },
-    {
-      "epoch": 4.95,
-      "learning_rate": 6.530151262598726e-05,
-      "loss": 1.8247,
-      "step": 344000
-    },
-    {
-      "epoch": 4.96,
-      "learning_rate": 6.524968973771803e-05,
-      "loss": 1.8293,
-      "step": 344500
-    },
-    {
-      "epoch": 4.97,
-      "learning_rate": 6.51977629959653e-05,
-      "loss": 1.8268,
-      "step": 345000
-    },
-    {
-      "epoch": 4.97,
-      "learning_rate": 6.514583625421255e-05,
-      "loss": 1.8235,
-      "step": 345500
-    },
-    {
-      "epoch": 4.98,
-      "learning_rate": 6.509390951245983e-05,
-      "loss": 1.8214,
-      "step": 346000
-    },
-    {
-      "epoch": 4.99,
-      "learning_rate": 6.504208662419059e-05,
-      "loss": 1.8261,
-      "step": 346500
-    },
-    {
-      "epoch": 4.99,
-      "learning_rate": 6.499015988243787e-05,
-      "loss": 1.8202,
-      "step": 347000
-    },
-    {
-      "epoch": 5.0,
-      "eval_accuracy": 0.6490041136483795,
-      "eval_loss": 1.6904325485229492,
-      "eval_runtime": 1266.0603,
-      "eval_samples_per_second": 425.68,
-      "eval_steps_per_second": 26.605,
-      "step": 347365
-    },
-    {
-      "epoch": 5.0,
-      "learning_rate": 6.493823314068512e-05,
-      "loss": 1.8242,
-      "step": 347500
-    },
-    {
-      "epoch": 5.01,
-      "learning_rate": 6.488630639893238e-05,
-      "loss": 1.817,
-      "step": 348000
-    },
-    {
-      "epoch": 5.02,
-      "learning_rate": 6.483448351066316e-05,
-      "loss": 1.8215,
-      "step": 348500
-    },
-    {
-      "epoch": 5.02,
-      "learning_rate": 6.478255676891043e-05,
-      "loss": 1.8152,
-      "step": 349000
-    },
-    {
-      "epoch": 5.03,
-      "learning_rate": 6.47306300271577e-05,
-      "loss": 1.8193,
-      "step": 349500
-    },
-    {
-      "epoch": 5.04,
-      "learning_rate": 6.467870328540495e-05,
-      "loss": 1.8207,
-      "step": 350000
-    },
-    {
-      "epoch": 5.05,
-      "learning_rate": 6.462688039713573e-05,
-      "loss": 1.8199,
-      "step": 350500
-    },
-    {
-      "epoch": 5.05,
-      "learning_rate": 6.457495365538298e-05,
-      "loss": 1.8181,
-      "step": 351000
-    },
-    {
-      "epoch": 5.06,
-      "learning_rate": 6.452302691363025e-05,
-      "loss": 1.8214,
-      "step": 351500
-    },
-    {
-      "epoch": 5.07,
-      "learning_rate": 6.447110017187752e-05,
-      "loss": 1.8212,
-      "step": 352000
-    },
-    {
-      "epoch": 5.07,
-      "learning_rate": 6.441927728360829e-05,
-      "loss": 1.8196,
-      "step": 352500
-    },
-    {
-      "epoch": 5.08,
-      "learning_rate": 6.436735054185555e-05,
-      "loss": 1.8239,
-      "step": 353000
-    },
-    {
-      "epoch": 5.09,
-      "learning_rate": 6.431542380010281e-05,
-      "loss": 1.82,
-      "step": 353500
-    },
-    {
-      "epoch": 5.1,
-      "learning_rate": 6.426349705835009e-05,
-      "loss": 1.82,
-      "step": 354000
-    },
-    {
-      "epoch": 5.1,
-      "learning_rate": 6.421167417008084e-05,
-      "loss": 1.8192,
-      "step": 354500
-    },
-    {
-      "epoch": 5.11,
-      "learning_rate": 6.415974742832813e-05,
-      "loss": 1.8149,
-      "step": 355000
-    },
-    {
-      "epoch": 5.12,
-      "learning_rate": 6.410782068657538e-05,
-      "loss": 1.8183,
-      "step": 355500
-    },
-    {
-      "epoch": 5.12,
-      "learning_rate": 6.405589394482265e-05,
-      "loss": 1.8165,
-      "step": 356000
-    },
-    {
-      "epoch": 5.13,
-      "learning_rate": 6.400407105655342e-05,
-      "loss": 1.816,
-      "step": 356500
-    },
-    {
-      "epoch": 5.14,
-      "learning_rate": 6.395214431480068e-05,
-      "loss": 1.8177,
-      "step": 357000
-    },
-    {
-      "epoch": 5.15,
-      "learning_rate": 6.390021757304795e-05,
-      "loss": 1.817,
-      "step": 357500
-    },
-    {
-      "epoch": 5.15,
-      "learning_rate": 6.38482908312952e-05,
-      "loss": 1.8175,
-      "step": 358000
-    },
-    {
-      "epoch": 5.16,
-      "learning_rate": 6.379646794302599e-05,
-      "loss": 1.8133,
-      "step": 358500
-    },
-    {
-      "epoch": 5.17,
-      "learning_rate": 6.374454120127324e-05,
-      "loss": 1.8163,
-      "step": 359000
-    },
-    {
-      "epoch": 5.17,
-      "learning_rate": 6.369261445952052e-05,
-      "loss": 1.8156,
-      "step": 359500
-    },
-    {
-      "epoch": 5.18,
-      "learning_rate": 6.364068771776778e-05,
-      "loss": 1.8158,
-      "step": 360000
-    },
-    {
-      "epoch": 5.19,
-      "learning_rate": 6.358886482949856e-05,
-      "loss": 1.8141,
-      "step": 360500
-    },
-    {
-      "epoch": 5.2,
-      "learning_rate": 6.353693808774581e-05,
-      "loss": 1.8159,
-      "step": 361000
-    },
-    {
-      "epoch": 5.2,
-      "learning_rate": 6.348501134599307e-05,
-      "loss": 1.82,
-      "step": 361500
-    },
-    {
-      "epoch": 5.21,
-      "learning_rate": 6.343308460424035e-05,
-      "loss": 1.8148,
-      "step": 362000
-    },
-    {
-      "epoch": 5.22,
-      "learning_rate": 6.33812617159711e-05,
-      "loss": 1.8171,
-      "step": 362500
-    },
-    {
-      "epoch": 5.23,
-      "learning_rate": 6.332933497421838e-05,
-      "loss": 1.8167,
-      "step": 363000
-    },
-    {
-      "epoch": 5.23,
-      "learning_rate": 6.327740823246564e-05,
-      "loss": 1.8164,
-      "step": 363500
-    },
-    {
-      "epoch": 5.24,
-      "learning_rate": 6.32254814907129e-05,
-      "loss": 1.8144,
-      "step": 364000
-    },
-    {
-      "epoch": 5.25,
-      "learning_rate": 6.317365860244367e-05,
-      "loss": 1.8186,
-      "step": 364500
-    },
-    {
-      "epoch": 5.25,
-      "learning_rate": 6.312173186069094e-05,
-      "loss": 1.8184,
-      "step": 365000
-    },
-    {
-      "epoch": 5.26,
-      "learning_rate": 6.306980511893821e-05,
-      "loss": 1.8155,
-      "step": 365500
-    },
-    {
-      "epoch": 5.27,
-      "learning_rate": 6.301787837718546e-05,
-      "loss": 1.8184,
-      "step": 366000
-    },
-    {
-      "epoch": 5.28,
-      "learning_rate": 6.296605548891624e-05,
-      "loss": 1.8176,
-      "step": 366500
-    },
-    {
-      "epoch": 5.28,
-      "learning_rate": 6.29141287471635e-05,
-      "loss": 1.8143,
-      "step": 367000
-    },
-    {
-      "epoch": 5.29,
-      "learning_rate": 6.286220200541077e-05,
-      "loss": 1.8174,
-      "step": 367500
-    },
-    {
-      "epoch": 5.3,
-      "learning_rate": 6.281027526365803e-05,
-      "loss": 1.8124,
-      "step": 368000
-    },
-    {
-      "epoch": 5.3,
-      "learning_rate": 6.275845237538881e-05,
-      "loss": 1.8168,
-      "step": 368500
-    },
-    {
-      "epoch": 5.31,
-      "learning_rate": 6.270652563363607e-05,
-      "loss": 1.8124,
-      "step": 369000
-    },
-    {
-      "epoch": 5.32,
-      "learning_rate": 6.265459889188334e-05,
-      "loss": 1.8129,
-      "step": 369500
-    },
-    {
-      "epoch": 5.33,
-      "learning_rate": 6.26026721501306e-05,
-      "loss": 1.815,
-      "step": 370000
-    },
-    {
-      "epoch": 5.33,
-      "learning_rate": 6.255084926186137e-05,
-      "loss": 1.8112,
-      "step": 370500
-    },
-    {
-      "epoch": 5.34,
-      "learning_rate": 6.249892252010864e-05,
-      "loss": 1.8129,
-      "step": 371000
-    },
-    {
-      "epoch": 5.35,
-      "learning_rate": 6.24469957783559e-05,
-      "loss": 1.8123,
-      "step": 371500
-    },
-    {
-      "epoch": 5.35,
-      "learning_rate": 6.239506903660316e-05,
-      "loss": 1.8152,
-      "step": 372000
-    },
-    {
-      "epoch": 5.36,
-      "learning_rate": 6.234324614833393e-05,
-      "loss": 1.8154,
-      "step": 372500
-    },
-    {
-      "epoch": 5.37,
-      "learning_rate": 6.22913194065812e-05,
-      "loss": 1.8118,
-      "step": 373000
-    },
-    {
-      "epoch": 5.38,
-      "learning_rate": 6.223939266482847e-05,
-      "loss": 1.8102,
-      "step": 373500
-    },
-    {
-      "epoch": 5.38,
-      "learning_rate": 6.218746592307572e-05,
-      "loss": 1.8123,
-      "step": 374000
-    },
-    {
-      "epoch": 5.39,
-      "learning_rate": 6.21356430348065e-05,
-      "loss": 1.8139,
-      "step": 374500
-    },
-    {
-      "epoch": 5.4,
-      "learning_rate": 6.208371629305376e-05,
-      "loss": 1.8072,
-      "step": 375000
-    },
-    {
-      "epoch": 5.4,
-      "learning_rate": 6.203178955130102e-05,
-      "loss": 1.8103,
-      "step": 375500
-    },
-    {
-      "epoch": 5.41,
-      "learning_rate": 6.197986280954829e-05,
-      "loss": 1.8124,
-      "step": 376000
-    },
-    {
-      "epoch": 5.42,
-      "learning_rate": 6.192803992127906e-05,
-      "loss": 1.8093,
-      "step": 376500
-    },
-    {
-      "epoch": 5.43,
-      "learning_rate": 6.187611317952633e-05,
-      "loss": 1.8109,
-      "step": 377000
-    },
-    {
-      "epoch": 5.43,
-      "learning_rate": 6.18241864377736e-05,
-      "loss": 1.8066,
-      "step": 377500
-    },
-    {
-      "epoch": 5.44,
-      "learning_rate": 6.177225969602086e-05,
-      "loss": 1.8095,
-      "step": 378000
-    },
-    {
-      "epoch": 5.45,
-      "learning_rate": 6.172043680775163e-05,
-      "loss": 1.8108,
-      "step": 378500
-    },
-    {
-      "epoch": 5.46,
-      "learning_rate": 6.16685100659989e-05,
-      "loss": 1.8083,
-      "step": 379000
-    },
-    {
-      "epoch": 5.46,
-      "learning_rate": 6.161658332424615e-05,
-      "loss": 1.8126,
-      "step": 379500
-    },
-    {
-      "epoch": 5.47,
-      "learning_rate": 6.156465658249342e-05,
-      "loss": 1.8116,
-      "step": 380000
-    },
-    {
-      "epoch": 5.48,
-      "learning_rate": 6.151283369422419e-05,
-      "loss": 1.8093,
-      "step": 380500
-    },
-    {
-      "epoch": 5.48,
-      "learning_rate": 6.146090695247146e-05,
-      "loss": 1.8125,
-      "step": 381000
-    },
-    {
-      "epoch": 5.49,
-      "learning_rate": 6.140898021071872e-05,
-      "loss": 1.8105,
-      "step": 381500
-    },
-    {
-      "epoch": 5.5,
-      "learning_rate": 6.135705346896599e-05,
-      "loss": 1.8114,
-      "step": 382000
-    },
-    {
-      "epoch": 5.51,
-      "learning_rate": 6.130523058069676e-05,
-      "loss": 1.8111,
-      "step": 382500
-    },
-    {
-      "epoch": 5.51,
-      "learning_rate": 6.125330383894403e-05,
-      "loss": 1.8092,
-      "step": 383000
-    },
-    {
-      "epoch": 5.52,
-      "learning_rate": 6.120137709719128e-05,
-      "loss": 1.8081,
-      "step": 383500
-    },
-    {
-      "epoch": 5.53,
-      "learning_rate": 6.114945035543855e-05,
-      "loss": 1.8051,
-      "step": 384000
-    },
-    {
-      "epoch": 5.53,
-      "learning_rate": 6.109762746716932e-05,
-      "loss": 1.8085,
-      "step": 384500
-    },
-    {
-      "epoch": 5.54,
-      "learning_rate": 6.104570072541658e-05,
-      "loss": 1.8095,
-      "step": 385000
-    },
-    {
-      "epoch": 5.55,
-      "learning_rate": 6.0993773983663846e-05,
-      "loss": 1.8033,
-      "step": 385500
-    },
-    {
-      "epoch": 5.56,
-      "learning_rate": 6.094184724191111e-05,
-      "loss": 1.8116,
-      "step": 386000
-    },
-    {
-      "epoch": 5.56,
-      "learning_rate": 6.089002435364188e-05,
-      "loss": 1.8123,
-      "step": 386500
-    },
-    {
-      "epoch": 5.57,
-      "learning_rate": 6.083809761188914e-05,
-      "loss": 1.806,
-      "step": 387000
-    },
-    {
-      "epoch": 5.58,
-      "learning_rate": 6.078617087013642e-05,
-      "loss": 1.8117,
-      "step": 387500
-    },
-    {
-      "epoch": 5.58,
-      "learning_rate": 6.073424412838368e-05,
-      "loss": 1.8058,
-      "step": 388000
-    },
-    {
-      "epoch": 5.59,
-      "learning_rate": 6.068242124011445e-05,
-      "loss": 1.8067,
-      "step": 388500
-    },
-    {
-      "epoch": 5.6,
-      "learning_rate": 6.063049449836171e-05,
-      "loss": 1.8075,
-      "step": 389000
-    },
-    {
-      "epoch": 5.61,
-      "learning_rate": 6.057856775660898e-05,
-      "loss": 1.8102,
-      "step": 389500
-    },
-    {
-      "epoch": 5.61,
-      "learning_rate": 6.052664101485624e-05,
-      "loss": 1.8078,
-      "step": 390000
-    },
-    {
-      "epoch": 5.62,
-      "learning_rate": 6.0474714273103504e-05,
-      "loss": 1.8082,
-      "step": 390500
-    },
-    {
-      "epoch": 5.63,
-      "learning_rate": 6.042289138483428e-05,
-      "loss": 1.807,
-      "step": 391000
-    },
-    {
-      "epoch": 5.64,
-      "learning_rate": 6.037096464308154e-05,
-      "loss": 1.8013,
-      "step": 391500
-    },
-    {
-      "epoch": 5.64,
-      "learning_rate": 6.0319037901328814e-05,
-      "loss": 1.8072,
-      "step": 392000
-    },
-    {
-      "epoch": 5.65,
-      "learning_rate": 6.0267111159576075e-05,
-      "loss": 1.8062,
-      "step": 392500
-    },
-    {
-      "epoch": 5.66,
-      "learning_rate": 6.021528827130685e-05,
-      "loss": 1.8062,
-      "step": 393000
-    },
-    {
-      "epoch": 5.66,
-      "learning_rate": 6.016336152955411e-05,
-      "loss": 1.8034,
-      "step": 393500
-    },
-    {
-      "epoch": 5.67,
-      "learning_rate": 6.011143478780137e-05,
-      "loss": 1.808,
-      "step": 394000
-    },
-    {
-      "epoch": 5.68,
-      "learning_rate": 6.005950804604864e-05,
-      "loss": 1.8079,
-      "step": 394500
-    },
-    {
-      "epoch": 5.69,
-      "learning_rate": 6.0007685157779406e-05,
-      "loss": 1.8035,
-      "step": 395000
-    },
-    {
-      "epoch": 5.69,
-      "learning_rate": 5.9955758416026674e-05,
-      "loss": 1.8072,
-      "step": 395500
-    },
-    {
-      "epoch": 5.7,
-      "learning_rate": 5.9903831674273936e-05,
-      "loss": 1.8055,
-      "step": 396000
-    },
-    {
-      "epoch": 5.71,
-      "learning_rate": 5.98519049325212e-05,
-      "loss": 1.8083,
-      "step": 396500
-    },
-    {
-      "epoch": 5.71,
-      "learning_rate": 5.980008204425197e-05,
-      "loss": 1.8048,
-      "step": 397000
-    },
-    {
-      "epoch": 5.72,
-      "learning_rate": 5.9748155302499245e-05,
-      "loss": 1.8053,
-      "step": 397500
-    },
-    {
-      "epoch": 5.73,
-      "learning_rate": 5.96962285607465e-05,
-      "loss": 1.8088,
-      "step": 398000
-    },
-    {
-      "epoch": 5.74,
-      "learning_rate": 5.964430181899376e-05,
-      "loss": 1.8074,
-      "step": 398500
-    },
-    {
-      "epoch": 5.74,
-      "learning_rate": 5.9592478930724535e-05,
-      "loss": 1.8055,
-      "step": 399000
-    },
-    {
-      "epoch": 5.75,
-      "learning_rate": 5.9540552188971796e-05,
-      "loss": 1.8081,
-      "step": 399500
-    },
-    {
-      "epoch": 5.76,
-      "learning_rate": 5.948862544721907e-05,
-      "loss": 1.8091,
-      "step": 400000
-    },
-    {
-      "epoch": 5.76,
-      "learning_rate": 5.943669870546633e-05,
-      "loss": 1.8033,
-      "step": 400500
-    },
-    {
-      "epoch": 5.77,
-      "learning_rate": 5.9384875817197106e-05,
-      "loss": 1.8031,
-      "step": 401000
-    },
-    {
-      "epoch": 5.78,
-      "learning_rate": 5.933294907544437e-05,
-      "loss": 1.8068,
-      "step": 401500
-    },
-    {
-      "epoch": 5.79,
-      "learning_rate": 5.928102233369163e-05,
-      "loss": 1.8025,
-      "step": 402000
-    },
-    {
-      "epoch": 5.79,
-      "learning_rate": 5.92290955919389e-05,
-      "loss": 1.8037,
-      "step": 402500
-    },
-    {
-      "epoch": 5.8,
-      "learning_rate": 5.917727270366966e-05,
-      "loss": 1.8097,
-      "step": 403000
-    },
-    {
-      "epoch": 5.81,
-      "learning_rate": 5.912534596191693e-05,
-      "loss": 1.8004,
-      "step": 403500
-    },
-    {
-      "epoch": 5.82,
-      "learning_rate": 5.907341922016419e-05,
-      "loss": 1.8012,
-      "step": 404000
-    },
-    {
-      "epoch": 5.82,
-      "learning_rate": 5.9021492478411455e-05,
-      "loss": 1.8084,
-      "step": 404500
-    },
-    {
-      "epoch": 5.83,
-      "learning_rate": 5.896966959014223e-05,
-      "loss": 1.8011,
-      "step": 405000
-    },
-    {
-      "epoch": 5.84,
-      "learning_rate": 5.891774284838949e-05,
-      "loss": 1.8042,
-      "step": 405500
-    },
-    {
-      "epoch": 5.84,
-      "learning_rate": 5.8865816106636764e-05,
-      "loss": 1.8036,
-      "step": 406000
-    },
-    {
-      "epoch": 5.85,
-      "learning_rate": 5.8813889364884026e-05,
-      "loss": 1.8038,
-      "step": 406500
-    },
-    {
-      "epoch": 5.86,
-      "learning_rate": 5.87620664766148e-05,
-      "loss": 1.8015,
-      "step": 407000
-    },
-    {
-      "epoch": 5.87,
-      "learning_rate": 5.871013973486206e-05,
-      "loss": 1.8043,
-      "step": 407500
-    },
-    {
-      "epoch": 5.87,
-      "learning_rate": 5.865821299310933e-05,
-      "loss": 1.8004,
-      "step": 408000
-    },
-    {
-      "epoch": 5.88,
-      "learning_rate": 5.860628625135659e-05,
-      "loss": 1.8037,
-      "step": 408500
-    },
-    {
-      "epoch": 5.89,
-      "learning_rate": 5.855446336308736e-05,
-      "loss": 1.8021,
-      "step": 409000
-    },
-    {
-      "epoch": 5.89,
-      "learning_rate": 5.8502536621334625e-05,
-      "loss": 1.8039,
-      "step": 409500
-    },
-    {
-      "epoch": 5.9,
-      "learning_rate": 5.8450609879581886e-05,
-      "loss": 1.7982,
-      "step": 410000
-    },
-    {
-      "epoch": 5.91,
-      "learning_rate": 5.8398683137829154e-05,
-      "loss": 1.8009,
-      "step": 410500
-    },
-    {
-      "epoch": 5.92,
-      "learning_rate": 5.834686024955992e-05,
-      "loss": 1.7999,
-      "step": 411000
-    },
-    {
-      "epoch": 5.92,
-      "learning_rate": 5.8294933507807196e-05,
-      "loss": 1.8029,
-      "step": 411500
-    },
-    {
-      "epoch": 5.93,
-      "learning_rate": 5.824300676605445e-05,
-      "loss": 1.7964,
-      "step": 412000
-    },
-    {
-      "epoch": 5.94,
-      "learning_rate": 5.819108002430171e-05,
-      "loss": 1.801,
-      "step": 412500
-    },
-    {
-      "epoch": 5.94,
-      "learning_rate": 5.8139257136032485e-05,
-      "loss": 1.8013,
-      "step": 413000
-    },
-    {
-      "epoch": 5.95,
-      "learning_rate": 5.8087330394279746e-05,
-      "loss": 1.8027,
-      "step": 413500
-    },
-    {
-      "epoch": 5.96,
-      "learning_rate": 5.803540365252702e-05,
-      "loss": 1.7981,
-      "step": 414000
-    },
-    {
-      "epoch": 5.97,
-      "learning_rate": 5.798347691077428e-05,
-      "loss": 1.7972,
-      "step": 414500
-    },
-    {
-      "epoch": 5.97,
-      "learning_rate": 5.7931654022505056e-05,
-      "loss": 1.8025,
-      "step": 415000
-    },
-    {
-      "epoch": 5.98,
-      "learning_rate": 5.787972728075232e-05,
-      "loss": 1.8021,
-      "step": 415500
-    },
-    {
-      "epoch": 5.99,
-      "learning_rate": 5.782780053899958e-05,
-      "loss": 1.7997,
-      "step": 416000
-    },
-    {
-      "epoch": 6.0,
-      "learning_rate": 5.777587379724685e-05,
-      "loss": 1.7996,
-      "step": 416500
-    },
-    {
-      "epoch": 6.0,
-      "eval_accuracy": 0.6524074511202295,
-      "eval_loss": 1.670515775680542,
-      "eval_runtime": 1271.0271,
-      "eval_samples_per_second": 424.017,
-      "eval_steps_per_second": 26.501,
-      "step": 416838
-    },
-    {
-      "epoch": 6.0,
-      "learning_rate": 5.772405090897762e-05,
-      "loss": 1.7959,
-      "step": 417000
-    },
-    {
-      "epoch": 6.01,
-      "learning_rate": 5.767212416722488e-05,
-      "loss": 1.7976,
-      "step": 417500
-    },
-    {
-      "epoch": 6.02,
-      "learning_rate": 5.7620197425472143e-05,
-      "loss": 1.7948,
-      "step": 418000
-    },
-    {
-      "epoch": 6.02,
-      "learning_rate": 5.756827068371942e-05,
-      "loss": 1.7999,
-      "step": 418500
-    },
-    {
-      "epoch": 6.03,
-      "learning_rate": 5.751644779545018e-05,
-      "loss": 1.7928,
-      "step": 419000
-    },
-    {
-      "epoch": 6.04,
-      "learning_rate": 5.746452105369745e-05,
-      "loss": 1.7988,
-      "step": 419500
-    },
-    {
-      "epoch": 6.05,
-      "learning_rate": 5.7412594311944715e-05,
-      "loss": 1.7969,
-      "step": 420000
-    },
-    {
-      "epoch": 6.05,
-      "learning_rate": 5.7360667570191976e-05,
-      "loss": 1.797,
-      "step": 420500
-    },
-    {
-      "epoch": 6.06,
-      "learning_rate": 5.730884468192275e-05,
-      "loss": 1.7943,
-      "step": 421000
-    },
-    {
-      "epoch": 6.07,
-      "learning_rate": 5.725691794017001e-05,
-      "loss": 1.7949,
-      "step": 421500
-    },
-    {
-      "epoch": 6.07,
-      "learning_rate": 5.720499119841728e-05,
-      "loss": 1.8002,
-      "step": 422000
-    },
-    {
-      "epoch": 6.08,
-      "learning_rate": 5.715306445666454e-05,
-      "loss": 1.7974,
-      "step": 422500
-    },
-    {
-      "epoch": 6.09,
-      "learning_rate": 5.7101241568395313e-05,
-      "loss": 1.7974,
-      "step": 423000
-    },
-    {
-      "epoch": 6.1,
-      "learning_rate": 5.7049314826642575e-05,
-      "loss": 1.7962,
-      "step": 423500
-    },
-    {
-      "epoch": 6.1,
-      "learning_rate": 5.6997388084889836e-05,
-      "loss": 1.7945,
-      "step": 424000
-    },
-    {
-      "epoch": 6.11,
-      "learning_rate": 5.6945461343137105e-05,
-      "loss": 1.7977,
-      "step": 424500
-    },
-    {
-      "epoch": 6.12,
-      "learning_rate": 5.689363845486787e-05,
-      "loss": 1.7925,
-      "step": 425000
-    },
-    {
-      "epoch": 6.12,
-      "learning_rate": 5.684171171311514e-05,
-      "loss": 1.7992,
-      "step": 425500
-    },
-    {
-      "epoch": 6.13,
-      "learning_rate": 5.67897849713624e-05,
-      "loss": 1.7964,
-      "step": 426000
-    },
-    {
-      "epoch": 6.14,
-      "learning_rate": 5.673785822960966e-05,
-      "loss": 1.7952,
-      "step": 426500
-    },
-    {
-      "epoch": 6.15,
-      "learning_rate": 5.6686035341340435e-05,
-      "loss": 1.7971,
-      "step": 427000
-    },
-    {
-      "epoch": 6.15,
-      "learning_rate": 5.663410859958771e-05,
-      "loss": 1.7936,
-      "step": 427500
-    },
-    {
-      "epoch": 6.16,
-      "learning_rate": 5.658218185783497e-05,
-      "loss": 1.7953,
-      "step": 428000
-    },
-    {
-      "epoch": 6.17,
-      "learning_rate": 5.6530255116082233e-05,
-      "loss": 1.7987,
-      "step": 428500
-    },
-    {
-      "epoch": 6.18,
-      "learning_rate": 5.6478432227813006e-05,
-      "loss": 1.7965,
-      "step": 429000
-    },
-    {
-      "epoch": 6.18,
-      "learning_rate": 5.642650548606027e-05,
-      "loss": 1.7943,
-      "step": 429500
-    },
-    {
-      "epoch": 6.19,
-      "learning_rate": 5.6374578744307536e-05,
-      "loss": 1.7949,
-      "step": 430000
-    },
-    {
-      "epoch": 6.2,
-      "learning_rate": 5.63226520025548e-05,
-      "loss": 1.7938,
-      "step": 430500
-    },
-    {
-      "epoch": 6.2,
-      "learning_rate": 5.627082911428557e-05,
-      "loss": 1.7929,
-      "step": 431000
-    },
-    {
-      "epoch": 6.21,
-      "learning_rate": 5.621890237253283e-05,
-      "loss": 1.7953,
-      "step": 431500
-    },
-    {
-      "epoch": 6.22,
-      "learning_rate": 5.6166975630780094e-05,
-      "loss": 1.796,
-      "step": 432000
-    },
-    {
-      "epoch": 6.23,
-      "learning_rate": 5.611504888902737e-05,
-      "loss": 1.7938,
-      "step": 432500
-    },
-    {
-      "epoch": 6.23,
-      "learning_rate": 5.606322600075813e-05,
-      "loss": 1.7925,
-      "step": 433000
-    },
-    {
-      "epoch": 6.24,
-      "learning_rate": 5.6011299259005403e-05,
-      "loss": 1.7946,
-      "step": 433500
-    },
-    {
-      "epoch": 6.25,
-      "learning_rate": 5.5959372517252665e-05,
-      "loss": 1.7945,
-      "step": 434000
-    },
-    {
-      "epoch": 6.25,
-      "learning_rate": 5.590744577549992e-05,
-      "loss": 1.7896,
-      "step": 434500
-    },
-    {
-      "epoch": 6.26,
-      "learning_rate": 5.58556228872307e-05,
-      "loss": 1.7962,
-      "step": 435000
-    },
-    {
-      "epoch": 6.27,
-      "learning_rate": 5.5803696145477954e-05,
-      "loss": 1.7921,
-      "step": 435500
-    },
-    {
-      "epoch": 6.28,
-      "learning_rate": 5.5751873257208734e-05,
-      "loss": 1.7922,
-      "step": 436000
-    },
-    {
-      "epoch": 6.28,
-      "learning_rate": 5.569994651545599e-05,
-      "loss": 1.7935,
-      "step": 436500
-    },
-    {
-      "epoch": 6.29,
-      "learning_rate": 5.5648019773703264e-05,
-      "loss": 1.7973,
-      "step": 437000
-    },
-    {
-      "epoch": 6.3,
-      "learning_rate": 5.5596093031950525e-05,
-      "loss": 1.795,
-      "step": 437500
-    },
-    {
-      "epoch": 6.3,
-      "learning_rate": 5.5544166290197794e-05,
-      "loss": 1.7883,
-      "step": 438000
-    },
-    {
-      "epoch": 6.31,
-      "learning_rate": 5.5492239548445055e-05,
-      "loss": 1.7932,
-      "step": 438500
-    },
-    {
-      "epoch": 6.32,
-      "learning_rate": 5.544031280669232e-05,
-      "loss": 1.7926,
-      "step": 439000
-    },
-    {
-      "epoch": 6.33,
-      "learning_rate": 5.538838606493959e-05,
-      "loss": 1.7896,
-      "step": 439500
-    },
-    {
-      "epoch": 6.33,
-      "learning_rate": 5.533656317667035e-05,
-      "loss": 1.7884,
-      "step": 440000
-    },
-    {
-      "epoch": 6.34,
-      "learning_rate": 5.5284636434917626e-05,
-      "loss": 1.7907,
-      "step": 440500
-    },
-    {
-      "epoch": 6.35,
-      "learning_rate": 5.523270969316489e-05,
-      "loss": 1.7919,
-      "step": 441000
-    },
-    {
-      "epoch": 6.35,
-      "learning_rate": 5.518078295141215e-05,
-      "loss": 1.7958,
-      "step": 441500
-    },
-    {
-      "epoch": 6.36,
-      "learning_rate": 5.512896006314292e-05,
-      "loss": 1.7903,
-      "step": 442000
-    },
-    {
-      "epoch": 6.37,
-      "learning_rate": 5.5077033321390184e-05,
-      "loss": 1.7971,
-      "step": 442500
-    },
-    {
-      "epoch": 6.38,
-      "learning_rate": 5.502510657963745e-05,
-      "loss": 1.7891,
-      "step": 443000
-    },
-    {
-      "epoch": 6.38,
-      "learning_rate": 5.4973179837884714e-05,
-      "loss": 1.791,
-      "step": 443500
-    },
-    {
-      "epoch": 6.39,
-      "learning_rate": 5.492135694961549e-05,
-      "loss": 1.7917,
-      "step": 444000
-    },
-    {
-      "epoch": 6.4,
-      "learning_rate": 5.486943020786275e-05,
-      "loss": 1.7915,
-      "step": 444500
-    },
-    {
-      "epoch": 6.41,
-      "learning_rate": 5.481750346611001e-05,
-      "loss": 1.7892,
-      "step": 445000
-    },
-    {
-      "epoch": 6.41,
-      "learning_rate": 5.4765576724357285e-05,
-      "loss": 1.7902,
-      "step": 445500
-    },
-    {
-      "epoch": 6.42,
-      "learning_rate": 5.4713753836088044e-05,
-      "loss": 1.7873,
-      "step": 446000
-    },
-    {
-      "epoch": 6.43,
-      "learning_rate": 5.466182709433532e-05,
-      "loss": 1.7919,
-      "step": 446500
-    },
-    {
-      "epoch": 6.43,
-      "learning_rate": 5.4609900352582574e-05,
-      "loss": 1.7903,
-      "step": 447000
-    },
-    {
-      "epoch": 6.44,
-      "learning_rate": 5.455797361082985e-05,
-      "loss": 1.7929,
-      "step": 447500
-    },
-    {
-      "epoch": 6.45,
-      "learning_rate": 5.450615072256061e-05,
-      "loss": 1.7872,
-      "step": 448000
-    },
-    {
-      "epoch": 6.46,
-      "learning_rate": 5.4454223980807884e-05,
-      "loss": 1.7884,
-      "step": 448500
-    },
-    {
-      "epoch": 6.46,
-      "learning_rate": 5.4402297239055145e-05,
-      "loss": 1.791,
-      "step": 449000
-    },
-    {
-      "epoch": 6.47,
-      "learning_rate": 5.435037049730241e-05,
-      "loss": 1.7905,
-      "step": 449500
-    },
-    {
-      "epoch": 6.48,
-      "learning_rate": 5.429854760903318e-05,
-      "loss": 1.7938,
-      "step": 450000
-    },
-    {
-      "epoch": 6.48,
-      "learning_rate": 5.424662086728044e-05,
-      "loss": 1.7886,
-      "step": 450500
-    },
-    {
-      "epoch": 6.49,
-      "learning_rate": 5.419469412552771e-05,
-      "loss": 1.7929,
-      "step": 451000
-    },
-    {
-      "epoch": 6.5,
-      "learning_rate": 5.414276738377497e-05,
-      "loss": 1.7898,
-      "step": 451500
-    },
-    {
-      "epoch": 6.51,
-      "learning_rate": 5.4090944495505744e-05,
-      "loss": 1.7883,
-      "step": 452000
-    },
-    {
-      "epoch": 6.51,
-      "learning_rate": 5.4039017753753005e-05,
-      "loss": 1.7875,
-      "step": 452500
-    },
-    {
-      "epoch": 6.52,
-      "learning_rate": 5.398709101200027e-05,
-      "loss": 1.7869,
-      "step": 453000
-    },
-    {
-      "epoch": 6.53,
-      "learning_rate": 5.393516427024754e-05,
-      "loss": 1.7875,
-      "step": 453500
-    },
-    {
-      "epoch": 6.53,
-      "learning_rate": 5.38833413819783e-05,
-      "loss": 1.7891,
-      "step": 454000
-    },
-    {
-      "epoch": 6.54,
-      "learning_rate": 5.383141464022558e-05,
-      "loss": 1.7887,
-      "step": 454500
-    },
-    {
-      "epoch": 6.55,
-      "learning_rate": 5.377948789847284e-05,
-      "loss": 1.7904,
-      "step": 455000
-    },
-    {
-      "epoch": 6.56,
-      "learning_rate": 5.37275611567201e-05,
-      "loss": 1.7871,
-      "step": 455500
-    },
-    {
-      "epoch": 6.56,
-      "learning_rate": 5.367573826845087e-05,
-      "loss": 1.7887,
-      "step": 456000
-    },
-    {
-      "epoch": 6.57,
-      "learning_rate": 5.362381152669814e-05,
-      "loss": 1.7875,
-      "step": 456500
-    },
-    {
-      "epoch": 6.58,
-      "learning_rate": 5.35718847849454e-05,
-      "loss": 1.7883,
-      "step": 457000
-    },
-    {
-      "epoch": 6.59,
-      "learning_rate": 5.3519958043192664e-05,
-      "loss": 1.7863,
-      "step": 457500
-    },
-    {
-      "epoch": 6.59,
-      "learning_rate": 5.346813515492344e-05,
-      "loss": 1.7863,
-      "step": 458000
-    },
-    {
-      "epoch": 6.6,
-      "learning_rate": 5.34162084131707e-05,
-      "loss": 1.7886,
-      "step": 458500
-    },
-    {
-      "epoch": 6.61,
-      "learning_rate": 5.3364281671417974e-05,
-      "loss": 1.785,
-      "step": 459000
-    },
-    {
-      "epoch": 6.61,
-      "learning_rate": 5.331235492966523e-05,
-      "loss": 1.7838,
-      "step": 459500
-    },
-    {
-      "epoch": 6.62,
-      "learning_rate": 5.326053204139601e-05,
-      "loss": 1.7836,
-      "step": 460000
-    },
-    {
-      "epoch": 6.63,
-      "learning_rate": 5.320860529964326e-05,
-      "loss": 1.7848,
-      "step": 460500
-    },
-    {
-      "epoch": 6.64,
-      "learning_rate": 5.3156678557890524e-05,
-      "loss": 1.7871,
-      "step": 461000
-    },
-    {
-      "epoch": 6.64,
-      "learning_rate": 5.31047518161378e-05,
-      "loss": 1.7878,
-      "step": 461500
-    },
-    {
-      "epoch": 6.65,
-      "learning_rate": 5.305292892786856e-05,
-      "loss": 1.7856,
-      "step": 462000
-    },
-    {
-      "epoch": 6.66,
-      "learning_rate": 5.3001002186115834e-05,
-      "loss": 1.788,
-      "step": 462500
-    },
-    {
-      "epoch": 6.66,
-      "learning_rate": 5.2949075444363095e-05,
-      "loss": 1.7862,
-      "step": 463000
-    },
-    {
-      "epoch": 6.67,
-      "learning_rate": 5.289714870261036e-05,
-      "loss": 1.7813,
-      "step": 463500
-    },
-    {
-      "epoch": 6.68,
-      "learning_rate": 5.284532581434113e-05,
-      "loss": 1.7879,
-      "step": 464000
-    },
-    {
-      "epoch": 6.69,
-      "learning_rate": 5.279339907258839e-05,
-      "loss": 1.7848,
-      "step": 464500
-    },
-    {
-      "epoch": 6.69,
-      "learning_rate": 5.274147233083566e-05,
-      "loss": 1.7879,
-      "step": 465000
-    },
-    {
-      "epoch": 6.7,
-      "learning_rate": 5.268954558908292e-05,
-      "loss": 1.7831,
-      "step": 465500
-    },
-    {
-      "epoch": 6.71,
-      "learning_rate": 5.2637722700813694e-05,
-      "loss": 1.7868,
-      "step": 466000
-    },
-    {
-      "epoch": 6.71,
-      "learning_rate": 5.2585795959060956e-05,
-      "loss": 1.7855,
-      "step": 466500
-    },
-    {
-      "epoch": 6.72,
-      "learning_rate": 5.253386921730823e-05,
-      "loss": 1.7821,
-      "step": 467000
-    },
-    {
-      "epoch": 6.73,
-      "learning_rate": 5.248204632903899e-05,
-      "loss": 1.787,
-      "step": 467500
-    },
-    {
-      "epoch": 6.74,
-      "learning_rate": 5.2430119587286265e-05,
-      "loss": 1.7819,
-      "step": 468000
-    },
-    {
-      "epoch": 6.74,
-      "learning_rate": 5.237819284553353e-05,
-      "loss": 1.7921,
-      "step": 468500
-    },
-    {
-      "epoch": 6.75,
-      "learning_rate": 5.232626610378079e-05,
-      "loss": 1.7887,
-      "step": 469000
-    },
-    {
-      "epoch": 6.76,
-      "learning_rate": 5.227433936202806e-05,
-      "loss": 1.7878,
-      "step": 469500
-    },
-    {
-      "epoch": 6.77,
-      "learning_rate": 5.222241262027532e-05,
-      "loss": 1.7862,
-      "step": 470000
-    },
-    {
-      "epoch": 6.77,
-      "learning_rate": 5.217048587852258e-05,
-      "loss": 1.7885,
-      "step": 470500
-    },
-    {
-      "epoch": 6.78,
-      "learning_rate": 5.211855913676985e-05,
-      "loss": 1.782,
-      "step": 471000
-    },
-    {
-      "epoch": 6.79,
-      "learning_rate": 5.2066736248500614e-05,
-      "loss": 1.7865,
-      "step": 471500
-    },
-    {
-      "epoch": 6.79,
-      "learning_rate": 5.201480950674788e-05,
-      "loss": 1.7851,
-      "step": 472000
-    },
-    {
-      "epoch": 6.8,
-      "learning_rate": 5.1962882764995144e-05,
-      "loss": 1.7843,
-      "step": 472500
-    },
-    {
-      "epoch": 6.81,
-      "learning_rate": 5.1910956023242406e-05,
-      "loss": 1.7836,
-      "step": 473000
-    },
-    {
-      "epoch": 6.82,
-      "learning_rate": 5.185913313497318e-05,
-      "loss": 1.7843,
-      "step": 473500
-    },
-    {
-      "epoch": 6.82,
-      "learning_rate": 5.180720639322044e-05,
-      "loss": 1.7803,
-      "step": 474000
-    },
-    {
-      "epoch": 6.83,
-      "learning_rate": 5.1755279651467715e-05,
-      "loss": 1.783,
-      "step": 474500
-    },
-    {
-      "epoch": 6.84,
-      "learning_rate": 5.170335290971498e-05,
-      "loss": 1.7814,
-      "step": 475000
-    },
-    {
-      "epoch": 6.84,
-      "learning_rate": 5.165153002144575e-05,
-      "loss": 1.7837,
-      "step": 475500
-    },
-    {
-      "epoch": 6.85,
-      "learning_rate": 5.159960327969301e-05,
-      "loss": 1.7838,
-      "step": 476000
-    },
-    {
-      "epoch": 6.86,
-      "learning_rate": 5.154767653794028e-05,
-      "loss": 1.7855,
-      "step": 476500
-    },
-    {
-      "epoch": 6.87,
-      "learning_rate": 5.149574979618754e-05,
-      "loss": 1.7831,
-      "step": 477000
-    },
-    {
-      "epoch": 6.87,
-      "learning_rate": 5.1443926907918314e-05,
-      "loss": 1.7825,
-      "step": 477500
-    },
-    {
-      "epoch": 6.88,
-      "learning_rate": 5.1392000166165576e-05,
-      "loss": 1.7831,
-      "step": 478000
-    },
-    {
-      "epoch": 6.89,
-      "learning_rate": 5.134007342441284e-05,
-      "loss": 1.7813,
-      "step": 478500
-    },
-    {
-      "epoch": 6.89,
-      "learning_rate": 5.128814668266011e-05,
-      "loss": 1.7834,
-      "step": 479000
-    },
-    {
-      "epoch": 6.9,
-      "learning_rate": 5.123632379439087e-05,
-      "loss": 1.7812,
-      "step": 479500
-    },
-    {
-      "epoch": 6.91,
-      "learning_rate": 5.118439705263815e-05,
-      "loss": 1.7805,
-      "step": 480000
-    },
-    {
-      "epoch": 6.92,
-      "learning_rate": 5.113247031088541e-05,
-      "loss": 1.7826,
-      "step": 480500
-    },
-    {
-      "epoch": 6.92,
-      "learning_rate": 5.108054356913266e-05,
-      "loss": 1.7811,
-      "step": 481000
-    },
-    {
-      "epoch": 6.93,
-      "learning_rate": 5.102872068086344e-05,
-      "loss": 1.7838,
-      "step": 481500
-    },
-    {
-      "epoch": 6.94,
-      "learning_rate": 5.09767939391107e-05,
-      "loss": 1.7842,
-      "step": 482000
-    },
-    {
-      "epoch": 6.95,
-      "learning_rate": 5.092486719735797e-05,
-      "loss": 1.7795,
-      "step": 482500
-    },
-    {
-      "epoch": 6.95,
-      "learning_rate": 5.0872940455605234e-05,
-      "loss": 1.7821,
-      "step": 483000
-    },
-    {
-      "epoch": 6.96,
-      "learning_rate": 5.082111756733601e-05,
-      "loss": 1.7795,
-      "step": 483500
-    },
-    {
-      "epoch": 6.97,
-      "learning_rate": 5.076919082558327e-05,
-      "loss": 1.7815,
-      "step": 484000
-    },
-    {
-      "epoch": 6.97,
-      "learning_rate": 5.071726408383053e-05,
-      "loss": 1.7804,
-      "step": 484500
-    },
-    {
-      "epoch": 6.98,
-      "learning_rate": 5.06653373420778e-05,
-      "loss": 1.7786,
-      "step": 485000
-    },
-    {
-      "epoch": 6.99,
-      "learning_rate": 5.061351445380857e-05,
-      "loss": 1.7784,
-      "step": 485500
-    },
-    {
-      "epoch": 7.0,
-      "learning_rate": 5.056158771205583e-05,
-      "loss": 1.7767,
-      "step": 486000
-    },
-    {
-      "epoch": 7.0,
-      "eval_accuracy": 0.655838567605194,
-      "eval_loss": 1.647852897644043,
-      "eval_runtime": 1268.2736,
-      "eval_samples_per_second": 424.938,
-      "eval_steps_per_second": 26.559,
-      "step": 486311
-    },
-    {
-      "epoch": 7.0,
-      "learning_rate": 5.0509660970303095e-05,
-      "loss": 1.7796,
-      "step": 486500
-    },
-    {
-      "epoch": 7.01,
-      "learning_rate": 5.045773422855037e-05,
-      "loss": 1.7809,
-      "step": 487000
-    },
-    {
-      "epoch": 7.02,
-      "learning_rate": 5.040591134028113e-05,
-      "loss": 1.778,
-      "step": 487500
-    },
-    {
-      "epoch": 7.02,
-      "learning_rate": 5.0353984598528404e-05,
-      "loss": 1.7798,
-      "step": 488000
-    },
-    {
-      "epoch": 7.03,
-      "learning_rate": 5.0302057856775666e-05,
-      "loss": 1.7795,
-      "step": 488500
-    },
-    {
-      "epoch": 7.04,
-      "learning_rate": 5.025013111502293e-05,
-      "loss": 1.7764,
-      "step": 489000
-    },
-    {
-      "epoch": 7.05,
-      "learning_rate": 5.01983082267537e-05,
-      "loss": 1.7809,
-      "step": 489500
-    },
-    {
-      "epoch": 7.05,
-      "learning_rate": 5.014638148500096e-05,
-      "loss": 1.7751,
-      "step": 490000
-    },
-    {
-      "epoch": 7.06,
-      "learning_rate": 5.009445474324823e-05,
-      "loss": 1.7768,
-      "step": 490500
-    },
-    {
-      "epoch": 7.07,
-      "learning_rate": 5.004252800149549e-05,
-      "loss": 1.778,
-      "step": 491000
-    },
-    {
-      "epoch": 7.07,
-      "learning_rate": 4.999070511322626e-05,
-      "loss": 1.7785,
-      "step": 491500
-    },
-    {
-      "epoch": 7.08,
-      "learning_rate": 4.9938778371473526e-05,
-      "loss": 1.7711,
-      "step": 492000
-    },
-    {
-      "epoch": 7.09,
-      "learning_rate": 4.9886851629720794e-05,
-      "loss": 1.7772,
-      "step": 492500
-    },
-    {
-      "epoch": 7.1,
-      "learning_rate": 4.983492488796806e-05,
-      "loss": 1.7781,
-      "step": 493000
-    },
-    {
-      "epoch": 7.1,
-      "learning_rate": 4.978310199969883e-05,
-      "loss": 1.7742,
-      "step": 493500
-    },
-    {
-      "epoch": 7.11,
-      "learning_rate": 4.97311752579461e-05,
-      "loss": 1.7752,
-      "step": 494000
-    },
-    {
-      "epoch": 7.12,
-      "learning_rate": 4.967924851619335e-05,
-      "loss": 1.7816,
-      "step": 494500
-    },
-    {
-      "epoch": 7.13,
-      "learning_rate": 4.962732177444062e-05,
-      "loss": 1.7716,
-      "step": 495000
-    },
-    {
-      "epoch": 7.13,
-      "learning_rate": 4.9575498886171386e-05,
-      "loss": 1.7801,
-      "step": 495500
-    },
-    {
-      "epoch": 7.14,
-      "learning_rate": 4.9523572144418655e-05,
-      "loss": 1.7721,
-      "step": 496000
-    },
-    {
-      "epoch": 7.15,
-      "learning_rate": 4.947164540266592e-05,
-      "loss": 1.778,
-      "step": 496500
-    },
-    {
-      "epoch": 7.15,
-      "learning_rate": 4.9419718660913185e-05,
-      "loss": 1.7751,
-      "step": 497000
-    },
-    {
-      "epoch": 7.16,
-      "learning_rate": 4.936789577264396e-05,
-      "loss": 1.776,
-      "step": 497500
-    },
-    {
-      "epoch": 7.17,
-      "learning_rate": 4.9315969030891226e-05,
-      "loss": 1.774,
-      "step": 498000
-    },
-    {
-      "epoch": 7.18,
-      "learning_rate": 4.926404228913849e-05,
-      "loss": 1.7743,
-      "step": 498500
-    },
-    {
-      "epoch": 7.18,
-      "learning_rate": 4.921211554738575e-05,
-      "loss": 1.7742,
-      "step": 499000
-    },
-    {
-      "epoch": 7.19,
-      "learning_rate": 4.916029265911652e-05,
-      "loss": 1.7744,
-      "step": 499500
-    },
-    {
-      "epoch": 7.2,
-      "learning_rate": 4.9108365917363783e-05,
-      "loss": 1.7771,
-      "step": 500000
-    },
-    {
-      "epoch": 7.2,
-      "learning_rate": 4.905643917561105e-05,
-      "loss": 1.7769,
-      "step": 500500
-    },
-    {
-      "epoch": 7.21,
-      "learning_rate": 4.900451243385831e-05,
-      "loss": 1.775,
-      "step": 501000
-    },
-    {
-      "epoch": 7.22,
-      "learning_rate": 4.8952689545589086e-05,
-      "loss": 1.7775,
-      "step": 501500
-    },
-    {
-      "epoch": 7.23,
-      "learning_rate": 4.890076280383635e-05,
-      "loss": 1.7769,
-      "step": 502000
-    },
-    {
-      "epoch": 7.23,
-      "learning_rate": 4.8848836062083616e-05,
-      "loss": 1.7737,
-      "step": 502500
-    },
-    {
-      "epoch": 7.24,
-      "learning_rate": 4.879690932033088e-05,
-      "loss": 1.7752,
-      "step": 503000
-    },
-    {
-      "epoch": 7.25,
-      "learning_rate": 4.8744982578578146e-05,
-      "loss": 1.7762,
-      "step": 503500
-    },
-    {
-      "epoch": 7.25,
-      "learning_rate": 4.869315969030891e-05,
-      "loss": 1.7731,
-      "step": 504000
-    },
-    {
-      "epoch": 7.26,
-      "learning_rate": 4.864123294855618e-05,
-      "loss": 1.7755,
-      "step": 504500
-    },
-    {
-      "epoch": 7.27,
-      "learning_rate": 4.858930620680344e-05,
-      "loss": 1.7746,
-      "step": 505000
-    },
-    {
-      "epoch": 7.28,
-      "learning_rate": 4.8537483318534215e-05,
-      "loss": 1.7742,
-      "step": 505500
-    },
-    {
-      "epoch": 7.28,
-      "learning_rate": 4.8485556576781476e-05,
-      "loss": 1.7715,
-      "step": 506000
-    },
-    {
-      "epoch": 7.29,
-      "learning_rate": 4.8433629835028745e-05,
-      "loss": 1.777,
-      "step": 506500
-    },
-    {
-      "epoch": 7.3,
-      "learning_rate": 4.8381703093276006e-05,
-      "loss": 1.7754,
-      "step": 507000
-    },
-    {
-      "epoch": 7.3,
-      "learning_rate": 4.8329776351523275e-05,
-      "loss": 1.7773,
-      "step": 507500
-    },
-    {
-      "epoch": 7.31,
-      "learning_rate": 4.8277849609770536e-05,
-      "loss": 1.7759,
-      "step": 508000
-    },
-    {
-      "epoch": 7.32,
-      "learning_rate": 4.8225922868017804e-05,
-      "loss": 1.7767,
-      "step": 508500
-    },
-    {
-      "epoch": 7.33,
-      "learning_rate": 4.817399612626507e-05,
-      "loss": 1.7725,
-      "step": 509000
-    },
-    {
-      "epoch": 7.33,
-      "learning_rate": 4.812217323799584e-05,
-      "loss": 1.7718,
-      "step": 509500
-    },
-    {
-      "epoch": 7.34,
-      "learning_rate": 4.807024649624311e-05,
-      "loss": 1.7748,
-      "step": 510000
-    },
-    {
-      "epoch": 7.35,
-      "learning_rate": 4.801831975449036e-05,
-      "loss": 1.7744,
-      "step": 510500
-    },
-    {
-      "epoch": 7.36,
-      "learning_rate": 4.796639301273763e-05,
-      "loss": 1.7763,
-      "step": 511000
-    },
-    {
-      "epoch": 7.36,
-      "learning_rate": 4.79144662709849e-05,
-      "loss": 1.7711,
-      "step": 511500
-    },
-    {
-      "epoch": 7.37,
-      "learning_rate": 4.7862643382715665e-05,
-      "loss": 1.7741,
-      "step": 512000
-    },
-    {
-      "epoch": 7.38,
-      "learning_rate": 4.781071664096293e-05,
-      "loss": 1.7705,
-      "step": 512500
-    },
-    {
-      "epoch": 7.38,
-      "learning_rate": 4.77587898992102e-05,
-      "loss": 1.7703,
-      "step": 513000
-    },
-    {
-      "epoch": 7.39,
-      "learning_rate": 4.770686315745746e-05,
-      "loss": 1.7754,
-      "step": 513500
-    },
-    {
-      "epoch": 7.4,
-      "learning_rate": 4.7655040269188236e-05,
-      "loss": 1.7721,
-      "step": 514000
-    },
-    {
-      "epoch": 7.41,
-      "learning_rate": 4.76031135274355e-05,
-      "loss": 1.7728,
-      "step": 514500
-    },
-    {
-      "epoch": 7.41,
-      "learning_rate": 4.755118678568276e-05,
-      "loss": 1.7734,
-      "step": 515000
-    },
-    {
-      "epoch": 7.42,
-      "learning_rate": 4.749926004393003e-05,
-      "loss": 1.7709,
-      "step": 515500
-    },
-    {
-      "epoch": 7.43,
-      "learning_rate": 4.7447437155660793e-05,
-      "loss": 1.7707,
-      "step": 516000
-    },
-    {
-      "epoch": 7.43,
-      "learning_rate": 4.739551041390806e-05,
-      "loss": 1.7686,
-      "step": 516500
-    },
-    {
-      "epoch": 7.44,
-      "learning_rate": 4.734358367215533e-05,
-      "loss": 1.7732,
-      "step": 517000
-    },
-    {
-      "epoch": 7.45,
-      "learning_rate": 4.729165693040259e-05,
-      "loss": 1.7755,
-      "step": 517500
-    },
-    {
-      "epoch": 7.46,
-      "learning_rate": 4.7239834042133365e-05,
-      "loss": 1.7707,
-      "step": 518000
-    },
-    {
-      "epoch": 7.46,
-      "learning_rate": 4.7187907300380626e-05,
-      "loss": 1.77,
-      "step": 518500
-    },
-    {
-      "epoch": 7.47,
-      "learning_rate": 4.713598055862789e-05,
-      "loss": 1.7733,
-      "step": 519000
-    },
-    {
-      "epoch": 7.48,
-      "learning_rate": 4.7084053816875156e-05,
-      "loss": 1.7701,
-      "step": 519500
-    },
-    {
-      "epoch": 7.48,
-      "learning_rate": 4.703223092860592e-05,
-      "loss": 1.7704,
-      "step": 520000
-    },
-    {
-      "epoch": 7.49,
-      "learning_rate": 4.698030418685319e-05,
-      "loss": 1.7733,
-      "step": 520500
-    },
-    {
-      "epoch": 7.5,
-      "learning_rate": 4.692837744510045e-05,
-      "loss": 1.7719,
-      "step": 521000
-    },
-    {
-      "epoch": 7.51,
-      "learning_rate": 4.687645070334772e-05,
-      "loss": 1.7713,
-      "step": 521500
-    },
-    {
-      "epoch": 7.51,
-      "learning_rate": 4.682462781507849e-05,
-      "loss": 1.7694,
-      "step": 522000
-    },
-    {
-      "epoch": 7.52,
-      "learning_rate": 4.6772701073325755e-05,
-      "loss": 1.7745,
-      "step": 522500
-    },
-    {
-      "epoch": 7.53,
-      "learning_rate": 4.6720774331573016e-05,
-      "loss": 1.7699,
-      "step": 523000
-    },
-    {
-      "epoch": 7.54,
-      "learning_rate": 4.6668847589820285e-05,
-      "loss": 1.7734,
-      "step": 523500
-    },
-    {
-      "epoch": 7.54,
-      "learning_rate": 4.661702470155106e-05,
-      "loss": 1.7732,
-      "step": 524000
-    },
-    {
-      "epoch": 7.55,
-      "learning_rate": 4.656509795979832e-05,
-      "loss": 1.7699,
-      "step": 524500
-    },
-    {
-      "epoch": 7.56,
-      "learning_rate": 4.651317121804558e-05,
-      "loss": 1.7735,
-      "step": 525000
-    },
-    {
-      "epoch": 7.56,
-      "learning_rate": 4.646124447629285e-05,
-      "loss": 1.7715,
-      "step": 525500
-    },
-    {
-      "epoch": 7.57,
-      "learning_rate": 4.6409421588023615e-05,
-      "loss": 1.7697,
-      "step": 526000
-    },
-    {
-      "epoch": 7.58,
-      "learning_rate": 4.6357494846270883e-05,
-      "loss": 1.7693,
-      "step": 526500
-    },
-    {
-      "epoch": 7.59,
-      "learning_rate": 4.630556810451815e-05,
-      "loss": 1.769,
-      "step": 527000
-    },
-    {
-      "epoch": 7.59,
-      "learning_rate": 4.625364136276541e-05,
-      "loss": 1.769,
-      "step": 527500
-    },
-    {
-      "epoch": 7.6,
-      "learning_rate": 4.6201818474496186e-05,
-      "loss": 1.7745,
-      "step": 528000
-    },
-    {
-      "epoch": 7.61,
-      "learning_rate": 4.614989173274345e-05,
-      "loss": 1.7707,
-      "step": 528500
-    },
-    {
-      "epoch": 7.61,
-      "learning_rate": 4.609796499099071e-05,
-      "loss": 1.7688,
-      "step": 529000
-    },
-    {
-      "epoch": 7.62,
-      "learning_rate": 4.604614210272148e-05,
-      "loss": 1.7706,
-      "step": 529500
-    },
-    {
-      "epoch": 7.63,
-      "learning_rate": 4.5994215360968744e-05,
-      "loss": 1.773,
-      "step": 530000
-    },
-    {
-      "epoch": 7.64,
-      "learning_rate": 4.594228861921601e-05,
-      "loss": 1.7672,
-      "step": 530500
-    },
-    {
-      "epoch": 7.64,
-      "learning_rate": 4.589036187746328e-05,
-      "loss": 1.7676,
-      "step": 531000
-    },
-    {
-      "epoch": 7.65,
-      "learning_rate": 4.583843513571054e-05,
-      "loss": 1.7676,
-      "step": 531500
-    },
-    {
-      "epoch": 7.66,
-      "learning_rate": 4.5786508393957803e-05,
-      "loss": 1.7707,
-      "step": 532000
-    },
-    {
-      "epoch": 7.66,
-      "learning_rate": 4.573458165220507e-05,
-      "loss": 1.7719,
-      "step": 532500
-    },
-    {
-      "epoch": 7.67,
-      "learning_rate": 4.568265491045234e-05,
-      "loss": 1.7694,
-      "step": 533000
-    },
-    {
-      "epoch": 7.68,
-      "learning_rate": 4.5630832022183106e-05,
-      "loss": 1.7724,
-      "step": 533500
-    },
-    {
-      "epoch": 7.69,
-      "learning_rate": 4.5578905280430375e-05,
-      "loss": 1.7717,
-      "step": 534000
-    },
-    {
-      "epoch": 7.69,
-      "learning_rate": 4.5526978538677636e-05,
-      "loss": 1.7678,
-      "step": 534500
-    },
-    {
-      "epoch": 7.7,
-      "learning_rate": 4.54750517969249e-05,
-      "loss": 1.7696,
-      "step": 535000
-    },
-    {
-      "epoch": 7.71,
-      "learning_rate": 4.542322890865567e-05,
-      "loss": 1.7684,
-      "step": 535500
-    },
-    {
-      "epoch": 7.72,
-      "learning_rate": 4.5371406020386444e-05,
-      "loss": 1.7627,
-      "step": 536000
-    },
-    {
-      "epoch": 7.72,
-      "learning_rate": 4.5319479278633705e-05,
-      "loss": 1.7686,
-      "step": 536500
-    },
-    {
-      "epoch": 7.73,
-      "learning_rate": 4.526755253688097e-05,
-      "loss": 1.7653,
-      "step": 537000
-    },
-    {
-      "epoch": 7.74,
-      "learning_rate": 4.5215625795128235e-05,
-      "loss": 1.766,
-      "step": 537500
-    },
-    {
-      "epoch": 7.74,
-      "learning_rate": 4.51636990533755e-05,
-      "loss": 1.7649,
-      "step": 538000
-    },
-    {
-      "epoch": 7.75,
-      "learning_rate": 4.5111772311622765e-05,
-      "loss": 1.7658,
-      "step": 538500
-    },
-    {
-      "epoch": 7.76,
-      "learning_rate": 4.5059845569870026e-05,
-      "loss": 1.7638,
-      "step": 539000
-    },
-    {
-      "epoch": 7.77,
-      "learning_rate": 4.5007918828117295e-05,
-      "loss": 1.7658,
-      "step": 539500
-    },
-    {
-      "epoch": 7.77,
-      "learning_rate": 4.495609593984806e-05,
-      "loss": 1.7654,
-      "step": 540000
-    },
-    {
-      "epoch": 7.78,
-      "learning_rate": 4.490416919809533e-05,
-      "loss": 1.7675,
-      "step": 540500
-    },
-    {
-      "epoch": 7.79,
-      "learning_rate": 4.485224245634259e-05,
-      "loss": 1.767,
-      "step": 541000
-    },
-    {
-      "epoch": 7.79,
-      "learning_rate": 4.480031571458986e-05,
-      "loss": 1.7705,
-      "step": 541500
-    },
-    {
-      "epoch": 7.8,
-      "learning_rate": 4.474849282632063e-05,
-      "loss": 1.7689,
-      "step": 542000
-    },
-    {
-      "epoch": 7.81,
-      "learning_rate": 4.4696566084567893e-05,
-      "loss": 1.7689,
-      "step": 542500
-    },
-    {
-      "epoch": 7.82,
-      "learning_rate": 4.464463934281516e-05,
-      "loss": 1.7701,
-      "step": 543000
-    },
-    {
-      "epoch": 7.82,
-      "learning_rate": 4.459271260106242e-05,
-      "loss": 1.7675,
-      "step": 543500
-    },
-    {
-      "epoch": 7.83,
-      "learning_rate": 4.4540889712793196e-05,
-      "loss": 1.7674,
-      "step": 544000
-    },
-    {
-      "epoch": 7.84,
-      "learning_rate": 4.448896297104046e-05,
-      "loss": 1.7654,
-      "step": 544500
-    },
-    {
-      "epoch": 7.84,
-      "learning_rate": 4.443703622928772e-05,
-      "loss": 1.7649,
-      "step": 545000
-    },
-    {
-      "epoch": 7.85,
-      "learning_rate": 4.438510948753499e-05,
-      "loss": 1.7676,
-      "step": 545500
-    },
-    {
-      "epoch": 7.86,
-      "learning_rate": 4.4333286599265754e-05,
-      "loss": 1.7621,
-      "step": 546000
-    },
-    {
-      "epoch": 7.87,
-      "learning_rate": 4.428135985751302e-05,
-      "loss": 1.7628,
-      "step": 546500
-    },
-    {
-      "epoch": 7.87,
-      "learning_rate": 4.422943311576029e-05,
-      "loss": 1.7692,
-      "step": 547000
-    },
-    {
-      "epoch": 7.88,
-      "learning_rate": 4.417750637400755e-05,
-      "loss": 1.7658,
-      "step": 547500
-    },
-    {
-      "epoch": 7.89,
-      "learning_rate": 4.4125683485738325e-05,
-      "loss": 1.7659,
-      "step": 548000
-    },
-    {
-      "epoch": 7.9,
-      "learning_rate": 4.4073756743985586e-05,
-      "loss": 1.7643,
-      "step": 548500
-    },
-    {
-      "epoch": 7.9,
-      "learning_rate": 4.402183000223285e-05,
-      "loss": 1.7654,
-      "step": 549000
-    },
-    {
-      "epoch": 7.91,
-      "learning_rate": 4.3969903260480116e-05,
-      "loss": 1.7654,
-      "step": 549500
-    },
-    {
-      "epoch": 7.92,
-      "learning_rate": 4.391808037221088e-05,
-      "loss": 1.7648,
-      "step": 550000
-    },
-    {
-      "epoch": 7.92,
-      "learning_rate": 4.386615363045815e-05,
-      "loss": 1.764,
-      "step": 550500
-    },
-    {
-      "epoch": 7.93,
-      "learning_rate": 4.381422688870542e-05,
-      "loss": 1.7654,
-      "step": 551000
-    },
-    {
-      "epoch": 7.94,
-      "learning_rate": 4.376230014695268e-05,
-      "loss": 1.7628,
-      "step": 551500
-    },
-    {
-      "epoch": 7.95,
-      "learning_rate": 4.3710477258683454e-05,
-      "loss": 1.7696,
-      "step": 552000
-    },
-    {
-      "epoch": 7.95,
-      "learning_rate": 4.365855051693072e-05,
-      "loss": 1.7657,
-      "step": 552500
-    },
-    {
-      "epoch": 7.96,
-      "learning_rate": 4.360662377517798e-05,
-      "loss": 1.7621,
-      "step": 553000
-    },
-    {
-      "epoch": 7.97,
-      "learning_rate": 4.3554697033425245e-05,
-      "loss": 1.7617,
-      "step": 553500
-    },
-    {
-      "epoch": 7.97,
-      "learning_rate": 4.350287414515601e-05,
-      "loss": 1.7636,
-      "step": 554000
-    },
-    {
-      "epoch": 7.98,
-      "learning_rate": 4.345094740340328e-05,
-      "loss": 1.7631,
-      "step": 554500
-    },
-    {
-      "epoch": 7.99,
-      "learning_rate": 4.339902066165055e-05,
-      "loss": 1.7652,
-      "step": 555000
-    },
-    {
-      "epoch": 8.0,
-      "learning_rate": 4.334709391989781e-05,
-      "loss": 1.7663,
-      "step": 555500
-    },
-    {
-      "epoch": 8.0,
-      "eval_accuracy": 0.6577083685785928,
-      "eval_loss": 1.633870005607605,
-      "eval_runtime": 1283.7415,
-      "eval_samples_per_second": 419.817,
-      "eval_steps_per_second": 26.239,
-      "step": 555784
-    },
-    {
-      "epoch": 8.0,
-      "learning_rate": 4.329527103162858e-05,
-      "loss": 1.7639,
-      "step": 556000
-    },
-    {
-      "epoch": 8.01,
-      "learning_rate": 4.324334428987585e-05,
-      "loss": 1.757,
-      "step": 556500
-    },
-    {
-      "epoch": 8.02,
-      "learning_rate": 4.319141754812311e-05,
-      "loss": 1.7587,
-      "step": 557000
-    },
-    {
-      "epoch": 8.02,
-      "learning_rate": 4.3139490806370374e-05,
-      "loss": 1.761,
-      "step": 557500
-    },
-    {
-      "epoch": 8.03,
-      "learning_rate": 4.3087667918101147e-05,
-      "loss": 1.7578,
-      "step": 558000
-    },
-    {
-      "epoch": 8.04,
-      "learning_rate": 4.303574117634841e-05,
-      "loss": 1.7618,
-      "step": 558500
-    },
-    {
-      "epoch": 8.05,
-      "learning_rate": 4.2983814434595676e-05,
-      "loss": 1.7611,
-      "step": 559000
-    },
-    {
-      "epoch": 8.05,
-      "learning_rate": 4.293188769284294e-05,
-      "loss": 1.7625,
-      "step": 559500
-    },
-    {
-      "epoch": 8.06,
-      "learning_rate": 4.288006480457371e-05,
-      "loss": 1.7631,
-      "step": 560000
-    },
-    {
-      "epoch": 8.07,
-      "learning_rate": 4.282813806282097e-05,
-      "loss": 1.7608,
-      "step": 560500
-    },
-    {
-      "epoch": 8.08,
-      "learning_rate": 4.277621132106824e-05,
-      "loss": 1.7602,
-      "step": 561000
-    },
-    {
-      "epoch": 8.08,
-      "learning_rate": 4.27242845793155e-05,
-      "loss": 1.7622,
-      "step": 561500
-    },
-    {
-      "epoch": 8.09,
-      "learning_rate": 4.2672461691046275e-05,
-      "loss": 1.7612,
-      "step": 562000
-    },
-    {
-      "epoch": 8.1,
-      "learning_rate": 4.262053494929354e-05,
-      "loss": 1.7629,
-      "step": 562500
-    },
-    {
-      "epoch": 8.1,
-      "learning_rate": 4.2568608207540805e-05,
-      "loss": 1.76,
-      "step": 563000
-    },
-    {
-      "epoch": 8.11,
-      "learning_rate": 4.251678531927157e-05,
-      "loss": 1.757,
-      "step": 563500
-    },
-    {
-      "epoch": 8.12,
-      "learning_rate": 4.246485857751884e-05,
-      "loss": 1.76,
-      "step": 564000
-    },
-    {
-      "epoch": 8.13,
-      "learning_rate": 4.24129318357661e-05,
-      "loss": 1.7598,
-      "step": 564500
-    },
-    {
-      "epoch": 8.13,
-      "learning_rate": 4.236100509401337e-05,
-      "loss": 1.7594,
-      "step": 565000
-    },
-    {
-      "epoch": 8.14,
-      "learning_rate": 4.230907835226063e-05,
-      "loss": 1.7601,
-      "step": 565500
-    },
-    {
-      "epoch": 8.15,
-      "learning_rate": 4.22571516105079e-05,
-      "loss": 1.7602,
-      "step": 566000
-    },
-    {
-      "epoch": 8.15,
-      "learning_rate": 4.220522486875516e-05,
-      "loss": 1.7642,
-      "step": 566500
-    },
-    {
-      "epoch": 8.16,
-      "learning_rate": 4.215329812700243e-05,
-      "loss": 1.7609,
-      "step": 567000
-    },
-    {
-      "epoch": 8.17,
-      "learning_rate": 4.2101475238733195e-05,
-      "loss": 1.7595,
-      "step": 567500
-    },
-    {
-      "epoch": 8.18,
-      "learning_rate": 4.2049548496980464e-05,
-      "loss": 1.762,
-      "step": 568000
-    },
-    {
-      "epoch": 8.18,
-      "learning_rate": 4.1997621755227725e-05,
-      "loss": 1.758,
-      "step": 568500
-    },
-    {
-      "epoch": 8.19,
-      "learning_rate": 4.1945695013474987e-05,
-      "loss": 1.7594,
-      "step": 569000
-    },
-    {
-      "epoch": 8.2,
-      "learning_rate": 4.1893872125205766e-05,
-      "loss": 1.7586,
-      "step": 569500
-    },
-    {
-      "epoch": 8.2,
-      "learning_rate": 4.184194538345302e-05,
-      "loss": 1.7621,
-      "step": 570000
-    },
-    {
-      "epoch": 8.21,
-      "learning_rate": 4.179001864170029e-05,
-      "loss": 1.7619,
-      "step": 570500
-    },
-    {
-      "epoch": 8.22,
-      "learning_rate": 4.173809189994756e-05,
-      "loss": 1.7591,
-      "step": 571000
-    },
-    {
-      "epoch": 8.23,
-      "learning_rate": 4.1686269011678324e-05,
-      "loss": 1.7582,
-      "step": 571500
-    },
-    {
-      "epoch": 8.23,
-      "learning_rate": 4.163434226992559e-05,
-      "loss": 1.7603,
-      "step": 572000
-    },
-    {
-      "epoch": 8.24,
-      "learning_rate": 4.158251938165636e-05,
-      "loss": 1.7545,
-      "step": 572500
-    },
-    {
-      "epoch": 8.25,
-      "learning_rate": 4.153059263990363e-05,
-      "loss": 1.7649,
-      "step": 573000
-    },
-    {
-      "epoch": 8.26,
-      "learning_rate": 4.1478665898150895e-05,
-      "loss": 1.756,
-      "step": 573500
-    },
-    {
-      "epoch": 8.26,
-      "learning_rate": 4.1426739156398157e-05,
-      "loss": 1.7582,
-      "step": 574000
-    },
-    {
-      "epoch": 8.27,
-      "learning_rate": 4.137481241464542e-05,
-      "loss": 1.7578,
-      "step": 574500
-    },
-    {
-      "epoch": 8.28,
-      "learning_rate": 4.1322885672892686e-05,
-      "loss": 1.763,
-      "step": 575000
-    },
-    {
-      "epoch": 8.28,
-      "learning_rate": 4.1270958931139955e-05,
-      "loss": 1.7579,
-      "step": 575500
-    },
-    {
-      "epoch": 8.29,
-      "learning_rate": 4.1219032189387216e-05,
-      "loss": 1.7617,
-      "step": 576000
-    },
-    {
-      "epoch": 8.3,
-      "learning_rate": 4.116720930111799e-05,
-      "loss": 1.7528,
-      "step": 576500
-    },
-    {
-      "epoch": 8.31,
-      "learning_rate": 4.111528255936525e-05,
-      "loss": 1.7591,
-      "step": 577000
-    },
-    {
-      "epoch": 8.31,
-      "learning_rate": 4.106335581761251e-05,
-      "loss": 1.7559,
-      "step": 577500
-    },
-    {
-      "epoch": 8.32,
-      "learning_rate": 4.101142907585978e-05,
-      "loss": 1.7591,
-      "step": 578000
-    },
-    {
-      "epoch": 8.33,
-      "learning_rate": 4.095960618759055e-05,
-      "loss": 1.7552,
-      "step": 578500
-    },
-    {
-      "epoch": 8.33,
-      "learning_rate": 4.0907679445837815e-05,
-      "loss": 1.7548,
-      "step": 579000
-    },
-    {
-      "epoch": 8.34,
-      "learning_rate": 4.0855752704085077e-05,
-      "loss": 1.7547,
-      "step": 579500
-    },
-    {
-      "epoch": 8.35,
-      "learning_rate": 4.0803825962332345e-05,
-      "loss": 1.7593,
-      "step": 580000
-    },
-    {
-      "epoch": 8.36,
-      "learning_rate": 4.075200307406312e-05,
-      "loss": 1.7594,
-      "step": 580500
-    },
-    {
-      "epoch": 8.36,
-      "learning_rate": 4.070007633231038e-05,
-      "loss": 1.7571,
-      "step": 581000
-    },
-    {
-      "epoch": 8.37,
-      "learning_rate": 4.064814959055764e-05,
-      "loss": 1.7594,
-      "step": 581500
-    },
-    {
-      "epoch": 8.38,
-      "learning_rate": 4.059622284880491e-05,
-      "loss": 1.7547,
-      "step": 582000
-    },
-    {
-      "epoch": 8.38,
-      "learning_rate": 4.0544399960535675e-05,
-      "loss": 1.7556,
-      "step": 582500
-    },
-    {
-      "epoch": 8.39,
-      "learning_rate": 4.0492473218782944e-05,
-      "loss": 1.7571,
-      "step": 583000
-    },
-    {
-      "epoch": 8.4,
-      "learning_rate": 4.0440546477030205e-05,
-      "loss": 1.7569,
-      "step": 583500
-    },
-    {
-      "epoch": 8.41,
-      "learning_rate": 4.0388619735277474e-05,
-      "loss": 1.7576,
-      "step": 584000
-    },
-    {
-      "epoch": 8.41,
-      "learning_rate": 4.033679684700824e-05,
-      "loss": 1.7582,
-      "step": 584500
-    },
-    {
-      "epoch": 8.42,
-      "learning_rate": 4.028487010525551e-05,
-      "loss": 1.7555,
-      "step": 585000
-    },
-    {
-      "epoch": 8.43,
-      "learning_rate": 4.0232943363502776e-05,
-      "loss": 1.7571,
-      "step": 585500
-    },
-    {
-      "epoch": 8.43,
-      "learning_rate": 4.018101662175004e-05,
-      "loss": 1.7567,
-      "step": 586000
-    },
-    {
-      "epoch": 8.44,
-      "learning_rate": 4.012919373348081e-05,
-      "loss": 1.753,
-      "step": 586500
-    },
-    {
-      "epoch": 8.45,
-      "learning_rate": 4.007726699172807e-05,
-      "loss": 1.7573,
-      "step": 587000
-    },
-    {
-      "epoch": 8.46,
-      "learning_rate": 4.0025340249975334e-05,
-      "loss": 1.7568,
-      "step": 587500
-    },
-    {
-      "epoch": 8.46,
-      "learning_rate": 3.99734135082226e-05,
-      "loss": 1.7521,
-      "step": 588000
-    },
-    {
-      "epoch": 8.47,
-      "learning_rate": 3.992159061995337e-05,
-      "loss": 1.7587,
-      "step": 588500
-    },
-    {
-      "epoch": 8.48,
-      "learning_rate": 3.986966387820064e-05,
-      "loss": 1.7561,
-      "step": 589000
-    },
-    {
-      "epoch": 8.49,
-      "learning_rate": 3.9817737136447905e-05,
-      "loss": 1.7552,
-      "step": 589500
-    },
-    {
-      "epoch": 8.49,
-      "learning_rate": 3.9765810394695167e-05,
-      "loss": 1.7615,
-      "step": 590000
-    },
-    {
-      "epoch": 8.5,
-      "learning_rate": 3.971398750642594e-05,
-      "loss": 1.7554,
-      "step": 590500
-    },
-    {
-      "epoch": 8.51,
-      "learning_rate": 3.96620607646732e-05,
-      "loss": 1.7579,
-      "step": 591000
-    },
-    {
-      "epoch": 8.51,
-      "learning_rate": 3.961013402292046e-05,
-      "loss": 1.7531,
-      "step": 591500
-    },
-    {
-      "epoch": 8.52,
-      "learning_rate": 3.955820728116773e-05,
-      "loss": 1.7594,
-      "step": 592000
-    },
-    {
-      "epoch": 8.53,
-      "learning_rate": 3.95063843928985e-05,
-      "loss": 1.7561,
-      "step": 592500
-    },
-    {
-      "epoch": 8.54,
-      "learning_rate": 3.9454457651145765e-05,
-      "loss": 1.758,
-      "step": 593000
-    },
-    {
-      "epoch": 8.54,
-      "learning_rate": 3.9402530909393034e-05,
-      "loss": 1.7567,
-      "step": 593500
-    },
-    {
-      "epoch": 8.55,
-      "learning_rate": 3.9350604167640295e-05,
-      "loss": 1.7568,
-      "step": 594000
-    },
-    {
-      "epoch": 8.56,
-      "learning_rate": 3.929878127937107e-05,
-      "loss": 1.7541,
-      "step": 594500
-    },
-    {
-      "epoch": 8.56,
-      "learning_rate": 3.924685453761833e-05,
-      "loss": 1.7548,
-      "step": 595000
-    },
-    {
-      "epoch": 8.57,
-      "learning_rate": 3.919492779586559e-05,
-      "loss": 1.7566,
-      "step": 595500
-    },
-    {
-      "epoch": 8.58,
-      "learning_rate": 3.914300105411286e-05,
-      "loss": 1.7522,
-      "step": 596000
-    },
-    {
-      "epoch": 8.59,
-      "learning_rate": 3.9091178165843626e-05,
-      "loss": 1.7527,
-      "step": 596500
-    },
-    {
-      "epoch": 8.59,
-      "learning_rate": 3.9039251424090894e-05,
-      "loss": 1.7552,
-      "step": 597000
-    },
-    {
-      "epoch": 8.6,
-      "learning_rate": 3.898732468233816e-05,
-      "loss": 1.7583,
-      "step": 597500
-    },
-    {
-      "epoch": 8.61,
-      "learning_rate": 3.893550179406893e-05,
-      "loss": 1.7533,
-      "step": 598000
-    },
-    {
-      "epoch": 8.61,
-      "learning_rate": 3.88835750523162e-05,
-      "loss": 1.7533,
-      "step": 598500
-    },
-    {
-      "epoch": 8.62,
-      "learning_rate": 3.883164831056346e-05,
-      "loss": 1.7521,
-      "step": 599000
-    },
-    {
-      "epoch": 8.63,
-      "learning_rate": 3.877972156881072e-05,
-      "loss": 1.7587,
-      "step": 599500
-    },
-    {
-      "epoch": 8.64,
-      "learning_rate": 3.872779482705799e-05,
-      "loss": 1.7544,
-      "step": 600000
-    },
-    {
-      "epoch": 8.64,
-      "learning_rate": 3.8675868085305257e-05,
-      "loss": 1.7521,
-      "step": 600500
-    },
-    {
-      "epoch": 8.65,
-      "learning_rate": 3.862394134355252e-05,
-      "loss": 1.7533,
-      "step": 601000
-    },
-    {
-      "epoch": 8.66,
-      "learning_rate": 3.857201460179978e-05,
-      "loss": 1.755,
-      "step": 601500
-    },
-    {
-      "epoch": 8.67,
-      "learning_rate": 3.852019171353055e-05,
-      "loss": 1.7522,
-      "step": 602000
-    },
-    {
-      "epoch": 8.67,
-      "learning_rate": 3.846826497177782e-05,
-      "loss": 1.7543,
-      "step": 602500
-    },
-    {
-      "epoch": 8.68,
-      "learning_rate": 3.841633823002508e-05,
-      "loss": 1.7516,
-      "step": 603000
-    },
-    {
-      "epoch": 8.69,
-      "learning_rate": 3.8364411488272344e-05,
-      "loss": 1.7528,
-      "step": 603500
-    },
-    {
-      "epoch": 8.69,
-      "learning_rate": 3.831258860000312e-05,
-      "loss": 1.752,
-      "step": 604000
-    },
-    {
-      "epoch": 8.7,
-      "learning_rate": 3.826066185825038e-05,
-      "loss": 1.7555,
-      "step": 604500
-    },
-    {
-      "epoch": 8.71,
-      "learning_rate": 3.820873511649765e-05,
-      "loss": 1.753,
-      "step": 605000
-    },
-    {
-      "epoch": 8.72,
-      "learning_rate": 3.8156808374744915e-05,
-      "loss": 1.7504,
-      "step": 605500
-    },
-    {
-      "epoch": 8.72,
-      "learning_rate": 3.810498548647568e-05,
-      "loss": 1.7575,
-      "step": 606000
-    },
-    {
-      "epoch": 8.73,
-      "learning_rate": 3.8053162598206454e-05,
-      "loss": 1.7528,
-      "step": 606500
-    },
-    {
-      "epoch": 8.74,
-      "learning_rate": 3.8001235856453716e-05,
-      "loss": 1.7548,
-      "step": 607000
-    },
-    {
-      "epoch": 8.74,
-      "learning_rate": 3.7949309114700984e-05,
-      "loss": 1.7483,
-      "step": 607500
-    },
-    {
-      "epoch": 8.75,
-      "learning_rate": 3.7897382372948246e-05,
-      "loss": 1.7564,
-      "step": 608000
-    },
-    {
-      "epoch": 8.76,
-      "learning_rate": 3.784545563119551e-05,
-      "loss": 1.7495,
-      "step": 608500
-    },
-    {
-      "epoch": 8.77,
-      "learning_rate": 3.7793528889442775e-05,
-      "loss": 1.7505,
-      "step": 609000
-    },
-    {
-      "epoch": 8.77,
-      "learning_rate": 3.7741602147690044e-05,
-      "loss": 1.7513,
-      "step": 609500
-    },
-    {
-      "epoch": 8.78,
-      "learning_rate": 3.7689675405937305e-05,
-      "loss": 1.7502,
-      "step": 610000
-    },
-    {
-      "epoch": 8.79,
-      "learning_rate": 3.763785251766808e-05,
-      "loss": 1.7521,
-      "step": 610500
-    },
-    {
-      "epoch": 8.79,
-      "learning_rate": 3.758592577591534e-05,
-      "loss": 1.7514,
-      "step": 611000
-    },
-    {
-      "epoch": 8.8,
-      "learning_rate": 3.75339990341626e-05,
-      "loss": 1.7505,
-      "step": 611500
-    },
-    {
-      "epoch": 8.81,
-      "learning_rate": 3.748207229240987e-05,
-      "loss": 1.7482,
-      "step": 612000
-    },
-    {
-      "epoch": 8.82,
-      "learning_rate": 3.743035325762415e-05,
-      "loss": 1.7525,
-      "step": 612500
-    },
-    {
-      "epoch": 8.82,
-      "learning_rate": 3.737842651587141e-05,
-      "loss": 1.7561,
-      "step": 613000
-    },
-    {
-      "epoch": 8.83,
-      "learning_rate": 3.732649977411867e-05,
-      "loss": 1.7582,
-      "step": 613500
-    },
-    {
-      "epoch": 8.84,
-      "learning_rate": 3.727457303236594e-05,
-      "loss": 1.7514,
-      "step": 614000
-    },
-    {
-      "epoch": 8.85,
-      "learning_rate": 3.722264629061321e-05,
-      "loss": 1.7517,
-      "step": 614500
-    },
-    {
-      "epoch": 8.85,
-      "learning_rate": 3.7170719548860475e-05,
-      "loss": 1.7498,
-      "step": 615000
-    },
-    {
-      "epoch": 8.86,
-      "learning_rate": 3.711879280710773e-05,
-      "loss": 1.7558,
-      "step": 615500
-    },
-    {
-      "epoch": 8.87,
-      "learning_rate": 3.7066866065355e-05,
-      "loss": 1.7502,
-      "step": 616000
-    },
-    {
-      "epoch": 8.87,
-      "learning_rate": 3.7015043177085765e-05,
-      "loss": 1.753,
-      "step": 616500
-    },
-    {
-      "epoch": 8.88,
-      "learning_rate": 3.696311643533303e-05,
-      "loss": 1.7497,
-      "step": 617000
-    },
-    {
-      "epoch": 8.89,
-      "learning_rate": 3.69111896935803e-05,
-      "loss": 1.7474,
-      "step": 617500
-    },
-    {
-      "epoch": 8.9,
-      "learning_rate": 3.685936680531107e-05,
-      "loss": 1.7486,
-      "step": 618000
-    },
-    {
-      "epoch": 8.9,
-      "learning_rate": 3.6807440063558336e-05,
-      "loss": 1.7488,
-      "step": 618500
-    },
-    {
-      "epoch": 8.91,
-      "learning_rate": 3.67555133218056e-05,
-      "loss": 1.7491,
-      "step": 619000
-    },
-    {
-      "epoch": 8.92,
-      "learning_rate": 3.6703586580052865e-05,
-      "loss": 1.7489,
-      "step": 619500
-    },
-    {
-      "epoch": 8.92,
-      "learning_rate": 3.665165983830013e-05,
-      "loss": 1.7524,
-      "step": 620000
-    },
-    {
-      "epoch": 8.93,
-      "learning_rate": 3.6599733096547395e-05,
-      "loss": 1.7538,
-      "step": 620500
-    },
-    {
-      "epoch": 8.94,
-      "learning_rate": 3.654780635479466e-05,
-      "loss": 1.7516,
-      "step": 621000
-    },
-    {
-      "epoch": 8.95,
-      "learning_rate": 3.6495879613041925e-05,
-      "loss": 1.749,
-      "step": 621500
-    },
-    {
-      "epoch": 8.95,
-      "learning_rate": 3.644405672477269e-05,
-      "loss": 1.7493,
-      "step": 622000
-    },
-    {
-      "epoch": 8.96,
-      "learning_rate": 3.639212998301996e-05,
-      "loss": 1.7492,
-      "step": 622500
-    },
-    {
-      "epoch": 8.97,
-      "learning_rate": 3.634020324126722e-05,
-      "loss": 1.7473,
-      "step": 623000
-    },
-    {
-      "epoch": 8.97,
-      "learning_rate": 3.628827649951448e-05,
-      "loss": 1.7485,
-      "step": 623500
-    },
-    {
-      "epoch": 8.98,
-      "learning_rate": 3.6236453611245256e-05,
-      "loss": 1.7497,
-      "step": 624000
-    },
-    {
-      "epoch": 8.99,
-      "learning_rate": 3.6184526869492524e-05,
-      "loss": 1.7524,
-      "step": 624500
-    },
-    {
-      "epoch": 9.0,
-      "learning_rate": 3.613270398122329e-05,
-      "loss": 1.7524,
-      "step": 625000
-    },
-    {
-      "epoch": 9.0,
-      "eval_accuracy": 0.6610600568746429,
-      "eval_loss": 1.6158946752548218,
-      "eval_runtime": 1285.3183,
-      "eval_samples_per_second": 419.302,
-      "eval_steps_per_second": 26.207,
-      "step": 625257
-    },
-    {
-      "epoch": 9.0,
-      "learning_rate": 3.608077723947056e-05,
-      "loss": 1.7455,
-      "step": 625500
-    },
-    {
-      "epoch": 9.01,
-      "learning_rate": 3.602885049771782e-05,
-      "loss": 1.7504,
-      "step": 626000
-    },
-    {
-      "epoch": 9.02,
-      "learning_rate": 3.597692375596509e-05,
-      "loss": 1.7482,
-      "step": 626500
-    },
-    {
-      "epoch": 9.03,
-      "learning_rate": 3.592499701421235e-05,
-      "loss": 1.748,
-      "step": 627000
-    },
-    {
-      "epoch": 9.03,
-      "learning_rate": 3.587307027245961e-05,
-      "loss": 1.7454,
-      "step": 627500
-    },
-    {
-      "epoch": 9.04,
-      "learning_rate": 3.582114353070688e-05,
-      "loss": 1.7465,
-      "step": 628000
-    },
-    {
-      "epoch": 9.05,
-      "learning_rate": 3.576921678895415e-05,
-      "loss": 1.7527,
-      "step": 628500
-    },
-    {
-      "epoch": 9.05,
-      "learning_rate": 3.5717393900684914e-05,
-      "loss": 1.7445,
-      "step": 629000
-    },
-    {
-      "epoch": 9.06,
-      "learning_rate": 3.566546715893218e-05,
-      "loss": 1.7462,
-      "step": 629500
-    },
-    {
-      "epoch": 9.07,
-      "learning_rate": 3.561354041717945e-05,
-      "loss": 1.745,
-      "step": 630000
-    },
-    {
-      "epoch": 9.08,
-      "learning_rate": 3.5561613675426705e-05,
-      "loss": 1.7473,
-      "step": 630500
-    },
-    {
-      "epoch": 9.08,
-      "learning_rate": 3.5509686933673974e-05,
-      "loss": 1.7444,
-      "step": 631000
-    },
-    {
-      "epoch": 9.09,
-      "learning_rate": 3.545786404540474e-05,
-      "loss": 1.7444,
-      "step": 631500
-    },
-    {
-      "epoch": 9.1,
-      "learning_rate": 3.540593730365201e-05,
-      "loss": 1.7499,
-      "step": 632000
-    },
-    {
-      "epoch": 9.1,
-      "learning_rate": 3.5354010561899277e-05,
-      "loss": 1.7466,
-      "step": 632500
-    },
-    {
-      "epoch": 9.11,
-      "learning_rate": 3.530208382014654e-05,
-      "loss": 1.7498,
-      "step": 633000
-    },
-    {
-      "epoch": 9.12,
-      "learning_rate": 3.52501570783938e-05,
-      "loss": 1.7483,
-      "step": 633500
-    },
-    {
-      "epoch": 9.13,
-      "learning_rate": 3.519833419012458e-05,
-      "loss": 1.7468,
-      "step": 634000
-    },
-    {
-      "epoch": 9.13,
-      "learning_rate": 3.5146407448371834e-05,
-      "loss": 1.7432,
-      "step": 634500
-    },
-    {
-      "epoch": 9.14,
-      "learning_rate": 3.50944807066191e-05,
-      "loss": 1.7428,
-      "step": 635000
-    },
-    {
-      "epoch": 9.15,
-      "learning_rate": 3.504255396486637e-05,
-      "loss": 1.749,
-      "step": 635500
-    },
-    {
-      "epoch": 9.15,
-      "learning_rate": 3.499062722311363e-05,
-      "loss": 1.7494,
-      "step": 636000
-    },
-    {
-      "epoch": 9.16,
-      "learning_rate": 3.49387004813609e-05,
-      "loss": 1.7472,
-      "step": 636500
-    },
-    {
-      "epoch": 9.17,
-      "learning_rate": 3.488677373960816e-05,
-      "loss": 1.7467,
-      "step": 637000
-    },
-    {
-      "epoch": 9.18,
-      "learning_rate": 3.4834846997855424e-05,
-      "loss": 1.7455,
-      "step": 637500
-    },
-    {
-      "epoch": 9.18,
-      "learning_rate": 3.4783024109586197e-05,
-      "loss": 1.7445,
-      "step": 638000
-    },
-    {
-      "epoch": 9.19,
-      "learning_rate": 3.473109736783346e-05,
-      "loss": 1.7454,
-      "step": 638500
-    },
-    {
-      "epoch": 9.2,
-      "learning_rate": 3.4679170626080726e-05,
-      "loss": 1.7441,
-      "step": 639000
-    },
-    {
-      "epoch": 9.21,
-      "learning_rate": 3.46273477378115e-05,
-      "loss": 1.7443,
-      "step": 639500
-    },
-    {
-      "epoch": 9.21,
-      "learning_rate": 3.457542099605876e-05,
-      "loss": 1.7463,
-      "step": 640000
-    },
-    {
-      "epoch": 9.22,
-      "learning_rate": 3.452349425430603e-05,
-      "loss": 1.7419,
-      "step": 640500
-    },
-    {
-      "epoch": 9.23,
-      "learning_rate": 3.447156751255329e-05,
-      "loss": 1.7453,
-      "step": 641000
-    },
-    {
-      "epoch": 9.23,
-      "learning_rate": 3.441964077080055e-05,
-      "loss": 1.7438,
-      "step": 641500
-    },
-    {
-      "epoch": 9.24,
-      "learning_rate": 3.436771402904782e-05,
-      "loss": 1.7424,
-      "step": 642000
-    },
-    {
-      "epoch": 9.25,
-      "learning_rate": 3.431578728729509e-05,
-      "loss": 1.745,
-      "step": 642500
-    },
-    {
-      "epoch": 9.26,
-      "learning_rate": 3.426386054554235e-05,
-      "loss": 1.7461,
-      "step": 643000
-    },
-    {
-      "epoch": 9.26,
-      "learning_rate": 3.421203765727312e-05,
-      "loss": 1.7383,
-      "step": 643500
-    },
-    {
-      "epoch": 9.27,
-      "learning_rate": 3.4160110915520385e-05,
-      "loss": 1.7443,
-      "step": 644000
-    },
-    {
-      "epoch": 9.28,
-      "learning_rate": 3.4108184173767646e-05,
-      "loss": 1.7418,
-      "step": 644500
-    },
-    {
-      "epoch": 9.28,
-      "learning_rate": 3.4056257432014915e-05,
-      "loss": 1.7446,
-      "step": 645000
-    },
-    {
-      "epoch": 9.29,
-      "learning_rate": 3.400443454374568e-05,
-      "loss": 1.743,
-      "step": 645500
-    },
-    {
-      "epoch": 9.3,
-      "learning_rate": 3.395250780199295e-05,
-      "loss": 1.7434,
-      "step": 646000
-    },
-    {
-      "epoch": 9.31,
-      "learning_rate": 3.390058106024022e-05,
-      "loss": 1.7457,
-      "step": 646500
-    },
-    {
-      "epoch": 9.31,
-      "learning_rate": 3.384865431848748e-05,
-      "loss": 1.7425,
-      "step": 647000
-    },
-    {
-      "epoch": 9.32,
-      "learning_rate": 3.379683143021825e-05,
-      "loss": 1.7444,
-      "step": 647500
-    },
-    {
-      "epoch": 9.33,
-      "learning_rate": 3.3744904688465514e-05,
-      "loss": 1.747,
-      "step": 648000
-    },
-    {
-      "epoch": 9.33,
-      "learning_rate": 3.3692977946712775e-05,
-      "loss": 1.7448,
-      "step": 648500
-    },
-    {
-      "epoch": 9.34,
-      "learning_rate": 3.3641155058443555e-05,
-      "loss": 1.744,
-      "step": 649000
-    },
-    {
-      "epoch": 9.35,
-      "learning_rate": 3.358922831669081e-05,
-      "loss": 1.7399,
-      "step": 649500
-    },
-    {
-      "epoch": 9.36,
-      "learning_rate": 3.353730157493808e-05,
-      "loss": 1.7487,
-      "step": 650000
-    },
-    {
-      "epoch": 9.36,
-      "learning_rate": 3.3485374833185346e-05,
-      "loss": 1.7477,
-      "step": 650500
-    },
-    {
-      "epoch": 9.37,
-      "learning_rate": 3.343344809143261e-05,
-      "loss": 1.7445,
-      "step": 651000
-    },
-    {
-      "epoch": 9.38,
-      "learning_rate": 3.3381521349679876e-05,
-      "loss": 1.7417,
-      "step": 651500
-    },
-    {
-      "epoch": 9.38,
-      "learning_rate": 3.332969846141064e-05,
-      "loss": 1.7456,
-      "step": 652000
-    },
-    {
-      "epoch": 9.39,
-      "learning_rate": 3.327777171965791e-05,
-      "loss": 1.7447,
-      "step": 652500
-    },
-    {
-      "epoch": 9.4,
-      "learning_rate": 3.322584497790517e-05,
-      "loss": 1.7372,
-      "step": 653000
-    },
-    {
-      "epoch": 9.41,
-      "learning_rate": 3.317391823615244e-05,
-      "loss": 1.7448,
-      "step": 653500
-    },
-    {
-      "epoch": 9.41,
-      "learning_rate": 3.31219914943997e-05,
-      "loss": 1.7469,
-      "step": 654000
-    },
-    {
-      "epoch": 9.42,
-      "learning_rate": 3.307006475264697e-05,
-      "loss": 1.7433,
-      "step": 654500
-    },
-    {
-      "epoch": 9.43,
-      "learning_rate": 3.3018241864377736e-05,
-      "loss": 1.7427,
-      "step": 655000
-    },
-    {
-      "epoch": 9.44,
-      "learning_rate": 3.2966315122625005e-05,
-      "loss": 1.7441,
-      "step": 655500
-    },
-    {
-      "epoch": 9.44,
-      "learning_rate": 3.2914388380872266e-05,
-      "loss": 1.7456,
-      "step": 656000
-    },
-    {
-      "epoch": 9.45,
-      "learning_rate": 3.286246163911953e-05,
-      "loss": 1.7416,
-      "step": 656500
-    },
-    {
-      "epoch": 9.46,
-      "learning_rate": 3.2810534897366796e-05,
-      "loss": 1.7468,
-      "step": 657000
-    },
-    {
-      "epoch": 9.46,
-      "learning_rate": 3.2758608155614064e-05,
-      "loss": 1.7435,
-      "step": 657500
-    },
-    {
-      "epoch": 9.47,
-      "learning_rate": 3.2706681413861326e-05,
-      "loss": 1.744,
-      "step": 658000
-    },
-    {
-      "epoch": 9.48,
-      "learning_rate": 3.265475467210859e-05,
-      "loss": 1.7436,
-      "step": 658500
-    },
-    {
-      "epoch": 9.49,
-      "learning_rate": 3.260293178383936e-05,
-      "loss": 1.7418,
-      "step": 659000
-    },
-    {
-      "epoch": 9.49,
-      "learning_rate": 3.255100504208662e-05,
-      "loss": 1.7398,
-      "step": 659500
-    },
-    {
-      "epoch": 9.5,
-      "learning_rate": 3.249907830033389e-05,
-      "loss": 1.7423,
-      "step": 660000
-    },
-    {
-      "epoch": 9.51,
-      "learning_rate": 3.244715155858116e-05,
-      "loss": 1.7419,
-      "step": 660500
-    },
-    {
-      "epoch": 9.51,
-      "learning_rate": 3.2395328670311925e-05,
-      "loss": 1.745,
-      "step": 661000
-    },
-    {
-      "epoch": 9.52,
-      "learning_rate": 3.234340192855919e-05,
-      "loss": 1.7421,
-      "step": 661500
-    },
-    {
-      "epoch": 9.53,
-      "learning_rate": 3.2291475186806454e-05,
-      "loss": 1.7403,
-      "step": 662000
-    },
-    {
-      "epoch": 9.54,
-      "learning_rate": 3.2239548445053716e-05,
-      "loss": 1.7404,
-      "step": 662500
-    },
-    {
-      "epoch": 9.54,
-      "learning_rate": 3.2187725556784496e-05,
-      "loss": 1.7439,
-      "step": 663000
-    },
-    {
-      "epoch": 9.55,
-      "learning_rate": 3.213579881503175e-05,
-      "loss": 1.7412,
-      "step": 663500
-    },
-    {
-      "epoch": 9.56,
-      "learning_rate": 3.208387207327902e-05,
-      "loss": 1.7418,
-      "step": 664000
-    },
-    {
-      "epoch": 9.56,
-      "learning_rate": 3.203194533152629e-05,
-      "loss": 1.7413,
-      "step": 664500
-    },
-    {
-      "epoch": 9.57,
-      "learning_rate": 3.198012244325705e-05,
-      "loss": 1.7445,
-      "step": 665000
-    },
-    {
-      "epoch": 9.58,
-      "learning_rate": 3.192819570150432e-05,
-      "loss": 1.7415,
-      "step": 665500
-    },
-    {
-      "epoch": 9.59,
-      "learning_rate": 3.187626895975158e-05,
-      "loss": 1.7374,
-      "step": 666000
-    },
-    {
-      "epoch": 9.59,
-      "learning_rate": 3.182434221799885e-05,
-      "loss": 1.7452,
-      "step": 666500
-    },
-    {
-      "epoch": 9.6,
-      "learning_rate": 3.177251932972962e-05,
-      "loss": 1.7432,
-      "step": 667000
-    },
-    {
-      "epoch": 9.61,
-      "learning_rate": 3.1720592587976886e-05,
-      "loss": 1.7409,
-      "step": 667500
-    },
-    {
-      "epoch": 9.62,
-      "learning_rate": 3.166866584622415e-05,
-      "loss": 1.7428,
-      "step": 668000
-    },
-    {
-      "epoch": 9.62,
-      "learning_rate": 3.1616739104471416e-05,
-      "loss": 1.7441,
-      "step": 668500
-    },
-    {
-      "epoch": 9.63,
-      "learning_rate": 3.156491621620218e-05,
-      "loss": 1.7421,
-      "step": 669000
-    },
-    {
-      "epoch": 9.64,
-      "learning_rate": 3.151298947444945e-05,
-      "loss": 1.738,
-      "step": 669500
-    },
-    {
-      "epoch": 9.64,
-      "learning_rate": 3.1461166586180217e-05,
-      "loss": 1.7426,
-      "step": 670000
-    },
-    {
-      "epoch": 9.65,
-      "learning_rate": 3.1409239844427485e-05,
-      "loss": 1.7417,
-      "step": 670500
-    },
-    {
-      "epoch": 9.66,
-      "learning_rate": 3.1357313102674746e-05,
-      "loss": 1.7429,
-      "step": 671000
-    },
-    {
-      "epoch": 9.67,
-      "learning_rate": 3.1305386360922015e-05,
-      "loss": 1.7447,
-      "step": 671500
-    },
-    {
-      "epoch": 9.67,
-      "learning_rate": 3.1253459619169276e-05,
-      "loss": 1.7419,
-      "step": 672000
-    },
-    {
-      "epoch": 9.68,
-      "learning_rate": 3.1201532877416544e-05,
-      "loss": 1.7406,
-      "step": 672500
-    },
-    {
-      "epoch": 9.69,
-      "learning_rate": 3.1149606135663806e-05,
-      "loss": 1.7409,
-      "step": 673000
-    },
-    {
-      "epoch": 9.69,
-      "learning_rate": 3.1097679393911074e-05,
-      "loss": 1.7345,
-      "step": 673500
-    },
-    {
-      "epoch": 9.7,
-      "learning_rate": 3.104585650564184e-05,
-      "loss": 1.7444,
-      "step": 674000
-    },
-    {
-      "epoch": 9.71,
-      "learning_rate": 3.099392976388911e-05,
-      "loss": 1.7404,
-      "step": 674500
-    },
-    {
-      "epoch": 9.72,
-      "learning_rate": 3.094200302213637e-05,
-      "loss": 1.7388,
-      "step": 675000
-    },
-    {
-      "epoch": 9.72,
-      "learning_rate": 3.089007628038363e-05,
-      "loss": 1.7414,
-      "step": 675500
-    },
-    {
-      "epoch": 9.73,
-      "learning_rate": 3.08381495386309e-05,
-      "loss": 1.7428,
-      "step": 676000
-    },
-    {
-      "epoch": 9.74,
-      "learning_rate": 3.078622279687817e-05,
-      "loss": 1.7414,
-      "step": 676500
-    },
-    {
-      "epoch": 9.74,
-      "learning_rate": 3.073429605512543e-05,
-      "loss": 1.7389,
-      "step": 677000
-    },
-    {
-      "epoch": 9.75,
-      "learning_rate": 3.068236931337269e-05,
-      "loss": 1.7396,
-      "step": 677500
-    },
-    {
-      "epoch": 9.76,
-      "learning_rate": 3.063054642510347e-05,
-      "loss": 1.7383,
-      "step": 678000
-    },
-    {
-      "epoch": 9.77,
-      "learning_rate": 3.057872353683424e-05,
-      "loss": 1.7364,
-      "step": 678500
-    },
-    {
-      "epoch": 9.77,
-      "learning_rate": 3.0526796795081506e-05,
-      "loss": 1.7393,
-      "step": 679000
-    },
-    {
-      "epoch": 9.78,
-      "learning_rate": 3.0474870053328764e-05,
-      "loss": 1.7383,
-      "step": 679500
-    },
-    {
-      "epoch": 9.79,
-      "learning_rate": 3.042294331157603e-05,
-      "loss": 1.7368,
-      "step": 680000
-    },
-    {
-      "epoch": 9.8,
-      "learning_rate": 3.0371016569823297e-05,
-      "loss": 1.7395,
-      "step": 680500
-    },
-    {
-      "epoch": 9.8,
-      "learning_rate": 3.031908982807056e-05,
-      "loss": 1.7427,
-      "step": 681000
-    },
-    {
-      "epoch": 9.81,
-      "learning_rate": 3.026726693980133e-05,
-      "loss": 1.7401,
-      "step": 681500
-    },
-    {
-      "epoch": 9.82,
-      "learning_rate": 3.0215340198048593e-05,
-      "loss": 1.7372,
-      "step": 682000
-    },
-    {
-      "epoch": 9.82,
-      "learning_rate": 3.0163413456295858e-05,
-      "loss": 1.7387,
-      "step": 682500
-    },
-    {
-      "epoch": 9.83,
-      "learning_rate": 3.0111486714543123e-05,
-      "loss": 1.741,
-      "step": 683000
-    },
-    {
-      "epoch": 9.84,
-      "learning_rate": 3.005955997279039e-05,
-      "loss": 1.7371,
-      "step": 683500
-    },
-    {
-      "epoch": 9.85,
-      "learning_rate": 3.000773708452116e-05,
-      "loss": 1.739,
-      "step": 684000
-    },
-    {
-      "epoch": 9.85,
-      "learning_rate": 2.9955810342768426e-05,
-      "loss": 1.737,
-      "step": 684500
-    },
-    {
-      "epoch": 9.86,
-      "learning_rate": 2.9903883601015687e-05,
-      "loss": 1.7383,
-      "step": 685000
-    },
-    {
-      "epoch": 9.87,
-      "learning_rate": 2.9851956859262952e-05,
-      "loss": 1.7391,
-      "step": 685500
-    },
-    {
-      "epoch": 9.87,
-      "learning_rate": 2.980003011751022e-05,
-      "loss": 1.7381,
-      "step": 686000
-    },
-    {
-      "epoch": 9.88,
-      "learning_rate": 2.974810337575748e-05,
-      "loss": 1.7364,
-      "step": 686500
-    },
-    {
-      "epoch": 9.89,
-      "learning_rate": 2.9696176634004747e-05,
-      "loss": 1.7382,
-      "step": 687000
-    },
-    {
-      "epoch": 9.9,
-      "learning_rate": 2.9644249892252012e-05,
-      "loss": 1.7347,
-      "step": 687500
-    },
-    {
-      "epoch": 9.9,
-      "learning_rate": 2.959242700398278e-05,
-      "loss": 1.7368,
-      "step": 688000
-    },
-    {
-      "epoch": 9.91,
-      "learning_rate": 2.9540500262230046e-05,
-      "loss": 1.7391,
-      "step": 688500
-    },
-    {
-      "epoch": 9.92,
-      "learning_rate": 2.9488677373960816e-05,
-      "loss": 1.7401,
-      "step": 689000
-    },
-    {
-      "epoch": 9.92,
-      "learning_rate": 2.943675063220808e-05,
-      "loss": 1.7389,
-      "step": 689500
-    },
-    {
-      "epoch": 9.93,
-      "learning_rate": 2.938482389045535e-05,
-      "loss": 1.7362,
-      "step": 690000
-    },
-    {
-      "epoch": 9.94,
-      "learning_rate": 2.933289714870261e-05,
-      "loss": 1.7357,
-      "step": 690500
-    },
-    {
-      "epoch": 9.95,
-      "learning_rate": 2.9280970406949876e-05,
-      "loss": 1.7325,
-      "step": 691000
-    },
-    {
-      "epoch": 9.95,
-      "learning_rate": 2.9229043665197144e-05,
-      "loss": 1.7402,
-      "step": 691500
-    },
-    {
-      "epoch": 9.96,
-      "learning_rate": 2.917722077692791e-05,
-      "loss": 1.7362,
-      "step": 692000
-    },
-    {
-      "epoch": 9.97,
-      "learning_rate": 2.912529403517518e-05,
-      "loss": 1.7369,
-      "step": 692500
-    },
-    {
-      "epoch": 9.98,
-      "learning_rate": 2.9073367293422443e-05,
-      "loss": 1.7375,
-      "step": 693000
-    },
-    {
-      "epoch": 9.98,
-      "learning_rate": 2.9021440551669705e-05,
-      "loss": 1.7335,
-      "step": 693500
-    },
-    {
-      "epoch": 9.99,
-      "learning_rate": 2.896951380991697e-05,
-      "loss": 1.7384,
-      "step": 694000
-    },
-    {
-      "epoch": 10.0,
-      "learning_rate": 2.8917587068164238e-05,
-      "loss": 1.7398,
-      "step": 694500
-    },
-    {
-      "epoch": 10.0,
-      "eval_accuracy": 0.6627300200297518,
-      "eval_loss": 1.6019505262374878,
-      "eval_runtime": 1291.9437,
-      "eval_samples_per_second": 417.152,
-      "eval_steps_per_second": 26.072,
-      "step": 694730
-    },
-    {
-      "epoch": 10.0,
-      "learning_rate": 2.8865660326411496e-05,
-      "loss": 1.7321,
-      "step": 695000
-    },
-    {
-      "epoch": 10.01,
-      "learning_rate": 2.8813733584658764e-05,
-      "loss": 1.7308,
-      "step": 695500
-    },
-    {
-      "epoch": 10.02,
-      "learning_rate": 2.8761910696389534e-05,
-      "loss": 1.7356,
-      "step": 696000
-    },
-    {
-      "epoch": 10.03,
-      "learning_rate": 2.8710087808120307e-05,
-      "loss": 1.735,
-      "step": 696500
-    },
-    {
-      "epoch": 10.03,
-      "learning_rate": 2.8658161066367572e-05,
-      "loss": 1.7381,
-      "step": 697000
-    },
-    {
-      "epoch": 10.04,
-      "learning_rate": 2.8606234324614833e-05,
-      "loss": 1.7334,
-      "step": 697500
-    },
-    {
-      "epoch": 10.05,
-      "learning_rate": 2.85543075828621e-05,
-      "loss": 1.7388,
-      "step": 698000
-    },
-    {
-      "epoch": 10.05,
-      "learning_rate": 2.8502380841109367e-05,
-      "loss": 1.7346,
-      "step": 698500
-    },
-    {
-      "epoch": 10.06,
-      "learning_rate": 2.8450454099356628e-05,
-      "loss": 1.7342,
-      "step": 699000
-    },
-    {
-      "epoch": 10.07,
-      "learning_rate": 2.8398527357603893e-05,
-      "loss": 1.7311,
-      "step": 699500
-    },
-    {
-      "epoch": 10.08,
-      "learning_rate": 2.834660061585116e-05,
-      "loss": 1.7361,
-      "step": 700000
-    },
-    {
-      "epoch": 10.08,
-      "learning_rate": 2.8294777727581928e-05,
-      "loss": 1.7332,
-      "step": 700500
-    },
-    {
-      "epoch": 10.09,
-      "learning_rate": 2.8242850985829196e-05,
-      "loss": 1.7365,
-      "step": 701000
-    },
-    {
-      "epoch": 10.1,
-      "learning_rate": 2.819092424407646e-05,
-      "loss": 1.7311,
-      "step": 701500
-    },
-    {
-      "epoch": 10.1,
-      "learning_rate": 2.8138997502323722e-05,
-      "loss": 1.7365,
-      "step": 702000
-    },
-    {
-      "epoch": 10.11,
-      "learning_rate": 2.8087174614054495e-05,
-      "loss": 1.7301,
-      "step": 702500
-    },
-    {
-      "epoch": 10.12,
-      "learning_rate": 2.8035247872301757e-05,
-      "loss": 1.7319,
-      "step": 703000
-    },
-    {
-      "epoch": 10.13,
-      "learning_rate": 2.7983321130549022e-05,
-      "loss": 1.7305,
-      "step": 703500
-    },
-    {
-      "epoch": 10.13,
-      "learning_rate": 2.793139438879629e-05,
-      "loss": 1.7322,
-      "step": 704000
-    },
-    {
-      "epoch": 10.14,
-      "learning_rate": 2.7879571500527056e-05,
-      "loss": 1.7325,
-      "step": 704500
-    },
-    {
-      "epoch": 10.15,
-      "learning_rate": 2.7827644758774325e-05,
-      "loss": 1.7343,
-      "step": 705000
-    },
-    {
-      "epoch": 10.16,
-      "learning_rate": 2.777582187050509e-05,
-      "loss": 1.7342,
-      "step": 705500
-    },
-    {
-      "epoch": 10.16,
-      "learning_rate": 2.772389512875236e-05,
-      "loss": 1.7341,
-      "step": 706000
-    },
-    {
-      "epoch": 10.17,
-      "learning_rate": 2.7671968386999624e-05,
-      "loss": 1.7324,
-      "step": 706500
-    },
-    {
-      "epoch": 10.18,
-      "learning_rate": 2.7620041645246886e-05,
-      "loss": 1.7331,
-      "step": 707000
-    },
-    {
-      "epoch": 10.18,
-      "learning_rate": 2.756821875697766e-05,
-      "loss": 1.7344,
-      "step": 707500
-    },
-    {
-      "epoch": 10.19,
-      "learning_rate": 2.751629201522492e-05,
-      "loss": 1.7285,
-      "step": 708000
-    },
-    {
-      "epoch": 10.2,
-      "learning_rate": 2.746436527347219e-05,
-      "loss": 1.7308,
-      "step": 708500
-    },
-    {
-      "epoch": 10.21,
-      "learning_rate": 2.7412438531719453e-05,
-      "loss": 1.7294,
-      "step": 709000
-    },
-    {
-      "epoch": 10.21,
-      "learning_rate": 2.7360511789966715e-05,
-      "loss": 1.7349,
-      "step": 709500
-    },
-    {
-      "epoch": 10.22,
-      "learning_rate": 2.730858504821398e-05,
-      "loss": 1.7308,
-      "step": 710000
-    },
-    {
-      "epoch": 10.23,
-      "learning_rate": 2.7256658306461248e-05,
-      "loss": 1.7317,
-      "step": 710500
-    },
-    {
-      "epoch": 10.23,
-      "learning_rate": 2.7204731564708513e-05,
-      "loss": 1.7291,
-      "step": 711000
-    },
-    {
-      "epoch": 10.24,
-      "learning_rate": 2.7152908676439283e-05,
-      "loss": 1.7321,
-      "step": 711500
-    },
-    {
-      "epoch": 10.25,
-      "learning_rate": 2.7100981934686547e-05,
-      "loss": 1.73,
-      "step": 712000
-    },
-    {
-      "epoch": 10.26,
-      "learning_rate": 2.704905519293381e-05,
-      "loss": 1.7363,
-      "step": 712500
-    },
-    {
-      "epoch": 10.26,
-      "learning_rate": 2.6997128451181074e-05,
-      "loss": 1.7345,
-      "step": 713000
-    },
-    {
-      "epoch": 10.27,
-      "learning_rate": 2.6945201709428342e-05,
-      "loss": 1.7328,
-      "step": 713500
-    },
-    {
-      "epoch": 10.28,
-      "learning_rate": 2.6893274967675604e-05,
-      "loss": 1.7307,
-      "step": 714000
-    },
-    {
-      "epoch": 10.28,
-      "learning_rate": 2.684134822592287e-05,
-      "loss": 1.7372,
-      "step": 714500
-    },
-    {
-      "epoch": 10.29,
-      "learning_rate": 2.6789421484170137e-05,
-      "loss": 1.7352,
-      "step": 715000
-    },
-    {
-      "epoch": 10.3,
-      "learning_rate": 2.6737598595900903e-05,
-      "loss": 1.7272,
-      "step": 715500
-    },
-    {
-      "epoch": 10.31,
-      "learning_rate": 2.668567185414817e-05,
-      "loss": 1.7286,
-      "step": 716000
-    },
-    {
-      "epoch": 10.31,
-      "learning_rate": 2.6633745112395436e-05,
-      "loss": 1.7269,
-      "step": 716500
-    },
-    {
-      "epoch": 10.32,
-      "learning_rate": 2.6581818370642698e-05,
-      "loss": 1.7301,
-      "step": 717000
-    },
-    {
-      "epoch": 10.33,
-      "learning_rate": 2.652999548237347e-05,
-      "loss": 1.7297,
-      "step": 717500
-    },
-    {
-      "epoch": 10.33,
-      "learning_rate": 2.647817259410424e-05,
-      "loss": 1.7337,
-      "step": 718000
-    },
-    {
-      "epoch": 10.34,
-      "learning_rate": 2.6426245852351505e-05,
-      "loss": 1.7335,
-      "step": 718500
-    },
-    {
-      "epoch": 10.35,
-      "learning_rate": 2.6374319110598767e-05,
-      "loss": 1.7327,
-      "step": 719000
-    },
-    {
-      "epoch": 10.36,
-      "learning_rate": 2.6322392368846032e-05,
-      "loss": 1.7291,
-      "step": 719500
-    },
-    {
-      "epoch": 10.36,
-      "learning_rate": 2.62704656270933e-05,
-      "loss": 1.7325,
-      "step": 720000
-    },
-    {
-      "epoch": 10.37,
-      "learning_rate": 2.621853888534056e-05,
-      "loss": 1.7332,
-      "step": 720500
-    },
-    {
-      "epoch": 10.38,
-      "learning_rate": 2.6166612143587826e-05,
-      "loss": 1.7302,
-      "step": 721000
-    },
-    {
-      "epoch": 10.39,
-      "learning_rate": 2.61147892553186e-05,
-      "loss": 1.731,
-      "step": 721500
-    },
-    {
-      "epoch": 10.39,
-      "learning_rate": 2.606286251356586e-05,
-      "loss": 1.734,
-      "step": 722000
-    },
-    {
-      "epoch": 10.4,
-      "learning_rate": 2.6010935771813126e-05,
-      "loss": 1.7329,
-      "step": 722500
-    },
-    {
-      "epoch": 10.41,
-      "learning_rate": 2.5959009030060394e-05,
-      "loss": 1.7322,
-      "step": 723000
-    },
-    {
-      "epoch": 10.41,
-      "learning_rate": 2.590718614179116e-05,
-      "loss": 1.732,
-      "step": 723500
-    },
-    {
-      "epoch": 10.42,
-      "learning_rate": 2.585525940003843e-05,
-      "loss": 1.7299,
-      "step": 724000
-    },
-    {
-      "epoch": 10.43,
-      "learning_rate": 2.580333265828569e-05,
-      "loss": 1.7322,
-      "step": 724500
-    },
-    {
-      "epoch": 10.44,
-      "learning_rate": 2.5751405916532955e-05,
-      "loss": 1.7266,
-      "step": 725000
-    },
-    {
-      "epoch": 10.44,
-      "learning_rate": 2.5699479174780223e-05,
-      "loss": 1.7318,
-      "step": 725500
-    },
-    {
-      "epoch": 10.45,
-      "learning_rate": 2.564755243302749e-05,
-      "loss": 1.7294,
-      "step": 726000
-    },
-    {
-      "epoch": 10.46,
-      "learning_rate": 2.5595729544758258e-05,
-      "loss": 1.7306,
-      "step": 726500
-    },
-    {
-      "epoch": 10.46,
-      "learning_rate": 2.5543802803005523e-05,
-      "loss": 1.7302,
-      "step": 727000
-    },
-    {
-      "epoch": 10.47,
-      "learning_rate": 2.5491876061252784e-05,
-      "loss": 1.7315,
-      "step": 727500
-    },
-    {
-      "epoch": 10.48,
-      "learning_rate": 2.543994931950005e-05,
-      "loss": 1.7284,
-      "step": 728000
-    },
-    {
-      "epoch": 10.49,
-      "learning_rate": 2.5388022577747318e-05,
-      "loss": 1.7283,
-      "step": 728500
-    },
-    {
-      "epoch": 10.49,
-      "learning_rate": 2.533609583599458e-05,
-      "loss": 1.7341,
-      "step": 729000
-    },
-    {
-      "epoch": 10.5,
-      "learning_rate": 2.5284272947725352e-05,
-      "loss": 1.733,
-      "step": 729500
-    },
-    {
-      "epoch": 10.51,
-      "learning_rate": 2.5232346205972614e-05,
-      "loss": 1.7285,
-      "step": 730000
-    },
-    {
-      "epoch": 10.51,
-      "learning_rate": 2.518041946421988e-05,
-      "loss": 1.7285,
-      "step": 730500
-    },
-    {
-      "epoch": 10.52,
-      "learning_rate": 2.5128492722467143e-05,
-      "loss": 1.7273,
-      "step": 731000
-    },
-    {
-      "epoch": 10.53,
-      "learning_rate": 2.5076565980714412e-05,
-      "loss": 1.7291,
-      "step": 731500
-    },
-    {
-      "epoch": 10.54,
-      "learning_rate": 2.5024639238961673e-05,
-      "loss": 1.7309,
-      "step": 732000
-    },
-    {
-      "epoch": 10.54,
-      "learning_rate": 2.4972712497208938e-05,
-      "loss": 1.7282,
-      "step": 732500
-    },
-    {
-      "epoch": 10.55,
-      "learning_rate": 2.4920785755456203e-05,
-      "loss": 1.7303,
-      "step": 733000
-    },
-    {
-      "epoch": 10.56,
-      "learning_rate": 2.4868962867186973e-05,
-      "loss": 1.729,
-      "step": 733500
-    },
-    {
-      "epoch": 10.57,
-      "learning_rate": 2.481703612543424e-05,
-      "loss": 1.7329,
-      "step": 734000
-    },
-    {
-      "epoch": 10.57,
-      "learning_rate": 2.4765109383681502e-05,
-      "loss": 1.7288,
-      "step": 734500
-    },
-    {
-      "epoch": 10.58,
-      "learning_rate": 2.4713182641928767e-05,
-      "loss": 1.7302,
-      "step": 735000
-    },
-    {
-      "epoch": 10.59,
-      "learning_rate": 2.4661255900176032e-05,
-      "loss": 1.7321,
-      "step": 735500
-    },
-    {
-      "epoch": 10.59,
-      "learning_rate": 2.4609433011906802e-05,
-      "loss": 1.7292,
-      "step": 736000
-    },
-    {
-      "epoch": 10.6,
-      "learning_rate": 2.4557506270154067e-05,
-      "loss": 1.7265,
-      "step": 736500
-    },
-    {
-      "epoch": 10.61,
-      "learning_rate": 2.4505579528401332e-05,
-      "loss": 1.7326,
-      "step": 737000
-    },
-    {
-      "epoch": 10.62,
-      "learning_rate": 2.4453652786648597e-05,
-      "loss": 1.7307,
-      "step": 737500
-    },
-    {
-      "epoch": 10.62,
-      "learning_rate": 2.440172604489586e-05,
-      "loss": 1.7275,
-      "step": 738000
-    },
-    {
-      "epoch": 10.63,
-      "learning_rate": 2.434979930314313e-05,
-      "loss": 1.7292,
-      "step": 738500
-    },
-    {
-      "epoch": 10.64,
-      "learning_rate": 2.4297976414873896e-05,
-      "loss": 1.7261,
-      "step": 739000
-    },
-    {
-      "epoch": 10.64,
-      "learning_rate": 2.4246049673121164e-05,
-      "loss": 1.7256,
-      "step": 739500
-    },
-    {
-      "epoch": 10.65,
-      "learning_rate": 2.4194122931368426e-05,
-      "loss": 1.7286,
-      "step": 740000
-    },
-    {
-      "epoch": 10.66,
-      "learning_rate": 2.414219618961569e-05,
-      "loss": 1.7312,
-      "step": 740500
-    },
-    {
-      "epoch": 10.67,
-      "learning_rate": 2.4090269447862956e-05,
-      "loss": 1.7292,
-      "step": 741000
-    },
-    {
-      "epoch": 10.67,
-      "learning_rate": 2.403844655959373e-05,
-      "loss": 1.725,
-      "step": 741500
-    },
-    {
-      "epoch": 10.68,
-      "learning_rate": 2.398651981784099e-05,
-      "loss": 1.7271,
-      "step": 742000
-    },
-    {
-      "epoch": 10.69,
-      "learning_rate": 2.3934593076088255e-05,
-      "loss": 1.7308,
-      "step": 742500
-    },
-    {
-      "epoch": 10.69,
-      "learning_rate": 2.388266633433552e-05,
-      "loss": 1.7253,
-      "step": 743000
-    },
-    {
-      "epoch": 10.7,
-      "learning_rate": 2.3830843446066293e-05,
-      "loss": 1.7264,
-      "step": 743500
-    },
-    {
-      "epoch": 10.71,
-      "learning_rate": 2.3778916704313555e-05,
-      "loss": 1.7239,
-      "step": 744000
-    },
-    {
-      "epoch": 10.72,
-      "learning_rate": 2.372698996256082e-05,
-      "loss": 1.7291,
-      "step": 744500
-    },
-    {
-      "epoch": 10.72,
-      "learning_rate": 2.3675063220808084e-05,
-      "loss": 1.7281,
-      "step": 745000
-    },
-    {
-      "epoch": 10.73,
-      "learning_rate": 2.362313647905535e-05,
-      "loss": 1.7274,
-      "step": 745500
-    },
-    {
-      "epoch": 10.74,
-      "learning_rate": 2.357131359078612e-05,
-      "loss": 1.7242,
-      "step": 746000
-    },
-    {
-      "epoch": 10.75,
-      "learning_rate": 2.3519386849033384e-05,
-      "loss": 1.7232,
-      "step": 746500
-    },
-    {
-      "epoch": 10.75,
-      "learning_rate": 2.3467460107280652e-05,
-      "loss": 1.7272,
-      "step": 747000
-    },
-    {
-      "epoch": 10.76,
-      "learning_rate": 2.3415533365527914e-05,
-      "loss": 1.7296,
-      "step": 747500
-    },
-    {
-      "epoch": 10.77,
-      "learning_rate": 2.336360662377518e-05,
-      "loss": 1.7294,
-      "step": 748000
-    },
-    {
-      "epoch": 10.77,
-      "learning_rate": 2.3311679882022443e-05,
-      "loss": 1.7216,
-      "step": 748500
-    },
-    {
-      "epoch": 10.78,
-      "learning_rate": 2.325975314026971e-05,
-      "loss": 1.723,
-      "step": 749000
-    },
-    {
-      "epoch": 10.79,
-      "learning_rate": 2.3207930252000478e-05,
-      "loss": 1.7262,
-      "step": 749500
-    },
-    {
-      "epoch": 10.8,
-      "learning_rate": 2.3156003510247743e-05,
-      "loss": 1.7261,
-      "step": 750000
-    },
-    {
-      "epoch": 10.8,
-      "learning_rate": 2.3104076768495008e-05,
-      "loss": 1.7268,
-      "step": 750500
-    },
-    {
-      "epoch": 10.81,
-      "learning_rate": 2.3052150026742273e-05,
-      "loss": 1.7271,
-      "step": 751000
-    },
-    {
-      "epoch": 10.82,
-      "learning_rate": 2.3000223284989538e-05,
-      "loss": 1.7255,
-      "step": 751500
-    },
-    {
-      "epoch": 10.82,
-      "learning_rate": 2.2948400396720307e-05,
-      "loss": 1.7262,
-      "step": 752000
-    },
-    {
-      "epoch": 10.83,
-      "learning_rate": 2.2896473654967572e-05,
-      "loss": 1.7263,
-      "step": 752500
-    },
-    {
-      "epoch": 10.84,
-      "learning_rate": 2.2844546913214837e-05,
-      "loss": 1.7283,
-      "step": 753000
-    },
-    {
-      "epoch": 10.85,
-      "learning_rate": 2.2792620171462102e-05,
-      "loss": 1.7261,
-      "step": 753500
-    },
-    {
-      "epoch": 10.85,
-      "learning_rate": 2.2740693429709367e-05,
-      "loss": 1.7245,
-      "step": 754000
-    },
-    {
-      "epoch": 10.86,
-      "learning_rate": 2.2688870541440136e-05,
-      "loss": 1.7284,
-      "step": 754500
-    },
-    {
-      "epoch": 10.87,
-      "learning_rate": 2.26369437996874e-05,
-      "loss": 1.7266,
-      "step": 755000
-    },
-    {
-      "epoch": 10.87,
-      "learning_rate": 2.258501705793467e-05,
-      "loss": 1.7276,
-      "step": 755500
-    },
-    {
-      "epoch": 10.88,
-      "learning_rate": 2.253309031618193e-05,
-      "loss": 1.7257,
-      "step": 756000
-    },
-    {
-      "epoch": 10.89,
-      "learning_rate": 2.2481267427912704e-05,
-      "loss": 1.7238,
-      "step": 756500
-    },
-    {
-      "epoch": 10.9,
-      "learning_rate": 2.2429340686159966e-05,
-      "loss": 1.7261,
-      "step": 757000
-    },
-    {
-      "epoch": 10.9,
-      "learning_rate": 2.237741394440723e-05,
-      "loss": 1.7229,
-      "step": 757500
-    },
-    {
-      "epoch": 10.91,
-      "learning_rate": 2.2325487202654495e-05,
-      "loss": 1.724,
-      "step": 758000
-    },
-    {
-      "epoch": 10.92,
-      "learning_rate": 2.227356046090176e-05,
-      "loss": 1.7255,
-      "step": 758500
-    },
-    {
-      "epoch": 10.93,
-      "learning_rate": 2.2221633719149025e-05,
-      "loss": 1.7255,
-      "step": 759000
-    },
-    {
-      "epoch": 10.93,
-      "learning_rate": 2.216970697739629e-05,
-      "loss": 1.7289,
-      "step": 759500
-    },
-    {
-      "epoch": 10.94,
-      "learning_rate": 2.211788408912706e-05,
-      "loss": 1.7237,
-      "step": 760000
-    },
-    {
-      "epoch": 10.95,
-      "learning_rate": 2.2065957347374325e-05,
-      "loss": 1.7219,
-      "step": 760500
-    },
-    {
-      "epoch": 10.95,
-      "learning_rate": 2.201403060562159e-05,
-      "loss": 1.7243,
-      "step": 761000
-    },
-    {
-      "epoch": 10.96,
-      "learning_rate": 2.1962103863868855e-05,
-      "loss": 1.728,
-      "step": 761500
-    },
-    {
-      "epoch": 10.97,
-      "learning_rate": 2.191017712211612e-05,
-      "loss": 1.7213,
-      "step": 762000
-    },
-    {
-      "epoch": 10.98,
-      "learning_rate": 2.1858250380363384e-05,
-      "loss": 1.7196,
-      "step": 762500
-    },
-    {
-      "epoch": 10.98,
-      "learning_rate": 2.1806427492094157e-05,
-      "loss": 1.7258,
-      "step": 763000
-    },
-    {
-      "epoch": 10.99,
-      "learning_rate": 2.175450075034142e-05,
-      "loss": 1.7255,
-      "step": 763500
-    },
-    {
-      "epoch": 11.0,
-      "learning_rate": 2.1702574008588684e-05,
-      "loss": 1.7229,
-      "step": 764000
-    },
-    {
-      "epoch": 11.0,
-      "eval_accuracy": 0.6645491401062097,
-      "eval_loss": 1.5919610261917114,
-      "eval_runtime": 1300.0829,
-      "eval_samples_per_second": 414.54,
-      "eval_steps_per_second": 25.909,
-      "step": 764203
-    },
-    {
-      "epoch": 11.0,
-      "learning_rate": 2.165064726683595e-05,
-      "loss": 1.7248,
-      "step": 764500
-    },
-    {
-      "epoch": 11.01,
-      "learning_rate": 2.1598824378566722e-05,
-      "loss": 1.7201,
-      "step": 765000
-    },
-    {
-      "epoch": 11.02,
-      "learning_rate": 2.1546897636813983e-05,
-      "loss": 1.7237,
-      "step": 765500
-    },
-    {
-      "epoch": 11.03,
-      "learning_rate": 2.1494970895061248e-05,
-      "loss": 1.724,
-      "step": 766000
-    },
-    {
-      "epoch": 11.03,
-      "learning_rate": 2.1443044153308513e-05,
-      "loss": 1.7209,
-      "step": 766500
-    },
-    {
-      "epoch": 11.04,
-      "learning_rate": 2.1391117411555778e-05,
-      "loss": 1.7219,
-      "step": 767000
-    },
-    {
-      "epoch": 11.05,
-      "learning_rate": 2.1339190669803043e-05,
-      "loss": 1.7215,
-      "step": 767500
-    },
-    {
-      "epoch": 11.05,
-      "learning_rate": 2.1287263928050308e-05,
-      "loss": 1.7206,
-      "step": 768000
-    },
-    {
-      "epoch": 11.06,
-      "learning_rate": 2.1235441039781077e-05,
-      "loss": 1.7198,
-      "step": 768500
-    },
-    {
-      "epoch": 11.07,
-      "learning_rate": 2.1183514298028342e-05,
-      "loss": 1.7244,
-      "step": 769000
-    },
-    {
-      "epoch": 11.08,
-      "learning_rate": 2.1131587556275607e-05,
-      "loss": 1.7213,
-      "step": 769500
-    },
-    {
-      "epoch": 11.08,
-      "learning_rate": 2.1079660814522872e-05,
-      "loss": 1.7236,
-      "step": 770000
-    },
-    {
-      "epoch": 11.09,
-      "learning_rate": 2.1027837926253645e-05,
-      "loss": 1.721,
-      "step": 770500
-    },
-    {
-      "epoch": 11.1,
-      "learning_rate": 2.0975911184500907e-05,
-      "loss": 1.7223,
-      "step": 771000
-    },
-    {
-      "epoch": 11.11,
-      "learning_rate": 2.092398444274817e-05,
-      "loss": 1.7238,
-      "step": 771500
-    },
-    {
-      "epoch": 11.11,
-      "learning_rate": 2.0872057700995436e-05,
-      "loss": 1.7226,
-      "step": 772000
-    },
-    {
-      "epoch": 11.12,
-      "learning_rate": 2.08201309592427e-05,
-      "loss": 1.7211,
-      "step": 772500
-    },
-    {
-      "epoch": 11.13,
-      "learning_rate": 2.0768204217489966e-05,
-      "loss": 1.7223,
-      "step": 773000
-    },
-    {
-      "epoch": 11.13,
-      "learning_rate": 2.0716381329220736e-05,
-      "loss": 1.7184,
-      "step": 773500
-    },
-    {
-      "epoch": 11.14,
-      "learning_rate": 2.0664454587468e-05,
-      "loss": 1.723,
-      "step": 774000
-    },
-    {
-      "epoch": 11.15,
-      "learning_rate": 2.0612527845715266e-05,
-      "loss": 1.7148,
-      "step": 774500
-    },
-    {
-      "epoch": 11.16,
-      "learning_rate": 2.056060110396253e-05,
-      "loss": 1.7249,
-      "step": 775000
-    },
-    {
-      "epoch": 11.16,
-      "learning_rate": 2.05087782156933e-05,
-      "loss": 1.7156,
-      "step": 775500
-    },
-    {
-      "epoch": 11.17,
-      "learning_rate": 2.0456851473940565e-05,
-      "loss": 1.7209,
-      "step": 776000
-    },
-    {
-      "epoch": 11.18,
-      "learning_rate": 2.040492473218783e-05,
-      "loss": 1.7263,
-      "step": 776500
-    },
-    {
-      "epoch": 11.18,
-      "learning_rate": 2.0352997990435095e-05,
-      "loss": 1.7217,
-      "step": 777000
-    },
-    {
-      "epoch": 11.19,
-      "learning_rate": 2.0301175102165865e-05,
-      "loss": 1.7213,
-      "step": 777500
-    },
-    {
-      "epoch": 11.2,
-      "learning_rate": 2.024924836041313e-05,
-      "loss": 1.7238,
-      "step": 778000
-    },
-    {
-      "epoch": 11.21,
-      "learning_rate": 2.0197321618660394e-05,
-      "loss": 1.7207,
-      "step": 778500
-    },
-    {
-      "epoch": 11.21,
-      "learning_rate": 2.014539487690766e-05,
-      "loss": 1.7236,
-      "step": 779000
-    },
-    {
-      "epoch": 11.22,
-      "learning_rate": 2.0093468135154924e-05,
-      "loss": 1.7216,
-      "step": 779500
-    },
-    {
-      "epoch": 11.23,
-      "learning_rate": 2.0041645246885697e-05,
-      "loss": 1.7229,
-      "step": 780000
-    },
-    {
-      "epoch": 11.23,
-      "learning_rate": 1.998971850513296e-05,
-      "loss": 1.7196,
-      "step": 780500
-    },
-    {
-      "epoch": 11.24,
-      "learning_rate": 1.9937791763380224e-05,
-      "loss": 1.7199,
-      "step": 781000
-    },
-    {
-      "epoch": 11.25,
-      "learning_rate": 1.988586502162749e-05,
-      "loss": 1.7186,
-      "step": 781500
-    },
-    {
-      "epoch": 11.26,
-      "learning_rate": 1.9833938279874753e-05,
-      "loss": 1.7208,
-      "step": 782000
-    },
-    {
-      "epoch": 11.26,
-      "learning_rate": 1.9782011538122018e-05,
-      "loss": 1.7177,
-      "step": 782500
-    },
-    {
-      "epoch": 11.27,
-      "learning_rate": 1.9730084796369283e-05,
-      "loss": 1.7212,
-      "step": 783000
-    },
-    {
-      "epoch": 11.28,
-      "learning_rate": 1.9678158054616548e-05,
-      "loss": 1.7219,
-      "step": 783500
-    },
-    {
-      "epoch": 11.28,
-      "learning_rate": 1.9626335166347318e-05,
-      "loss": 1.7191,
-      "step": 784000
-    },
-    {
-      "epoch": 11.29,
-      "learning_rate": 1.9574512278078087e-05,
-      "loss": 1.7213,
-      "step": 784500
-    },
-    {
-      "epoch": 11.3,
-      "learning_rate": 1.9522585536325352e-05,
-      "loss": 1.7227,
-      "step": 785000
-    },
-    {
-      "epoch": 11.31,
-      "learning_rate": 1.9470658794572617e-05,
-      "loss": 1.7204,
-      "step": 785500
-    },
-    {
-      "epoch": 11.31,
-      "learning_rate": 1.9418732052819882e-05,
-      "loss": 1.7223,
-      "step": 786000
-    },
-    {
-      "epoch": 11.32,
-      "learning_rate": 1.9366805311067147e-05,
-      "loss": 1.7194,
-      "step": 786500
-    },
-    {
-      "epoch": 11.33,
-      "learning_rate": 1.9314878569314412e-05,
-      "loss": 1.7205,
-      "step": 787000
-    },
-    {
-      "epoch": 11.34,
-      "learning_rate": 1.9262951827561677e-05,
-      "loss": 1.7195,
-      "step": 787500
-    },
-    {
-      "epoch": 11.34,
-      "learning_rate": 1.9211128939292446e-05,
-      "loss": 1.7208,
-      "step": 788000
-    },
-    {
-      "epoch": 11.35,
-      "learning_rate": 1.915920219753971e-05,
-      "loss": 1.7161,
-      "step": 788500
-    },
-    {
-      "epoch": 11.36,
-      "learning_rate": 1.9107275455786976e-05,
-      "loss": 1.7181,
-      "step": 789000
-    },
-    {
-      "epoch": 11.36,
-      "learning_rate": 1.905534871403424e-05,
-      "loss": 1.7177,
-      "step": 789500
-    },
-    {
-      "epoch": 11.37,
-      "learning_rate": 1.9003421972281506e-05,
-      "loss": 1.7162,
-      "step": 790000
-    },
-    {
-      "epoch": 11.38,
-      "learning_rate": 1.8951599084012276e-05,
-      "loss": 1.717,
-      "step": 790500
-    },
-    {
-      "epoch": 11.39,
-      "learning_rate": 1.889967234225954e-05,
-      "loss": 1.7229,
-      "step": 791000
-    },
-    {
-      "epoch": 11.39,
-      "learning_rate": 1.8847745600506805e-05,
-      "loss": 1.7197,
-      "step": 791500
-    },
-    {
-      "epoch": 11.4,
-      "learning_rate": 1.879581885875407e-05,
-      "loss": 1.7194,
-      "step": 792000
-    },
-    {
-      "epoch": 11.41,
-      "learning_rate": 1.8743892117001335e-05,
-      "loss": 1.7226,
-      "step": 792500
-    },
-    {
-      "epoch": 11.41,
-      "learning_rate": 1.8692069228732105e-05,
-      "loss": 1.7229,
-      "step": 793000
-    },
-    {
-      "epoch": 11.42,
-      "learning_rate": 1.864014248697937e-05,
-      "loss": 1.7167,
-      "step": 793500
-    },
-    {
-      "epoch": 11.43,
-      "learning_rate": 1.8588215745226638e-05,
-      "loss": 1.7203,
-      "step": 794000
-    },
-    {
-      "epoch": 11.44,
-      "learning_rate": 1.85362890034739e-05,
-      "loss": 1.723,
-      "step": 794500
-    },
-    {
-      "epoch": 11.44,
-      "learning_rate": 1.8484362261721164e-05,
-      "loss": 1.7216,
-      "step": 795000
-    },
-    {
-      "epoch": 11.45,
-      "learning_rate": 1.8432539373451934e-05,
-      "loss": 1.7197,
-      "step": 795500
-    },
-    {
-      "epoch": 11.46,
-      "learning_rate": 1.83806126316992e-05,
-      "loss": 1.7171,
-      "step": 796000
-    },
-    {
-      "epoch": 11.46,
-      "learning_rate": 1.8328685889946464e-05,
-      "loss": 1.7186,
-      "step": 796500
-    },
-    {
-      "epoch": 11.47,
-      "learning_rate": 1.827675914819373e-05,
-      "loss": 1.7188,
-      "step": 797000
-    },
-    {
-      "epoch": 11.48,
-      "learning_rate": 1.82249362599245e-05,
-      "loss": 1.7182,
-      "step": 797500
-    },
-    {
-      "epoch": 11.49,
-      "learning_rate": 1.8173009518171763e-05,
-      "loss": 1.7204,
-      "step": 798000
-    },
-    {
-      "epoch": 11.49,
-      "learning_rate": 1.8121082776419028e-05,
-      "loss": 1.7228,
-      "step": 798500
-    },
-    {
-      "epoch": 11.5,
-      "learning_rate": 1.8069156034666293e-05,
-      "loss": 1.7191,
-      "step": 799000
-    },
-    {
-      "epoch": 11.51,
-      "learning_rate": 1.8017229292913558e-05,
-      "loss": 1.7188,
-      "step": 799500
-    },
-    {
-      "epoch": 11.52,
-      "learning_rate": 1.7965406404644328e-05,
-      "loss": 1.7207,
-      "step": 800000
-    },
-    {
-      "epoch": 11.52,
-      "learning_rate": 1.7913479662891593e-05,
-      "loss": 1.7174,
-      "step": 800500
-    },
-    {
-      "epoch": 11.53,
-      "learning_rate": 1.7861552921138858e-05,
-      "loss": 1.7187,
-      "step": 801000
-    },
-    {
-      "epoch": 11.54,
-      "learning_rate": 1.7809626179386122e-05,
-      "loss": 1.7177,
-      "step": 801500
-    },
-    {
-      "epoch": 11.54,
-      "learning_rate": 1.7757699437633387e-05,
-      "loss": 1.7169,
-      "step": 802000
-    },
-    {
-      "epoch": 11.55,
-      "learning_rate": 1.770587654936416e-05,
-      "loss": 1.7187,
-      "step": 802500
-    },
-    {
-      "epoch": 11.56,
-      "learning_rate": 1.7653949807611422e-05,
-      "loss": 1.7142,
-      "step": 803000
-    },
-    {
-      "epoch": 11.57,
-      "learning_rate": 1.760202306585869e-05,
-      "loss": 1.7166,
-      "step": 803500
-    },
-    {
-      "epoch": 11.57,
-      "learning_rate": 1.755009632410595e-05,
-      "loss": 1.7169,
-      "step": 804000
-    },
-    {
-      "epoch": 11.58,
-      "learning_rate": 1.7498273435836725e-05,
-      "loss": 1.7156,
-      "step": 804500
-    },
-    {
-      "epoch": 11.59,
-      "learning_rate": 1.7446346694083986e-05,
-      "loss": 1.7156,
-      "step": 805000
-    },
-    {
-      "epoch": 11.59,
-      "learning_rate": 1.739441995233125e-05,
-      "loss": 1.7159,
-      "step": 805500
-    },
-    {
-      "epoch": 11.6,
-      "learning_rate": 1.7342493210578516e-05,
-      "loss": 1.717,
-      "step": 806000
-    },
-    {
-      "epoch": 11.61,
-      "learning_rate": 1.729056646882578e-05,
-      "loss": 1.716,
-      "step": 806500
-    },
-    {
-      "epoch": 11.62,
-      "learning_rate": 1.723874358055655e-05,
-      "loss": 1.7217,
-      "step": 807000
-    },
-    {
-      "epoch": 11.62,
-      "learning_rate": 1.7186816838803815e-05,
-      "loss": 1.7168,
-      "step": 807500
-    },
-    {
-      "epoch": 11.63,
-      "learning_rate": 1.713489009705108e-05,
-      "loss": 1.7147,
-      "step": 808000
-    },
-    {
-      "epoch": 11.64,
-      "learning_rate": 1.7082963355298345e-05,
-      "loss": 1.7156,
-      "step": 808500
-    },
-    {
-      "epoch": 11.64,
-      "learning_rate": 1.7031140467029115e-05,
-      "loss": 1.7187,
-      "step": 809000
-    },
-    {
-      "epoch": 11.65,
-      "learning_rate": 1.697921372527638e-05,
-      "loss": 1.7168,
-      "step": 809500
-    },
-    {
-      "epoch": 11.66,
-      "learning_rate": 1.6927286983523648e-05,
-      "loss": 1.7184,
-      "step": 810000
-    },
-    {
-      "epoch": 11.67,
-      "learning_rate": 1.687536024177091e-05,
-      "loss": 1.7142,
-      "step": 810500
-    },
-    {
-      "epoch": 11.67,
-      "learning_rate": 1.6823433500018178e-05,
-      "loss": 1.7171,
-      "step": 811000
-    },
-    {
-      "epoch": 11.68,
-      "learning_rate": 1.6771610611748944e-05,
-      "loss": 1.7163,
-      "step": 811500
-    },
-    {
-      "epoch": 11.69,
-      "learning_rate": 1.6719683869996212e-05,
-      "loss": 1.7208,
-      "step": 812000
-    },
-    {
-      "epoch": 11.7,
-      "learning_rate": 1.6667757128243474e-05,
-      "loss": 1.714,
-      "step": 812500
-    },
-    {
-      "epoch": 11.7,
-      "learning_rate": 1.661583038649074e-05,
-      "loss": 1.7158,
-      "step": 813000
-    },
-    {
-      "epoch": 11.71,
-      "learning_rate": 1.656400749822151e-05,
-      "loss": 1.7167,
-      "step": 813500
-    },
-    {
-      "epoch": 11.72,
-      "learning_rate": 1.6512080756468777e-05,
-      "loss": 1.7153,
-      "step": 814000
-    },
-    {
-      "epoch": 11.72,
-      "learning_rate": 1.6460154014716038e-05,
-      "loss": 1.7164,
-      "step": 814500
-    },
-    {
-      "epoch": 11.73,
-      "learning_rate": 1.6408227272963303e-05,
-      "loss": 1.7186,
-      "step": 815000
-    },
-    {
-      "epoch": 11.74,
-      "learning_rate": 1.6356404384694073e-05,
-      "loss": 1.7179,
-      "step": 815500
-    },
-    {
-      "epoch": 11.75,
-      "learning_rate": 1.630447764294134e-05,
-      "loss": 1.7159,
-      "step": 816000
-    },
-    {
-      "epoch": 11.75,
-      "learning_rate": 1.6252550901188603e-05,
-      "loss": 1.7159,
-      "step": 816500
-    },
-    {
-      "epoch": 11.76,
-      "learning_rate": 1.6200624159435868e-05,
-      "loss": 1.7179,
-      "step": 817000
-    },
-    {
-      "epoch": 11.77,
-      "learning_rate": 1.6148801271166637e-05,
-      "loss": 1.7141,
-      "step": 817500
-    },
-    {
-      "epoch": 11.77,
-      "learning_rate": 1.6096874529413902e-05,
-      "loss": 1.7203,
-      "step": 818000
-    },
-    {
-      "epoch": 11.78,
-      "learning_rate": 1.604494778766117e-05,
-      "loss": 1.7158,
-      "step": 818500
-    },
-    {
-      "epoch": 11.79,
-      "learning_rate": 1.5993021045908432e-05,
-      "loss": 1.7126,
-      "step": 819000
-    },
-    {
-      "epoch": 11.8,
-      "learning_rate": 1.59410943041557e-05,
-      "loss": 1.7146,
-      "step": 819500
-    },
-    {
-      "epoch": 11.8,
-      "learning_rate": 1.5889271415886466e-05,
-      "loss": 1.713,
-      "step": 820000
-    },
-    {
-      "epoch": 11.81,
-      "learning_rate": 1.5837344674133735e-05,
-      "loss": 1.715,
-      "step": 820500
-    },
-    {
-      "epoch": 11.82,
-      "learning_rate": 1.5785417932380996e-05,
-      "loss": 1.7166,
-      "step": 821000
-    },
-    {
-      "epoch": 11.82,
-      "learning_rate": 1.5733491190628264e-05,
-      "loss": 1.7166,
-      "step": 821500
-    },
-    {
-      "epoch": 11.83,
-      "learning_rate": 1.5681564448875526e-05,
-      "loss": 1.7136,
-      "step": 822000
-    },
-    {
-      "epoch": 11.84,
-      "learning_rate": 1.562963770712279e-05,
-      "loss": 1.7152,
-      "step": 822500
-    },
-    {
-      "epoch": 11.85,
-      "learning_rate": 1.557781481885356e-05,
-      "loss": 1.7161,
-      "step": 823000
-    },
-    {
-      "epoch": 11.85,
-      "learning_rate": 1.552588807710083e-05,
-      "loss": 1.7157,
-      "step": 823500
-    },
-    {
-      "epoch": 11.86,
-      "learning_rate": 1.547396133534809e-05,
-      "loss": 1.7162,
-      "step": 824000
-    },
-    {
-      "epoch": 11.87,
-      "learning_rate": 1.5422034593595355e-05,
-      "loss": 1.717,
-      "step": 824500
-    },
-    {
-      "epoch": 11.88,
-      "learning_rate": 1.5370211705326125e-05,
-      "loss": 1.7182,
-      "step": 825000
-    },
-    {
-      "epoch": 11.88,
-      "learning_rate": 1.5318284963573393e-05,
-      "loss": 1.7174,
-      "step": 825500
-    },
-    {
-      "epoch": 11.89,
-      "learning_rate": 1.5266358221820655e-05,
-      "loss": 1.7126,
-      "step": 826000
-    },
-    {
-      "epoch": 11.9,
-      "learning_rate": 1.521443148006792e-05,
-      "loss": 1.7112,
-      "step": 826500
-    },
-    {
-      "epoch": 11.9,
-      "learning_rate": 1.5162608591798691e-05,
-      "loss": 1.7143,
-      "step": 827000
-    },
-    {
-      "epoch": 11.91,
-      "learning_rate": 1.5110681850045954e-05,
-      "loss": 1.7127,
-      "step": 827500
-    },
-    {
-      "epoch": 11.92,
-      "learning_rate": 1.505875510829322e-05,
-      "loss": 1.72,
-      "step": 828000
-    },
-    {
-      "epoch": 11.93,
-      "learning_rate": 1.5006828366540484e-05,
-      "loss": 1.7146,
-      "step": 828500
-    },
-    {
-      "epoch": 11.93,
-      "learning_rate": 1.4955005478271255e-05,
-      "loss": 1.7156,
-      "step": 829000
-    },
-    {
-      "epoch": 11.94,
-      "learning_rate": 1.4903078736518518e-05,
-      "loss": 1.7144,
-      "step": 829500
-    },
-    {
-      "epoch": 11.95,
-      "learning_rate": 1.4851151994765785e-05,
-      "loss": 1.7136,
-      "step": 830000
-    },
-    {
-      "epoch": 11.95,
-      "learning_rate": 1.479922525301305e-05,
-      "loss": 1.713,
-      "step": 830500
-    },
-    {
-      "epoch": 11.96,
-      "learning_rate": 1.474740236474382e-05,
-      "loss": 1.7165,
-      "step": 831000
-    },
-    {
-      "epoch": 11.97,
-      "learning_rate": 1.4695475622991084e-05,
-      "loss": 1.7137,
-      "step": 831500
-    },
-    {
-      "epoch": 11.98,
-      "learning_rate": 1.4643548881238351e-05,
-      "loss": 1.7115,
-      "step": 832000
-    },
-    {
-      "epoch": 11.98,
-      "learning_rate": 1.4591622139485614e-05,
-      "loss": 1.7183,
-      "step": 832500
-    },
-    {
-      "epoch": 11.99,
-      "learning_rate": 1.4539695397732881e-05,
-      "loss": 1.7142,
-      "step": 833000
-    },
-    {
-      "epoch": 12.0,
-      "learning_rate": 1.4487872509463649e-05,
-      "loss": 1.7127,
-      "step": 833500
-    },
-    {
-      "epoch": 12.0,
-      "eval_accuracy": 0.6658040615937678,
-      "eval_loss": 1.5836162567138672,
-      "eval_runtime": 1303.2983,
-      "eval_samples_per_second": 413.518,
-      "eval_steps_per_second": 25.845,
-      "step": 833676
-    },
-    {
-      "epoch": 12.0,
-      "learning_rate": 1.4435945767710915e-05,
-      "loss": 1.7129,
-      "step": 834000
-    },
-    {
-      "epoch": 12.01,
-      "learning_rate": 1.4384019025958179e-05,
-      "loss": 1.7078,
-      "step": 834500
-    },
-    {
-      "epoch": 12.02,
-      "learning_rate": 1.4332092284205445e-05,
-      "loss": 1.7095,
-      "step": 835000
-    },
-    {
-      "epoch": 12.03,
-      "learning_rate": 1.4280165542452708e-05,
-      "loss": 1.7136,
-      "step": 835500
-    },
-    {
-      "epoch": 12.03,
-      "learning_rate": 1.422834265418348e-05,
-      "loss": 1.7119,
-      "step": 836000
-    },
-    {
-      "epoch": 12.04,
-      "learning_rate": 1.4176415912430743e-05,
-      "loss": 1.7085,
-      "step": 836500
-    },
-    {
-      "epoch": 12.05,
-      "learning_rate": 1.4124489170678006e-05,
-      "loss": 1.7114,
-      "step": 837000
-    },
-    {
-      "epoch": 12.06,
-      "learning_rate": 1.4072562428925273e-05,
-      "loss": 1.7107,
-      "step": 837500
-    },
-    {
-      "epoch": 12.06,
-      "learning_rate": 1.4020739540656044e-05,
-      "loss": 1.7106,
-      "step": 838000
-    },
-    {
-      "epoch": 12.07,
-      "learning_rate": 1.3968812798903307e-05,
-      "loss": 1.7099,
-      "step": 838500
-    },
-    {
-      "epoch": 12.08,
-      "learning_rate": 1.3916886057150572e-05,
-      "loss": 1.7152,
-      "step": 839000
-    },
-    {
-      "epoch": 12.08,
-      "learning_rate": 1.3864959315397839e-05,
-      "loss": 1.7166,
-      "step": 839500
-    },
-    {
-      "epoch": 12.09,
-      "learning_rate": 1.3813136427128608e-05,
-      "loss": 1.7123,
-      "step": 840000
-    },
-    {
-      "epoch": 12.1,
-      "learning_rate": 1.3761209685375873e-05,
-      "loss": 1.7127,
-      "step": 840500
-    },
-    {
-      "epoch": 12.11,
-      "learning_rate": 1.3709282943623137e-05,
-      "loss": 1.7144,
-      "step": 841000
-    },
-    {
-      "epoch": 12.11,
-      "learning_rate": 1.3657356201870403e-05,
-      "loss": 1.7121,
-      "step": 841500
-    },
-    {
-      "epoch": 12.12,
-      "learning_rate": 1.3605429460117666e-05,
-      "loss": 1.714,
-      "step": 842000
-    },
-    {
-      "epoch": 12.13,
-      "learning_rate": 1.3553502718364933e-05,
-      "loss": 1.7112,
-      "step": 842500
-    },
-    {
-      "epoch": 12.13,
-      "learning_rate": 1.3501679830095701e-05,
-      "loss": 1.7121,
-      "step": 843000
-    },
-    {
-      "epoch": 12.14,
-      "learning_rate": 1.3449753088342967e-05,
-      "loss": 1.706,
-      "step": 843500
-    },
-    {
-      "epoch": 12.15,
-      "learning_rate": 1.339782634659023e-05,
-      "loss": 1.7121,
-      "step": 844000
-    },
-    {
-      "epoch": 12.16,
-      "learning_rate": 1.3345899604837494e-05,
-      "loss": 1.7075,
-      "step": 844500
-    },
-    {
-      "epoch": 12.16,
-      "learning_rate": 1.3294076716568265e-05,
-      "loss": 1.7105,
-      "step": 845000
-    },
-    {
-      "epoch": 12.17,
-      "learning_rate": 1.3242149974815532e-05,
-      "loss": 1.7086,
-      "step": 845500
-    },
-    {
-      "epoch": 12.18,
-      "learning_rate": 1.3190223233062795e-05,
-      "loss": 1.7103,
-      "step": 846000
-    },
-    {
-      "epoch": 12.18,
-      "learning_rate": 1.313829649131006e-05,
-      "loss": 1.7082,
-      "step": 846500
-    },
-    {
-      "epoch": 12.19,
-      "learning_rate": 1.3086369749557325e-05,
-      "loss": 1.7101,
-      "step": 847000
-    },
-    {
-      "epoch": 12.2,
-      "learning_rate": 1.3034546861288096e-05,
-      "loss": 1.7145,
-      "step": 847500
-    },
-    {
-      "epoch": 12.21,
-      "learning_rate": 1.2982620119535361e-05,
-      "loss": 1.7118,
-      "step": 848000
-    },
-    {
-      "epoch": 12.21,
-      "learning_rate": 1.2930693377782624e-05,
-      "loss": 1.7077,
-      "step": 848500
-    },
-    {
-      "epoch": 12.22,
-      "learning_rate": 1.2878766636029891e-05,
-      "loss": 1.7097,
-      "step": 849000
-    },
-    {
-      "epoch": 12.23,
-      "learning_rate": 1.2826943747760659e-05,
-      "loss": 1.7084,
-      "step": 849500
-    },
-    {
-      "epoch": 12.23,
-      "learning_rate": 1.2775017006007925e-05,
-      "loss": 1.7086,
-      "step": 850000
-    },
-    {
-      "epoch": 12.24,
-      "learning_rate": 1.2723090264255189e-05,
-      "loss": 1.7108,
-      "step": 850500
-    },
-    {
-      "epoch": 12.25,
-      "learning_rate": 1.2671163522502455e-05,
-      "loss": 1.7097,
-      "step": 851000
-    },
-    {
-      "epoch": 12.26,
-      "learning_rate": 1.2619340634233223e-05,
-      "loss": 1.7124,
-      "step": 851500
-    },
-    {
-      "epoch": 12.26,
-      "learning_rate": 1.256741389248049e-05,
-      "loss": 1.7127,
-      "step": 852000
-    },
-    {
-      "epoch": 12.27,
-      "learning_rate": 1.2515487150727753e-05,
-      "loss": 1.7104,
-      "step": 852500
-    },
-    {
-      "epoch": 12.28,
-      "learning_rate": 1.2463560408975018e-05,
-      "loss": 1.7132,
-      "step": 853000
-    },
-    {
-      "epoch": 12.29,
-      "learning_rate": 1.2411633667222283e-05,
-      "loss": 1.7116,
-      "step": 853500
-    },
-    {
-      "epoch": 12.29,
-      "learning_rate": 1.2359810778953054e-05,
-      "loss": 1.7118,
-      "step": 854000
-    },
-    {
-      "epoch": 12.3,
-      "learning_rate": 1.2307884037200317e-05,
-      "loss": 1.7152,
-      "step": 854500
-    },
-    {
-      "epoch": 12.31,
-      "learning_rate": 1.2255957295447582e-05,
-      "loss": 1.7078,
-      "step": 855000
-    },
-    {
-      "epoch": 12.31,
-      "learning_rate": 1.2204030553694847e-05,
-      "loss": 1.7121,
-      "step": 855500
-    },
-    {
-      "epoch": 12.32,
-      "learning_rate": 1.2152207665425617e-05,
-      "loss": 1.7093,
-      "step": 856000
-    },
-    {
-      "epoch": 12.33,
-      "learning_rate": 1.2100280923672883e-05,
-      "loss": 1.708,
-      "step": 856500
-    },
-    {
-      "epoch": 12.34,
-      "learning_rate": 1.2048354181920148e-05,
-      "loss": 1.7122,
-      "step": 857000
-    },
-    {
-      "epoch": 12.34,
-      "learning_rate": 1.1996427440167413e-05,
-      "loss": 1.7071,
-      "step": 857500
-    },
-    {
-      "epoch": 12.35,
-      "learning_rate": 1.1944604551898183e-05,
-      "loss": 1.7148,
-      "step": 858000
-    },
-    {
-      "epoch": 12.36,
-      "learning_rate": 1.1892677810145448e-05,
-      "loss": 1.7069,
-      "step": 858500
-    },
-    {
-      "epoch": 12.36,
-      "learning_rate": 1.1840751068392713e-05,
-      "loss": 1.7117,
-      "step": 859000
-    },
-    {
-      "epoch": 12.37,
-      "learning_rate": 1.1788824326639977e-05,
-      "loss": 1.7114,
-      "step": 859500
-    },
-    {
-      "epoch": 12.38,
-      "learning_rate": 1.1737001438370747e-05,
-      "loss": 1.7095,
-      "step": 860000
-    },
-    {
-      "epoch": 12.39,
-      "learning_rate": 1.1685074696618012e-05,
-      "loss": 1.7103,
-      "step": 860500
-    },
-    {
-      "epoch": 12.39,
-      "learning_rate": 1.1633147954865277e-05,
-      "loss": 1.7089,
-      "step": 861000
-    },
-    {
-      "epoch": 12.4,
-      "learning_rate": 1.1581221213112542e-05,
-      "loss": 1.709,
-      "step": 861500
-    },
-    {
-      "epoch": 12.41,
-      "learning_rate": 1.1529294471359805e-05,
-      "loss": 1.7089,
-      "step": 862000
-    },
-    {
-      "epoch": 12.41,
-      "learning_rate": 1.147736772960707e-05,
-      "loss": 1.7028,
-      "step": 862500
-    },
-    {
-      "epoch": 12.42,
-      "learning_rate": 1.1425544841337841e-05,
-      "loss": 1.7084,
-      "step": 863000
-    },
-    {
-      "epoch": 12.43,
-      "learning_rate": 1.1373618099585104e-05,
-      "loss": 1.7113,
-      "step": 863500
-    },
-    {
-      "epoch": 12.44,
-      "learning_rate": 1.1321691357832371e-05,
-      "loss": 1.7077,
-      "step": 864000
-    },
-    {
-      "epoch": 12.44,
-      "learning_rate": 1.1269764616079636e-05,
-      "loss": 1.7084,
-      "step": 864500
-    },
-    {
-      "epoch": 12.45,
-      "learning_rate": 1.1217837874326901e-05,
-      "loss": 1.7084,
-      "step": 865000
-    },
-    {
-      "epoch": 12.46,
-      "learning_rate": 1.116601498605767e-05,
-      "loss": 1.7087,
-      "step": 865500
-    },
-    {
-      "epoch": 12.47,
-      "learning_rate": 1.1114088244304935e-05,
-      "loss": 1.7086,
-      "step": 866000
-    },
-    {
-      "epoch": 12.47,
-      "learning_rate": 1.10621615025522e-05,
-      "loss": 1.7121,
-      "step": 866500
-    },
-    {
-      "epoch": 12.48,
-      "learning_rate": 1.1010234760799465e-05,
-      "loss": 1.7092,
-      "step": 867000
-    },
-    {
-      "epoch": 12.49,
-      "learning_rate": 1.0958411872530235e-05,
-      "loss": 1.7075,
-      "step": 867500
-    },
-    {
-      "epoch": 12.49,
-      "learning_rate": 1.09064851307775e-05,
-      "loss": 1.7053,
-      "step": 868000
-    },
-    {
-      "epoch": 12.5,
-      "learning_rate": 1.0854558389024765e-05,
-      "loss": 1.7118,
-      "step": 868500
-    },
-    {
-      "epoch": 12.51,
-      "learning_rate": 1.080263164727203e-05,
-      "loss": 1.7089,
-      "step": 869000
-    },
-    {
-      "epoch": 12.52,
-      "learning_rate": 1.07508087590028e-05,
-      "loss": 1.7085,
-      "step": 869500
-    },
-    {
-      "epoch": 12.52,
-      "learning_rate": 1.0698882017250064e-05,
-      "loss": 1.7098,
-      "step": 870000
-    },
-    {
-      "epoch": 12.53,
-      "learning_rate": 1.0646955275497329e-05,
-      "loss": 1.7075,
-      "step": 870500
-    },
-    {
-      "epoch": 12.54,
-      "learning_rate": 1.0595028533744594e-05,
-      "loss": 1.7067,
-      "step": 871000
-    },
-    {
-      "epoch": 12.54,
-      "learning_rate": 1.0543101791991857e-05,
-      "loss": 1.7055,
-      "step": 871500
-    },
-    {
-      "epoch": 12.55,
-      "learning_rate": 1.0491175050239124e-05,
-      "loss": 1.7062,
-      "step": 872000
-    },
-    {
-      "epoch": 12.56,
-      "learning_rate": 1.0439352161969893e-05,
-      "loss": 1.708,
-      "step": 872500
-    },
-    {
-      "epoch": 12.57,
-      "learning_rate": 1.0387425420217158e-05,
-      "loss": 1.7085,
-      "step": 873000
-    },
-    {
-      "epoch": 12.57,
-      "learning_rate": 1.0335498678464423e-05,
-      "loss": 1.7083,
-      "step": 873500
-    },
-    {
-      "epoch": 12.58,
-      "learning_rate": 1.0283571936711688e-05,
-      "loss": 1.7091,
-      "step": 874000
-    },
-    {
-      "epoch": 12.59,
-      "learning_rate": 1.0231645194958953e-05,
-      "loss": 1.707,
-      "step": 874500
-    },
-    {
-      "epoch": 12.59,
-      "learning_rate": 1.0179822306689723e-05,
-      "loss": 1.7085,
-      "step": 875000
-    },
-    {
-      "epoch": 12.6,
-      "learning_rate": 1.0127895564936987e-05,
-      "loss": 1.7052,
-      "step": 875500
-    },
-    {
-      "epoch": 12.61,
-      "learning_rate": 1.0075968823184252e-05,
-      "loss": 1.7092,
-      "step": 876000
-    },
-    {
-      "epoch": 12.62,
-      "learning_rate": 1.0024042081431517e-05,
-      "loss": 1.705,
-      "step": 876500
-    },
-    {
-      "epoch": 12.62,
-      "learning_rate": 9.972219193162287e-06,
-      "loss": 1.7111,
-      "step": 877000
-    },
-    {
-      "epoch": 12.63,
-      "learning_rate": 9.920292451409552e-06,
-      "loss": 1.712,
-      "step": 877500
-    },
-    {
-      "epoch": 12.64,
-      "learning_rate": 9.868365709656817e-06,
-      "loss": 1.7061,
-      "step": 878000
-    },
-    {
-      "epoch": 12.65,
-      "learning_rate": 9.816438967904082e-06,
-      "loss": 1.7059,
-      "step": 878500
-    },
-    {
-      "epoch": 12.65,
-      "learning_rate": 9.764616079634851e-06,
-      "loss": 1.7086,
-      "step": 879000
-    },
-    {
-      "epoch": 12.66,
-      "learning_rate": 9.712689337882116e-06,
-      "loss": 1.7065,
-      "step": 879500
-    },
-    {
-      "epoch": 12.67,
-      "learning_rate": 9.660762596129381e-06,
-      "loss": 1.703,
-      "step": 880000
-    },
-    {
-      "epoch": 12.67,
-      "learning_rate": 9.608835854376646e-06,
-      "loss": 1.7063,
-      "step": 880500
-    },
-    {
-      "epoch": 12.68,
-      "learning_rate": 9.556909112623911e-06,
-      "loss": 1.7087,
-      "step": 881000
-    },
-    {
-      "epoch": 12.69,
-      "learning_rate": 9.504982370871176e-06,
-      "loss": 1.7095,
-      "step": 881500
-    },
-    {
-      "epoch": 12.7,
-      "learning_rate": 9.45305562911844e-06,
-      "loss": 1.7099,
-      "step": 882000
-    },
-    {
-      "epoch": 12.7,
-      "learning_rate": 9.40123274084921e-06,
-      "loss": 1.709,
-      "step": 882500
-    },
-    {
-      "epoch": 12.71,
-      "learning_rate": 9.349305999096475e-06,
-      "loss": 1.7032,
-      "step": 883000
-    },
-    {
-      "epoch": 12.72,
-      "learning_rate": 9.29737925734374e-06,
-      "loss": 1.7104,
-      "step": 883500
-    },
-    {
-      "epoch": 12.72,
-      "learning_rate": 9.245452515591005e-06,
-      "loss": 1.706,
-      "step": 884000
-    },
-    {
-      "epoch": 12.73,
-      "learning_rate": 9.19352577383827e-06,
-      "loss": 1.7081,
-      "step": 884500
-    },
-    {
-      "epoch": 12.74,
-      "learning_rate": 9.14170288556904e-06,
-      "loss": 1.7088,
-      "step": 885000
-    },
-    {
-      "epoch": 12.75,
-      "learning_rate": 9.089776143816304e-06,
-      "loss": 1.7035,
-      "step": 885500
-    },
-    {
-      "epoch": 12.75,
-      "learning_rate": 9.03784940206357e-06,
-      "loss": 1.7031,
-      "step": 886000
-    },
-    {
-      "epoch": 12.76,
-      "learning_rate": 8.985922660310834e-06,
-      "loss": 1.7079,
-      "step": 886500
-    },
-    {
-      "epoch": 12.77,
-      "learning_rate": 8.933995918558097e-06,
-      "loss": 1.7058,
-      "step": 887000
-    },
-    {
-      "epoch": 12.77,
-      "learning_rate": 8.882173030288869e-06,
-      "loss": 1.7032,
-      "step": 887500
-    },
-    {
-      "epoch": 12.78,
-      "learning_rate": 8.830246288536134e-06,
-      "loss": 1.7037,
-      "step": 888000
-    },
-    {
-      "epoch": 12.79,
-      "learning_rate": 8.778319546783399e-06,
-      "loss": 1.703,
-      "step": 888500
-    },
-    {
-      "epoch": 12.8,
-      "learning_rate": 8.726392805030663e-06,
-      "loss": 1.7074,
-      "step": 889000
-    },
-    {
-      "epoch": 12.8,
-      "learning_rate": 8.674466063277928e-06,
-      "loss": 1.7061,
-      "step": 889500
-    },
-    {
-      "epoch": 12.81,
-      "learning_rate": 8.622643175008698e-06,
-      "loss": 1.7065,
-      "step": 890000
-    },
-    {
-      "epoch": 12.82,
-      "learning_rate": 8.570716433255963e-06,
-      "loss": 1.7058,
-      "step": 890500
-    },
-    {
-      "epoch": 12.83,
-      "learning_rate": 8.518789691503228e-06,
-      "loss": 1.7054,
-      "step": 891000
-    },
-    {
-      "epoch": 12.83,
-      "learning_rate": 8.466862949750493e-06,
-      "loss": 1.7071,
-      "step": 891500
-    },
-    {
-      "epoch": 12.84,
-      "learning_rate": 8.415040061481262e-06,
-      "loss": 1.703,
-      "step": 892000
-    },
-    {
-      "epoch": 12.85,
-      "learning_rate": 8.363113319728527e-06,
-      "loss": 1.7054,
-      "step": 892500
-    },
-    {
-      "epoch": 12.85,
-      "learning_rate": 8.311186577975792e-06,
-      "loss": 1.7059,
-      "step": 893000
-    },
-    {
-      "epoch": 12.86,
-      "learning_rate": 8.259259836223057e-06,
-      "loss": 1.7043,
-      "step": 893500
-    },
-    {
-      "epoch": 12.87,
-      "learning_rate": 8.207436947953827e-06,
-      "loss": 1.6995,
-      "step": 894000
-    },
-    {
-      "epoch": 12.88,
-      "learning_rate": 8.155510206201092e-06,
-      "loss": 1.7058,
-      "step": 894500
-    },
-    {
-      "epoch": 12.88,
-      "learning_rate": 8.103583464448357e-06,
-      "loss": 1.7018,
-      "step": 895000
-    },
-    {
-      "epoch": 12.89,
-      "learning_rate": 8.051656722695621e-06,
-      "loss": 1.7065,
-      "step": 895500
-    },
-    {
-      "epoch": 12.9,
-      "learning_rate": 7.999729980942886e-06,
-      "loss": 1.7049,
-      "step": 896000
-    },
-    {
-      "epoch": 12.9,
-      "learning_rate": 7.947907092673656e-06,
-      "loss": 1.7068,
-      "step": 896500
-    },
-    {
-      "epoch": 12.91,
-      "learning_rate": 7.895980350920921e-06,
-      "loss": 1.7001,
-      "step": 897000
-    },
-    {
-      "epoch": 12.92,
-      "learning_rate": 7.844053609168186e-06,
-      "loss": 1.7098,
-      "step": 897500
-    },
-    {
-      "epoch": 12.93,
-      "learning_rate": 7.79212686741545e-06,
-      "loss": 1.7014,
-      "step": 898000
-    },
-    {
-      "epoch": 12.93,
-      "learning_rate": 7.740303979146222e-06,
-      "loss": 1.7061,
-      "step": 898500
-    },
-    {
-      "epoch": 12.94,
-      "learning_rate": 7.688377237393487e-06,
-      "loss": 1.705,
-      "step": 899000
-    },
-    {
-      "epoch": 12.95,
-      "learning_rate": 7.63645049564075e-06,
-      "loss": 1.7037,
-      "step": 899500
-    },
-    {
-      "epoch": 12.95,
-      "learning_rate": 7.584523753888015e-06,
-      "loss": 1.7083,
-      "step": 900000
-    },
-    {
-      "epoch": 12.96,
-      "learning_rate": 7.53259701213528e-06,
-      "loss": 1.7029,
-      "step": 900500
-    },
-    {
-      "epoch": 12.97,
-      "learning_rate": 7.4807741238660495e-06,
-      "loss": 1.7047,
-      "step": 901000
-    },
-    {
-      "epoch": 12.98,
-      "learning_rate": 7.4288473821133144e-06,
-      "loss": 1.7049,
-      "step": 901500
-    },
-    {
-      "epoch": 12.98,
-      "learning_rate": 7.376920640360579e-06,
-      "loss": 1.7073,
-      "step": 902000
-    },
-    {
-      "epoch": 12.99,
-      "learning_rate": 7.324993898607844e-06,
-      "loss": 1.7036,
-      "step": 902500
-    },
-    {
-      "epoch": 13.0,
-      "learning_rate": 7.273171010338614e-06,
-      "loss": 1.7011,
-      "step": 903000
-    },
-    {
-      "epoch": 13.0,
-      "eval_accuracy": 0.6677293660710026,
-      "eval_loss": 1.5736616849899292,
-      "eval_runtime": 1292.8595,
-      "eval_samples_per_second": 416.857,
-      "eval_steps_per_second": 26.054,
-      "step": 903149
-    },
-    {
-      "epoch": 13.01,
-      "learning_rate": 7.221244268585879e-06,
-      "loss": 1.7015,
-      "step": 903500
-    },
-    {
-      "epoch": 13.01,
-      "learning_rate": 7.1693175268331445e-06,
-      "loss": 1.7008,
-      "step": 904000
-    },
-    {
-      "epoch": 13.02,
-      "learning_rate": 7.1173907850804094e-06,
-      "loss": 1.7008,
-      "step": 904500
-    },
-    {
-      "epoch": 13.03,
-      "learning_rate": 7.065567896811179e-06,
-      "loss": 1.7008,
-      "step": 905000
-    },
-    {
-      "epoch": 13.03,
-      "learning_rate": 7.013641155058444e-06,
-      "loss": 1.7018,
-      "step": 905500
-    },
-    {
-      "epoch": 13.04,
-      "learning_rate": 6.961714413305709e-06,
-      "loss": 1.7048,
-      "step": 906000
-    },
-    {
-      "epoch": 13.05,
-      "learning_rate": 6.909787671552974e-06,
-      "loss": 1.7036,
-      "step": 906500
-    },
-    {
-      "epoch": 13.06,
-      "learning_rate": 6.857860929800238e-06,
-      "loss": 1.7012,
-      "step": 907000
-    },
-    {
-      "epoch": 13.06,
-      "learning_rate": 6.805934188047503e-06,
-      "loss": 1.6996,
-      "step": 907500
-    },
-    {
-      "epoch": 13.07,
-      "learning_rate": 6.754111299778273e-06,
-      "loss": 1.701,
-      "step": 908000
-    },
-    {
-      "epoch": 13.08,
-      "learning_rate": 6.702184558025538e-06,
-      "loss": 1.7037,
-      "step": 908500
-    },
-    {
-      "epoch": 13.08,
-      "learning_rate": 6.650257816272802e-06,
-      "loss": 1.7042,
-      "step": 909000
-    },
-    {
-      "epoch": 13.09,
-      "learning_rate": 6.598331074520067e-06,
-      "loss": 1.704,
-      "step": 909500
-    },
-    {
-      "epoch": 13.1,
-      "learning_rate": 6.546508186250838e-06,
-      "loss": 1.7035,
-      "step": 910000
-    },
-    {
-      "epoch": 13.11,
-      "learning_rate": 6.494581444498102e-06,
-      "loss": 1.7007,
-      "step": 910500
-    },
-    {
-      "epoch": 13.11,
-      "learning_rate": 6.4426547027453665e-06,
-      "loss": 1.703,
-      "step": 911000
-    },
-    {
-      "epoch": 13.12,
-      "learning_rate": 6.390727960992631e-06,
-      "loss": 1.7008,
-      "step": 911500
-    },
-    {
-      "epoch": 13.13,
-      "learning_rate": 6.338905072723401e-06,
-      "loss": 1.7057,
-      "step": 912000
-    },
-    {
-      "epoch": 13.13,
-      "learning_rate": 6.286978330970667e-06,
-      "loss": 1.7022,
-      "step": 912500
-    },
-    {
-      "epoch": 13.14,
-      "learning_rate": 6.235051589217932e-06,
-      "loss": 1.7011,
-      "step": 913000
-    },
-    {
-      "epoch": 13.15,
-      "learning_rate": 6.183124847465197e-06,
-      "loss": 1.7006,
-      "step": 913500
-    },
-    {
-      "epoch": 13.16,
-      "learning_rate": 6.131198105712461e-06,
-      "loss": 1.6992,
-      "step": 914000
-    },
-    {
-      "epoch": 13.16,
-      "learning_rate": 6.079375217443231e-06,
-      "loss": 1.702,
-      "step": 914500
-    },
-    {
-      "epoch": 13.17,
-      "learning_rate": 6.027448475690496e-06,
-      "loss": 1.6989,
-      "step": 915000
-    },
-    {
-      "epoch": 13.18,
-      "learning_rate": 5.975521733937761e-06,
-      "loss": 1.7008,
-      "step": 915500
-    },
-    {
-      "epoch": 13.18,
-      "learning_rate": 5.923594992185026e-06,
-      "loss": 1.7023,
-      "step": 916000
-    },
-    {
-      "epoch": 13.19,
-      "learning_rate": 5.871668250432291e-06,
-      "loss": 1.7007,
-      "step": 916500
-    },
-    {
-      "epoch": 13.2,
-      "learning_rate": 5.819845362163061e-06,
-      "loss": 1.6995,
-      "step": 917000
-    },
-    {
-      "epoch": 13.21,
-      "learning_rate": 5.767918620410325e-06,
-      "loss": 1.7002,
-      "step": 917500
-    },
-    {
-      "epoch": 13.21,
-      "learning_rate": 5.71599187865759e-06,
-      "loss": 1.7031,
-      "step": 918000
-    },
-    {
-      "epoch": 13.22,
-      "learning_rate": 5.664065136904855e-06,
-      "loss": 1.6987,
-      "step": 918500
-    },
-    {
-      "epoch": 13.23,
-      "learning_rate": 5.612138395152119e-06,
-      "loss": 1.7016,
-      "step": 919000
-    },
-    {
-      "epoch": 13.24,
-      "learning_rate": 5.56031550688289e-06,
-      "loss": 1.6996,
-      "step": 919500
-    },
-    {
-      "epoch": 13.24,
-      "learning_rate": 5.5083887651301545e-06,
-      "loss": 1.7036,
-      "step": 920000
-    },
-    {
-      "epoch": 13.25,
-      "learning_rate": 5.456462023377419e-06,
-      "loss": 1.7015,
-      "step": 920500
-    },
-    {
-      "epoch": 13.26,
-      "learning_rate": 5.404535281624684e-06,
-      "loss": 1.7019,
-      "step": 921000
-    },
-    {
-      "epoch": 13.26,
-      "learning_rate": 5.352712393355455e-06,
-      "loss": 1.7019,
-      "step": 921500
-    },
-    {
-      "epoch": 13.27,
-      "learning_rate": 5.30078565160272e-06,
-      "loss": 1.7024,
-      "step": 922000
-    },
-    {
-      "epoch": 13.28,
-      "learning_rate": 5.248858909849984e-06,
-      "loss": 1.7003,
-      "step": 922500
-    },
-    {
-      "epoch": 13.29,
-      "learning_rate": 5.196932168097249e-06,
-      "loss": 1.6995,
-      "step": 923000
-    },
-    {
-      "epoch": 13.29,
-      "learning_rate": 5.145109279828019e-06,
-      "loss": 1.6998,
-      "step": 923500
-    },
-    {
-      "epoch": 13.3,
-      "learning_rate": 5.093182538075283e-06,
-      "loss": 1.6999,
-      "step": 924000
-    },
-    {
-      "epoch": 13.31,
-      "learning_rate": 5.041255796322548e-06,
-      "loss": 1.6995,
-      "step": 924500
-    },
-    {
-      "epoch": 13.31,
-      "learning_rate": 4.989329054569814e-06,
-      "loss": 1.6965,
-      "step": 925000
-    },
-    {
-      "epoch": 13.32,
-      "learning_rate": 4.9375061663005835e-06,
-      "loss": 1.7033,
-      "step": 925500
-    },
-    {
-      "epoch": 13.33,
-      "learning_rate": 4.885579424547848e-06,
-      "loss": 1.6986,
-      "step": 926000
-    },
-    {
-      "epoch": 13.34,
-      "learning_rate": 4.833652682795113e-06,
-      "loss": 1.6984,
-      "step": 926500
-    },
-    {
-      "epoch": 13.34,
-      "learning_rate": 4.781725941042377e-06,
-      "loss": 1.7015,
-      "step": 927000
-    },
-    {
-      "epoch": 13.35,
-      "learning_rate": 4.729799199289642e-06,
-      "loss": 1.6982,
-      "step": 927500
-    },
-    {
-      "epoch": 13.36,
-      "learning_rate": 4.677976311020413e-06,
-      "loss": 1.7004,
-      "step": 928000
-    },
-    {
-      "epoch": 13.36,
-      "learning_rate": 4.626049569267677e-06,
-      "loss": 1.702,
-      "step": 928500
-    },
-    {
-      "epoch": 13.37,
-      "learning_rate": 4.5741228275149425e-06,
-      "loss": 1.7002,
-      "step": 929000
-    },
-    {
-      "epoch": 13.38,
-      "learning_rate": 4.5221960857622074e-06,
-      "loss": 1.7009,
-      "step": 929500
-    },
-    {
-      "epoch": 13.39,
-      "learning_rate": 4.4702693440094715e-06,
-      "loss": 1.6987,
-      "step": 930000
-    },
-    {
-      "epoch": 13.39,
-      "learning_rate": 4.418446455740242e-06,
-      "loss": 1.7007,
-      "step": 930500
-    },
-    {
-      "epoch": 13.4,
-      "learning_rate": 4.366519713987507e-06,
-      "loss": 1.6964,
-      "step": 931000
-    },
-    {
-      "epoch": 13.41,
-      "learning_rate": 4.314592972234771e-06,
-      "loss": 1.6999,
-      "step": 931500
-    },
-    {
-      "epoch": 13.42,
-      "learning_rate": 4.262666230482036e-06,
-      "loss": 1.7002,
-      "step": 932000
-    },
-    {
-      "epoch": 13.42,
-      "learning_rate": 4.210843342212806e-06,
-      "loss": 1.6982,
-      "step": 932500
-    },
-    {
-      "epoch": 13.43,
-      "learning_rate": 4.158916600460071e-06,
-      "loss": 1.697,
-      "step": 933000
-    },
-    {
-      "epoch": 13.44,
-      "learning_rate": 4.106989858707336e-06,
-      "loss": 1.7032,
-      "step": 933500
-    },
-    {
-      "epoch": 13.44,
-      "learning_rate": 4.055063116954601e-06,
-      "loss": 1.7021,
-      "step": 934000
-    },
-    {
-      "epoch": 13.45,
-      "learning_rate": 4.003136375201865e-06,
-      "loss": 1.7011,
-      "step": 934500
-    },
-    {
-      "epoch": 13.46,
-      "learning_rate": 3.9513134869326356e-06,
-      "loss": 1.7007,
-      "step": 935000
-    },
-    {
-      "epoch": 13.47,
-      "learning_rate": 3.8993867451799005e-06,
-      "loss": 1.6997,
-      "step": 935500
-    },
-    {
-      "epoch": 13.47,
-      "learning_rate": 3.847460003427165e-06,
-      "loss": 1.7,
-      "step": 936000
-    },
-    {
-      "epoch": 13.48,
-      "learning_rate": 3.79553326167443e-06,
-      "loss": 1.7009,
-      "step": 936500
-    },
-    {
-      "epoch": 13.49,
-      "learning_rate": 3.7436065199216947e-06,
-      "loss": 1.698,
-      "step": 937000
-    },
-    {
-      "epoch": 13.49,
-      "learning_rate": 3.6917836316524652e-06,
-      "loss": 1.703,
-      "step": 937500
-    },
-    {
-      "epoch": 13.5,
-      "learning_rate": 3.6398568898997293e-06,
-      "loss": 1.6989,
-      "step": 938000
-    },
-    {
-      "epoch": 13.51,
-      "learning_rate": 3.5879301481469946e-06,
-      "loss": 1.6989,
-      "step": 938500
-    },
-    {
-      "epoch": 13.52,
-      "learning_rate": 3.5360034063942595e-06,
-      "loss": 1.6994,
-      "step": 939000
-    },
-    {
-      "epoch": 13.52,
-      "learning_rate": 3.484180518125029e-06,
-      "loss": 1.7032,
-      "step": 939500
-    },
-    {
-      "epoch": 13.53,
-      "learning_rate": 3.432253776372294e-06,
-      "loss": 1.7012,
-      "step": 940000
-    },
-    {
-      "epoch": 13.54,
-      "learning_rate": 3.380327034619559e-06,
-      "loss": 1.6949,
-      "step": 940500
-    },
-    {
-      "epoch": 13.54,
-      "learning_rate": 3.3284002928668234e-06,
-      "loss": 1.6991,
-      "step": 941000
-    },
-    {
-      "epoch": 13.55,
-      "learning_rate": 3.2764735511140883e-06,
-      "loss": 1.6962,
-      "step": 941500
-    },
-    {
-      "epoch": 13.56,
-      "learning_rate": 3.224650662844859e-06,
-      "loss": 1.6984,
-      "step": 942000
-    },
-    {
-      "epoch": 13.57,
-      "learning_rate": 3.1727239210921233e-06,
-      "loss": 1.6974,
-      "step": 942500
-    },
-    {
-      "epoch": 13.57,
-      "learning_rate": 3.120797179339388e-06,
-      "loss": 1.6989,
-      "step": 943000
-    },
-    {
-      "epoch": 13.58,
-      "learning_rate": 3.0688704375866527e-06,
-      "loss": 1.6986,
-      "step": 943500
-    },
-    {
-      "epoch": 13.59,
-      "learning_rate": 3.017047549317423e-06,
-      "loss": 1.6972,
-      "step": 944000
-    },
-    {
-      "epoch": 13.6,
-      "learning_rate": 2.9651208075646876e-06,
-      "loss": 1.6985,
-      "step": 944500
-    },
-    {
-      "epoch": 13.6,
-      "learning_rate": 2.913194065811953e-06,
-      "loss": 1.6983,
-      "step": 945000
-    },
-    {
-      "epoch": 13.61,
-      "learning_rate": 2.8612673240592174e-06,
-      "loss": 1.7019,
-      "step": 945500
-    },
-    {
-      "epoch": 13.62,
-      "learning_rate": 2.809444435789988e-06,
-      "loss": 1.6978,
-      "step": 946000
-    },
-    {
-      "epoch": 13.62,
-      "learning_rate": 2.7575176940372524e-06,
-      "loss": 1.6995,
-      "step": 946500
-    },
-    {
-      "epoch": 13.63,
-      "learning_rate": 2.705590952284517e-06,
-      "loss": 1.6951,
-      "step": 947000
-    },
-    {
-      "epoch": 13.64,
-      "learning_rate": 2.6536642105317818e-06,
-      "loss": 1.7,
-      "step": 947500
-    },
-    {
-      "epoch": 13.65,
-      "learning_rate": 2.6017374687790467e-06,
-      "loss": 1.6994,
-      "step": 948000
-    },
-    {
-      "epoch": 13.65,
-      "learning_rate": 2.5499145805098167e-06,
-      "loss": 1.6966,
-      "step": 948500
-    },
-    {
-      "epoch": 13.66,
-      "learning_rate": 2.4979878387570816e-06,
-      "loss": 1.6966,
-      "step": 949000
-    },
-    {
-      "epoch": 13.67,
-      "learning_rate": 2.4460610970043465e-06,
-      "loss": 1.697,
-      "step": 949500
-    },
-    {
-      "epoch": 13.67,
-      "learning_rate": 2.394134355251611e-06,
-      "loss": 1.6988,
-      "step": 950000
-    },
-    {
-      "epoch": 13.68,
-      "learning_rate": 2.342207613498876e-06,
-      "loss": 1.6992,
-      "step": 950500
-    },
-    {
-      "epoch": 13.69,
-      "learning_rate": 2.290280871746141e-06,
-      "loss": 1.6961,
-      "step": 951000
-    },
-    {
-      "epoch": 13.7,
-      "learning_rate": 2.238457983476911e-06,
-      "loss": 1.7023,
-      "step": 951500
-    },
-    {
-      "epoch": 13.7,
-      "learning_rate": 2.1865312417241758e-06,
-      "loss": 1.7008,
-      "step": 952000
-    },
-    {
-      "epoch": 13.71,
-      "learning_rate": 2.1346044999714403e-06,
-      "loss": 1.7013,
-      "step": 952500
-    },
-    {
-      "epoch": 13.72,
-      "learning_rate": 2.082677758218705e-06,
-      "loss": 1.6989,
-      "step": 953000
-    },
-    {
-      "epoch": 13.72,
-      "learning_rate": 2.0307510164659696e-06,
-      "loss": 1.6959,
-      "step": 953500
-    },
-    {
-      "epoch": 13.73,
-      "learning_rate": 1.97892812819674e-06,
-      "loss": 1.6983,
-      "step": 954000
-    },
-    {
-      "epoch": 13.74,
-      "learning_rate": 1.9270013864440046e-06,
-      "loss": 1.7009,
-      "step": 954500
-    },
-    {
-      "epoch": 13.75,
-      "learning_rate": 1.8750746446912697e-06,
-      "loss": 1.6965,
-      "step": 955000
-    },
-    {
-      "epoch": 13.75,
-      "learning_rate": 1.8231479029385344e-06,
-      "loss": 1.697,
-      "step": 955500
-    },
-    {
-      "epoch": 13.76,
-      "learning_rate": 1.7713250146693047e-06,
-      "loss": 1.6969,
-      "step": 956000
-    },
-    {
-      "epoch": 13.77,
-      "learning_rate": 1.7193982729165694e-06,
-      "loss": 1.6936,
-      "step": 956500
-    },
-    {
-      "epoch": 13.78,
-      "learning_rate": 1.667471531163834e-06,
-      "loss": 1.7003,
-      "step": 957000
-    },
-    {
-      "epoch": 13.78,
-      "learning_rate": 1.6155447894110987e-06,
-      "loss": 1.6998,
-      "step": 957500
-    },
-    {
-      "epoch": 13.79,
-      "learning_rate": 1.563721901141869e-06,
-      "loss": 1.6969,
-      "step": 958000
-    },
-    {
-      "epoch": 13.8,
-      "learning_rate": 1.511795159389134e-06,
-      "loss": 1.6965,
-      "step": 958500
-    },
-    {
-      "epoch": 13.8,
-      "learning_rate": 1.4598684176363986e-06,
-      "loss": 1.6967,
-      "step": 959000
-    },
-    {
-      "epoch": 13.81,
-      "learning_rate": 1.4079416758836633e-06,
-      "loss": 1.6951,
-      "step": 959500
-    },
-    {
-      "epoch": 13.82,
-      "learning_rate": 1.3561187876144336e-06,
-      "loss": 1.6965,
-      "step": 960000
-    },
-    {
-      "epoch": 13.83,
-      "learning_rate": 1.3041920458616985e-06,
-      "loss": 1.696,
-      "step": 960500
-    },
-    {
-      "epoch": 13.83,
-      "learning_rate": 1.2522653041089632e-06,
-      "loss": 1.6984,
-      "step": 961000
-    },
-    {
-      "epoch": 13.84,
-      "learning_rate": 1.2003385623562278e-06,
-      "loss": 1.6992,
-      "step": 961500
-    },
-    {
-      "epoch": 13.85,
-      "learning_rate": 1.1484118206034927e-06,
-      "loss": 1.6976,
-      "step": 962000
-    },
-    {
-      "epoch": 13.85,
-      "learning_rate": 1.0965889323342628e-06,
-      "loss": 1.6984,
-      "step": 962500
-    },
-    {
-      "epoch": 13.86,
-      "learning_rate": 1.0446621905815277e-06,
-      "loss": 1.6995,
-      "step": 963000
-    },
-    {
-      "epoch": 13.87,
-      "learning_rate": 9.927354488287924e-07,
-      "loss": 1.6993,
-      "step": 963500
-    },
-    {
-      "epoch": 13.88,
-      "learning_rate": 9.408087070760572e-07,
-      "loss": 1.6974,
-      "step": 964000
-    },
-    {
-      "epoch": 13.88,
-      "learning_rate": 8.889858188068274e-07,
-      "loss": 1.6952,
-      "step": 964500
-    },
-    {
-      "epoch": 13.89,
-      "learning_rate": 8.370590770540923e-07,
-      "loss": 1.6941,
-      "step": 965000
-    },
-    {
-      "epoch": 13.9,
-      "learning_rate": 7.851323353013568e-07,
-      "loss": 1.6948,
-      "step": 965500
-    },
-    {
-      "epoch": 13.9,
-      "learning_rate": 7.332055935486216e-07,
-      "loss": 1.6992,
-      "step": 966000
-    },
-    {
-      "epoch": 13.91,
-      "learning_rate": 6.812788517958863e-07,
-      "loss": 1.7009,
-      "step": 966500
-    },
-    {
-      "epoch": 13.92,
-      "learning_rate": 6.294559635266566e-07,
-      "loss": 1.6947,
-      "step": 967000
-    },
-    {
-      "epoch": 13.93,
-      "learning_rate": 5.775292217739214e-07,
-      "loss": 1.7006,
-      "step": 967500
-    },
-    {
-      "epoch": 13.93,
-      "learning_rate": 5.256024800211861e-07,
-      "loss": 1.7017,
-      "step": 968000
-    },
-    {
-      "epoch": 13.94,
-      "learning_rate": 4.736757382684509e-07,
-      "loss": 1.6951,
-      "step": 968500
-    },
-    {
-      "epoch": 13.95,
-      "learning_rate": 4.218528499992211e-07,
-      "loss": 1.6955,
-      "step": 969000
-    },
-    {
-      "epoch": 13.96,
-      "learning_rate": 3.6992610824648584e-07,
-      "loss": 1.6968,
-      "step": 969500
-    },
-    {
-      "epoch": 13.96,
-      "learning_rate": 3.1799936649375064e-07,
-      "loss": 1.6982,
-      "step": 970000
-    },
-    {
-      "epoch": 13.97,
-      "learning_rate": 2.660726247410154e-07,
-      "loss": 1.6923,
-      "step": 970500
-    },
-    {
-      "epoch": 13.98,
-      "learning_rate": 2.1414588298828015e-07,
-      "loss": 1.6925,
-      "step": 971000
-    },
-    {
-      "epoch": 13.98,
-      "learning_rate": 1.6232299471905037e-07,
-      "loss": 1.695,
-      "step": 971500
-    },
-    {
-      "epoch": 13.99,
-      "learning_rate": 1.1039625296631512e-07,
-      "loss": 1.6977,
-      "step": 972000
-    },
-    {
-      "epoch": 14.0,
-      "learning_rate": 5.846951121357988e-08,
-      "loss": 1.7,
-      "step": 972500
-    },
-    {
-      "epoch": 14.0,
-      "eval_accuracy": 0.6690181349948218,
-      "eval_loss": 1.5665007829666138,
-      "eval_runtime": 1297.3439,
-      "eval_samples_per_second": 415.416,
-      "eval_steps_per_second": 25.964,
-      "step": 972622
-    }
-  ],
-  "max_steps": 972622,
-  "num_train_epochs": 14,
-  "total_flos": 7.363280880004055e+18,
-  "trial_name": null,
-  "trial_params": null
-}