diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12403 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 42.627562982224305,
+  "global_step": 1000000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.96e-07,
+      "loss": 3.5376,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992e-06,
+      "loss": 3.2204,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.99e-06,
+      "loss": 3.0852,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.988000000000001e-06,
+      "loss": 2.9628,
+      "step": 2000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.988e-06,
+      "loss": 2.6672,
+      "step": 2500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988e-06,
+      "loss": 2.1921,
+      "step": 3000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.988000000000001e-06,
+      "loss": 1.9085,
+      "step": 3500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.988e-06,
+      "loss": 1.8146,
+      "step": 4000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.988000000000001e-06,
+      "loss": 1.7548,
+      "step": 4500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.988000000000002e-06,
+      "loss": 1.7099,
+      "step": 5000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0988e-05,
+      "loss": 1.6733,
+      "step": 5500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1988000000000001e-05,
+      "loss": 1.6449,
+      "step": 6000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2988e-05,
+      "loss": 1.6214,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3988000000000002e-05,
+      "loss": 1.6005,
+      "step": 7000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4988000000000001e-05,
+      "loss": 1.5755,
+      "step": 7500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5988e-05,
+      "loss": 1.5608,
+      "step": 8000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6988e-05,
+      "loss": 1.5432,
+      "step": 8500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7988e-05,
+      "loss": 1.5307,
+      "step": 9000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8988000000000002e-05,
+      "loss": 1.5122,
+      "step": 9500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9988000000000002e-05,
+      "loss": 1.5002,
+      "step": 10000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.999647781881444e-05,
+      "loss": 1.4902,
+      "step": 10500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9992912858100306e-05,
+      "loss": 1.4755,
+      "step": 11000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9989347897386173e-05,
+      "loss": 1.4646,
+      "step": 11500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.998578293667204e-05,
+      "loss": 1.4532,
+      "step": 12000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9982225105879337e-05,
+      "loss": 1.4447,
+      "step": 12500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9978660145165204e-05,
+      "loss": 1.4381,
+      "step": 13000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.997509518445107e-05,
+      "loss": 1.4269,
+      "step": 13500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997153022373694e-05,
+      "loss": 1.4211,
+      "step": 14000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9967965263022805e-05,
+      "loss": 1.4155,
+      "step": 14500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.996440030230867e-05,
+      "loss": 1.4087,
+      "step": 15000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9960835341594536e-05,
+      "loss": 1.4009,
+      "step": 15500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9957270380880404e-05,
+      "loss": 1.3942,
+      "step": 16000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.995370542016627e-05,
+      "loss": 1.39,
+      "step": 16500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9950140459452138e-05,
+      "loss": 1.3859,
+      "step": 17000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9946582628659434e-05,
+      "loss": 1.3852,
+      "step": 17500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.99430176679453e-05,
+      "loss": 1.3728,
+      "step": 18000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.993945270723117e-05,
+      "loss": 1.3675,
+      "step": 18500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9935887746517036e-05,
+      "loss": 1.3643,
+      "step": 19000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9932329915724333e-05,
+      "loss": 1.3606,
+      "step": 19500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.99287649550102e-05,
+      "loss": 1.3543,
+      "step": 20000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9925199994296063e-05,
+      "loss": 1.35,
+      "step": 20500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.992163503358193e-05,
+      "loss": 1.3463,
+      "step": 21000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9918070072867798e-05,
+      "loss": 1.3437,
+      "step": 21500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9914505112153665e-05,
+      "loss": 1.3396,
+      "step": 22000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.991094728136096e-05,
+      "loss": 1.3366,
+      "step": 22500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.990738232064683e-05,
+      "loss": 1.3326,
+      "step": 23000
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7545876212630639,
+      "eval_loss": 1.1067582368850708,
+      "eval_runtime": 59.6311,
+      "eval_samples_per_second": 867.736,
+      "eval_steps_per_second": 13.567,
+      "step": 23459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9903817359932696e-05,
+      "loss": 1.3273,
+      "step": 23500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9900252399218563e-05,
+      "loss": 1.323,
+      "step": 24000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.989668743850443e-05,
+      "loss": 1.3248,
+      "step": 24500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9893129607711727e-05,
+      "loss": 1.3175,
+      "step": 25000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9889564646997594e-05,
+      "loss": 1.3164,
+      "step": 25500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9885999686283458e-05,
+      "loss": 1.3105,
+      "step": 26000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9882434725569325e-05,
+      "loss": 1.3113,
+      "step": 26500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.987887689477662e-05,
+      "loss": 1.3085,
+      "step": 27000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.987531193406249e-05,
+      "loss": 1.304,
+      "step": 27500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9871746973348356e-05,
+      "loss": 1.3041,
+      "step": 28000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9868182012634223e-05,
+      "loss": 1.3017,
+      "step": 28500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.986461705192009e-05,
+      "loss": 1.2977,
+      "step": 29000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9861066351048813e-05,
+      "loss": 1.2914,
+      "step": 29500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.985750139033468e-05,
+      "loss": 1.2891,
+      "step": 30000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9853936429620547e-05,
+      "loss": 1.2881,
+      "step": 30500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9850371468906414e-05,
+      "loss": 1.2862,
+      "step": 31000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.984680650819228e-05,
+      "loss": 1.286,
+      "step": 31500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.984324154747815e-05,
+      "loss": 1.2786,
+      "step": 32000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9839676586764016e-05,
+      "loss": 1.2819,
+      "step": 32500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9836111626049883e-05,
+      "loss": 1.2762,
+      "step": 33000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.983254666533575e-05,
+      "loss": 1.2765,
+      "step": 33500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9828981704621617e-05,
+      "loss": 1.2738,
+      "step": 34000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9825416743907484e-05,
+      "loss": 1.2717,
+      "step": 34500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.982185891311478e-05,
+      "loss": 1.2711,
+      "step": 35000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9818293952400648e-05,
+      "loss": 1.267,
+      "step": 35500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9814728991686515e-05,
+      "loss": 1.2626,
+      "step": 36000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9811164030972382e-05,
+      "loss": 1.2634,
+      "step": 36500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9807606200179676e-05,
+      "loss": 1.2639,
+      "step": 37000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9804041239465543e-05,
+      "loss": 1.2574,
+      "step": 37500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.980047627875141e-05,
+      "loss": 1.2568,
+      "step": 38000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9796911318037277e-05,
+      "loss": 1.2565,
+      "step": 38500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9793346357323144e-05,
+      "loss": 1.2556,
+      "step": 39000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.978978139660901e-05,
+      "loss": 1.2518,
+      "step": 39500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.978621643589488e-05,
+      "loss": 1.2532,
+      "step": 40000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9782651475180746e-05,
+      "loss": 1.2514,
+      "step": 40500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9779086514466613e-05,
+      "loss": 1.2484,
+      "step": 41000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.977552155375248e-05,
+      "loss": 1.2443,
+      "step": 41500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9771963722959777e-05,
+      "loss": 1.2452,
+      "step": 42000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.976839876224564e-05,
+      "loss": 1.2446,
+      "step": 42500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9764840931452937e-05,
+      "loss": 1.2429,
+      "step": 43000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9761275970738804e-05,
+      "loss": 1.2408,
+      "step": 43500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.975771101002467e-05,
+      "loss": 1.241,
+      "step": 44000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.975414604931054e-05,
+      "loss": 1.2361,
+      "step": 44500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9750588218517835e-05,
+      "loss": 1.238,
+      "step": 45000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9747023257803702e-05,
+      "loss": 1.2354,
+      "step": 45500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.974345829708957e-05,
+      "loss": 1.2313,
+      "step": 46000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9739893336375436e-05,
+      "loss": 1.2328,
+      "step": 46500
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.7677078867058438,
+      "eval_loss": 1.0249285697937012,
+      "eval_runtime": 61.6684,
+      "eval_samples_per_second": 839.069,
+      "eval_steps_per_second": 13.119,
+      "step": 46918
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.973633550558273e-05,
+      "loss": 1.2284,
+      "step": 47000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9732770544868597e-05,
+      "loss": 1.2291,
+      "step": 47500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9729205584154464e-05,
+      "loss": 1.229,
+      "step": 48000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.972564062344033e-05,
+      "loss": 1.2234,
+      "step": 48500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9722075662726198e-05,
+      "loss": 1.2242,
+      "step": 49000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9718517831933495e-05,
+      "loss": 1.2249,
+      "step": 49500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9714952871219362e-05,
+      "loss": 1.2214,
+      "step": 50000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.971138791050523e-05,
+      "loss": 1.2226,
+      "step": 50500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9707830079712522e-05,
+      "loss": 1.2197,
+      "step": 51000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9704265118998393e-05,
+      "loss": 1.2167,
+      "step": 51500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9700700158284257e-05,
+      "loss": 1.2162,
+      "step": 52000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9697135197570124e-05,
+      "loss": 1.2159,
+      "step": 52500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.969357023685599e-05,
+      "loss": 1.2151,
+      "step": 53000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9690005276141858e-05,
+      "loss": 1.2143,
+      "step": 53500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9686440315427725e-05,
+      "loss": 1.2121,
+      "step": 54000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9682875354713592e-05,
+      "loss": 1.207,
+      "step": 54500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.967931039399946e-05,
+      "loss": 1.2098,
+      "step": 55000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9675752563206756e-05,
+      "loss": 1.2087,
+      "step": 55500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9672187602492623e-05,
+      "loss": 1.2042,
+      "step": 56000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.966862264177849e-05,
+      "loss": 1.2088,
+      "step": 56500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9665057681064358e-05,
+      "loss": 1.204,
+      "step": 57000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9661492720350225e-05,
+      "loss": 1.2049,
+      "step": 57500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9657927759636092e-05,
+      "loss": 1.2028,
+      "step": 58000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.965436279892196e-05,
+      "loss": 1.2024,
+      "step": 58500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9650797838207826e-05,
+      "loss": 1.2015,
+      "step": 59000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.964724000741512e-05,
+      "loss": 1.2018,
+      "step": 59500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9643675046700987e-05,
+      "loss": 1.1964,
+      "step": 60000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9640110085986854e-05,
+      "loss": 1.1938,
+      "step": 60500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.963654512527272e-05,
+      "loss": 1.1993,
+      "step": 61000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9632980164558588e-05,
+      "loss": 1.1946,
+      "step": 61500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9629422333765885e-05,
+      "loss": 1.1979,
+      "step": 62000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9625857373051752e-05,
+      "loss": 1.1936,
+      "step": 62500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.962229241233762e-05,
+      "loss": 1.1933,
+      "step": 63000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9618727451623486e-05,
+      "loss": 1.1927,
+      "step": 63500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.961516962083078e-05,
+      "loss": 1.1913,
+      "step": 64000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9611604660116647e-05,
+      "loss": 1.1916,
+      "step": 64500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9608039699402514e-05,
+      "loss": 1.191,
+      "step": 65000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.960447473868838e-05,
+      "loss": 1.1887,
+      "step": 65500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9600916907895677e-05,
+      "loss": 1.1864,
+      "step": 66000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.959735907710297e-05,
+      "loss": 1.1876,
+      "step": 66500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9593801246310267e-05,
+      "loss": 1.189,
+      "step": 67000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9590236285596134e-05,
+      "loss": 1.1842,
+      "step": 67500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9586671324882e-05,
+      "loss": 1.1851,
+      "step": 68000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.958310636416787e-05,
+      "loss": 1.1837,
+      "step": 68500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9579541403453736e-05,
+      "loss": 1.1844,
+      "step": 69000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9575976442739603e-05,
+      "loss": 1.1822,
+      "step": 69500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.957241148202547e-05,
+      "loss": 1.1818,
+      "step": 70000
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.7751185001590098,
+      "eval_loss": 0.9822936058044434,
+      "eval_runtime": 62.3828,
+      "eval_samples_per_second": 829.459,
+      "eval_steps_per_second": 12.968,
+      "step": 70377
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9568846521311337e-05,
+      "loss": 1.1791,
+      "step": 70500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9565281560597205e-05,
+      "loss": 1.1815,
+      "step": 71000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.956171659988307e-05,
+      "loss": 1.1775,
+      "step": 71500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.955815163916894e-05,
+      "loss": 1.1774,
+      "step": 72000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9554586678454806e-05,
+      "loss": 1.1761,
+      "step": 72500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.9551021717740673e-05,
+      "loss": 1.1747,
+      "step": 73000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.954746388694797e-05,
+      "loss": 1.1712,
+      "step": 73500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9543906056155263e-05,
+      "loss": 1.1718,
+      "step": 74000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.954034109544113e-05,
+      "loss": 1.1729,
+      "step": 74500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.9536776134726997e-05,
+      "loss": 1.172,
+      "step": 75000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9533211174012864e-05,
+      "loss": 1.1722,
+      "step": 75500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.952964621329873e-05,
+      "loss": 1.1705,
+      "step": 76000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.95260812525846e-05,
+      "loss": 1.1667,
+      "step": 76500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.9522516291870466e-05,
+      "loss": 1.1702,
+      "step": 77000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.9518951331156333e-05,
+      "loss": 1.171,
+      "step": 77500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.95153863704422e-05,
+      "loss": 1.1679,
+      "step": 78000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.9511821409728067e-05,
+      "loss": 1.1668,
+      "step": 78500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.950827070885679e-05,
+      "loss": 1.1681,
+      "step": 79000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.9504705748142657e-05,
+      "loss": 1.1668,
+      "step": 79500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.9501140787428524e-05,
+      "loss": 1.1635,
+      "step": 80000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.9497582956635818e-05,
+      "loss": 1.1664,
+      "step": 80500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9494017995921685e-05,
+      "loss": 1.1625,
+      "step": 81000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.9490453035207552e-05,
+      "loss": 1.1653,
+      "step": 81500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.948688807449342e-05,
+      "loss": 1.1629,
+      "step": 82000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9483323113779286e-05,
+      "loss": 1.1641,
+      "step": 82500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.9479758153065157e-05,
+      "loss": 1.1621,
+      "step": 83000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9476193192351024e-05,
+      "loss": 1.1597,
+      "step": 83500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.947262823163689e-05,
+      "loss": 1.16,
+      "step": 84000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9469063270922758e-05,
+      "loss": 1.1583,
+      "step": 84500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9465498310208625e-05,
+      "loss": 1.1592,
+      "step": 85000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.946194047941592e-05,
+      "loss": 1.1562,
+      "step": 85500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9458375518701786e-05,
+      "loss": 1.1578,
+      "step": 86000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8455252525252528e-05,
+      "loss": 1.1569,
+      "step": 86500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8445151515151517e-05,
+      "loss": 1.1544,
+      "step": 87000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8435070707070707e-05,
+      "loss": 1.1557,
+      "step": 87500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.84249696969697e-05,
+      "loss": 1.1524,
+      "step": 88000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.841488888888889e-05,
+      "loss": 1.1061,
+      "step": 88500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.840478787878788e-05,
+      "loss": 1.1047,
+      "step": 89000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.839468686868687e-05,
+      "loss": 1.1074,
+      "step": 89500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.838458585858586e-05,
+      "loss": 1.1053,
+      "step": 90000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.837448484848485e-05,
+      "loss": 1.1541,
+      "step": 90500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.836438383838384e-05,
+      "loss": 1.1524,
+      "step": 91000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.835428282828283e-05,
+      "loss": 1.1488,
+      "step": 91500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.834418181818182e-05,
+      "loss": 1.1508,
+      "step": 92000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.833410101010101e-05,
+      "loss": 1.1487,
+      "step": 92500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8324000000000003e-05,
+      "loss": 1.1491,
+      "step": 93000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8313919191919193e-05,
+      "loss": 1.1513,
+      "step": 93500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8303818181818182e-05,
+      "loss": 1.1469,
+      "step": 94000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8293717171717174e-05,
+      "loss": 1.1486,
+      "step": 94500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8283616161616162e-05,
+      "loss": 1.1467,
+      "step": 95000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8273515151515154e-05,
+      "loss": 1.1441,
+      "step": 95500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.8263414141414143e-05,
+      "loss": 1.1458,
+      "step": 96000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.825331313131313e-05,
+      "loss": 1.1457,
+      "step": 96500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.824321212121212e-05,
+      "loss": 1.1442,
+      "step": 97000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.8233111111111112e-05,
+      "loss": 1.1459,
+      "step": 97500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.8223010101010104e-05,
+      "loss": 1.1439,
+      "step": 98000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.8212929292929295e-05,
+      "loss": 1.1419,
+      "step": 98500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.8202828282828283e-05,
+      "loss": 1.1424,
+      "step": 99000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.8192727272727275e-05,
+      "loss": 1.1424,
+      "step": 99500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.8182646464646466e-05,
+      "loss": 1.1398,
+      "step": 100000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.8172545454545454e-05,
+      "loss": 1.1388,
+      "step": 100500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.8162444444444446e-05,
+      "loss": 1.1398,
+      "step": 101000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.8152343434343435e-05,
+      "loss": 1.14,
+      "step": 101500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.8142242424242427e-05,
+      "loss": 1.1382,
+      "step": 102000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.8132141414141416e-05,
+      "loss": 1.1369,
+      "step": 102500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8122040404040408e-05,
+      "loss": 1.1373,
+      "step": 103000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.8111939393939396e-05,
+      "loss": 1.1358,
+      "step": 103500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.8101858585858587e-05,
+      "loss": 1.1351,
+      "step": 104000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.809177777777778e-05,
+      "loss": 1.1375,
+      "step": 104500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.808167676767677e-05,
+      "loss": 1.1339,
+      "step": 105000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8071575757575758e-05,
+      "loss": 1.1358,
+      "step": 105500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.806147474747475e-05,
+      "loss": 1.1348,
+      "step": 106000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.805137373737374e-05,
+      "loss": 1.1341,
+      "step": 106500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.804127272727273e-05,
+      "loss": 1.1336,
+      "step": 107000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.803117171717172e-05,
+      "loss": 1.1334,
+      "step": 107500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8021070707070708e-05,
+      "loss": 1.1341,
+      "step": 108000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.80109898989899e-05,
+      "loss": 1.1342,
+      "step": 108500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.800088888888889e-05,
+      "loss": 1.1308,
+      "step": 109000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7990787878787882e-05,
+      "loss": 1.1319,
+      "step": 109500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.798068686868687e-05,
+      "loss": 1.1297,
+      "step": 110000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.797060606060606e-05,
+      "loss": 1.1298,
+      "step": 110500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7960525252525255e-05,
+      "loss": 1.1319,
+      "step": 111000
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.7826185119761669,
+      "eval_loss": 0.9372493624687195,
+      "eval_runtime": 54.8263,
+      "eval_samples_per_second": 943.78,
+      "eval_steps_per_second": 14.756,
+      "step": 111459
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7950424242424244e-05,
+      "loss": 1.1277,
+      "step": 111500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7940323232323232e-05,
+      "loss": 1.1274,
+      "step": 112000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7930222222222224e-05,
+      "loss": 1.1262,
+      "step": 112500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7920121212121213e-05,
+      "loss": 1.127,
+      "step": 113000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7910020202020205e-05,
+      "loss": 1.1277,
+      "step": 113500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7899919191919194e-05,
+      "loss": 1.1258,
+      "step": 114000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7889818181818182e-05,
+      "loss": 1.1259,
+      "step": 114500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7879717171717174e-05,
+      "loss": 1.1255,
+      "step": 115000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7869636363636365e-05,
+      "loss": 1.1258,
+      "step": 115500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7859535353535357e-05,
+      "loss": 1.1248,
+      "step": 116000
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7849434343434345e-05,
+      "loss": 1.125,
+      "step": 116500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7839333333333334e-05,
+      "loss": 1.1236,
+      "step": 117000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7829232323232322e-05,
+      "loss": 1.124,
+      "step": 117500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7819131313131314e-05,
+      "loss": 1.1236,
+      "step": 118000
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7809030303030303e-05,
+      "loss": 1.1227,
+      "step": 118500
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7798949494949497e-05,
+      "loss": 1.1235,
+      "step": 119000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7788848484848486e-05,
+      "loss": 1.1203,
+      "step": 119500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7778747474747474e-05,
+      "loss": 1.1216,
+      "step": 120000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7768646464646466e-05,
+      "loss": 1.1224,
+      "step": 120500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7758545454545455e-05,
+      "loss": 1.1203,
+      "step": 121000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7748444444444447e-05,
+      "loss": 1.1159,
+      "step": 121500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.7738343434343435e-05,
+      "loss": 1.1201,
+      "step": 122000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.7728242424242427e-05,
+      "loss": 1.1205,
+      "step": 122500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.7718161616161618e-05,
+      "loss": 1.1184,
+      "step": 123000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.770806060606061e-05,
+      "loss": 1.1201,
+      "step": 123500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.76979595959596e-05,
+      "loss": 1.1181,
+      "step": 124000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7687858585858587e-05,
+      "loss": 1.1159,
+      "step": 124500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.7677777777777778e-05,
+      "loss": 1.1166,
+      "step": 125000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.766767676767677e-05,
+      "loss": 1.1158,
+      "step": 125500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.765757575757576e-05,
+      "loss": 1.1158,
+      "step": 126000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.764749494949495e-05,
+      "loss": 1.1157,
+      "step": 126500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.763739393939394e-05,
+      "loss": 1.1149,
+      "step": 127000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.7627292929292933e-05,
+      "loss": 1.1146,
+      "step": 127500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7617212121212123e-05,
+      "loss": 1.1142,
+      "step": 128000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.7607111111111112e-05,
+      "loss": 1.1163,
+      "step": 128500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.7597010101010104e-05,
+      "loss": 1.1126,
+      "step": 129000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.7586909090909092e-05,
+      "loss": 1.1118,
+      "step": 129500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.7576808080808084e-05,
+      "loss": 1.1124,
+      "step": 130000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.7566707070707073e-05,
+      "loss": 1.1132,
+      "step": 130500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.755660606060606e-05,
+      "loss": 1.1121,
+      "step": 131000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.754650505050505e-05,
+      "loss": 1.1129,
+      "step": 131500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7536404040404042e-05,
+      "loss": 1.1145,
+      "step": 132000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.752630303030303e-05,
+      "loss": 1.1118,
+      "step": 132500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7516202020202023e-05,
+      "loss": 1.1116,
+      "step": 133000
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.750610101010101e-05,
+      "loss": 1.111,
+      "step": 133500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.7496000000000003e-05,
+      "loss": 1.1093,
+      "step": 134000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7485898989898992e-05,
+      "loss": 1.1116,
+      "step": 134500
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.7848298822164493,
+      "eval_loss": 0.922777533531189,
+      "eval_runtime": 56.1658,
+      "eval_samples_per_second": 921.273,
+      "eval_steps_per_second": 14.404,
+      "step": 134918
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.747579797979798e-05,
+      "loss": 1.1099,
+      "step": 135000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.7465717171717174e-05,
+      "loss": 1.1078,
+      "step": 135500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.7455616161616163e-05,
+      "loss": 1.1091,
+      "step": 136000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.7445515151515155e-05,
+      "loss": 1.1063,
+      "step": 136500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.7435414141414144e-05,
+      "loss": 1.108,
+      "step": 137000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.7425333333333334e-05,
+      "loss": 1.1104,
+      "step": 137500
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.7415232323232326e-05,
+      "loss": 1.1049,
+      "step": 138000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.7405151515151517e-05,
+      "loss": 1.1059,
+      "step": 138500
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.7395050505050505e-05,
+      "loss": 1.1059,
+      "step": 139000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.7384949494949497e-05,
+      "loss": 1.1055,
+      "step": 139500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.7374848484848486e-05,
+      "loss": 1.1065,
+      "step": 140000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.7364747474747478e-05,
+      "loss": 1.105,
+      "step": 140500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.7354646464646466e-05,
+      "loss": 1.1027,
+      "step": 141000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.7344545454545455e-05,
+      "loss": 1.1037,
+      "step": 141500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.7334444444444444e-05,
+      "loss": 1.1026,
+      "step": 142000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.7324363636363637e-05,
+      "loss": 1.1021,
+      "step": 142500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.731426262626263e-05,
+      "loss": 1.1034,
+      "step": 143000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.7304161616161618e-05,
+      "loss": 1.0995,
+      "step": 143500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.7294060606060607e-05,
+      "loss": 1.1042,
+      "step": 144000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.7283959595959595e-05,
+      "loss": 1.1014,
+      "step": 144500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.727387878787879e-05,
+      "loss": 1.1014,
+      "step": 145000
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.7263777777777778e-05,
+      "loss": 1.1014,
+      "step": 145500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.7253676767676766e-05,
+      "loss": 1.0997,
+      "step": 146000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.724357575757576e-05,
+      "loss": 1.1014,
+      "step": 146500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.7233494949494952e-05,
+      "loss": 1.1006,
+      "step": 147000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.722339393939394e-05,
+      "loss": 1.0981,
+      "step": 147500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.721329292929293e-05,
+      "loss": 1.0996,
+      "step": 148000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.720319191919192e-05,
+      "loss": 1.0985,
+      "step": 148500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.719309090909091e-05,
+      "loss": 1.0991,
+      "step": 149000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.7182989898989902e-05,
+      "loss": 1.1012,
+      "step": 149500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.717288888888889e-05,
+      "loss": 1.0994,
+      "step": 150000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.7162787878787883e-05,
+      "loss": 1.0969,
+      "step": 150500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.7152707070707073e-05,
+      "loss": 1.0978,
+      "step": 151000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.7142626262626264e-05,
+      "loss": 1.0999,
+      "step": 151500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.7132525252525252e-05,
+      "loss": 1.0999,
+      "step": 152000
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.7122424242424244e-05,
+      "loss": 1.0973,
+      "step": 152500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.7112323232323233e-05,
+      "loss": 1.0991,
+      "step": 153000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.7102222222222225e-05,
+      "loss": 1.0963,
+      "step": 153500
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.7092121212121213e-05,
+      "loss": 1.0951,
+      "step": 154000
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.7082020202020205e-05,
+      "loss": 1.0958,
+      "step": 154500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.7071919191919194e-05,
+      "loss": 1.0967,
+      "step": 155000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.7061838383838385e-05,
+      "loss": 1.0961,
+      "step": 155500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.7051737373737377e-05,
+      "loss": 1.0977,
+      "step": 156000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.7041636363636365e-05,
+      "loss": 1.0919,
+      "step": 156500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.7031535353535357e-05,
+      "loss": 1.0936,
+      "step": 157000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.7021434343434346e-05,
+      "loss": 1.0936,
+      "step": 157500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.7011333333333334e-05,
+      "loss": 1.0914,
+      "step": 158000
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.7873718456746811,
+      "eval_loss": 0.9092118740081787,
+      "eval_runtime": 55.9111,
+      "eval_samples_per_second": 925.47,
+      "eval_steps_per_second": 14.469,
+      "step": 158377
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.7001232323232323e-05,
+      "loss": 1.0931,
+      "step": 158500
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.6991131313131315e-05,
+      "loss": 1.0902,
+      "step": 159000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.6981050505050505e-05,
+      "loss": 1.0955,
+      "step": 159500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.6970949494949494e-05,
+      "loss": 1.0928,
+      "step": 160000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.6960868686868688e-05,
+      "loss": 1.0925,
+      "step": 160500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.695076767676768e-05,
+      "loss": 1.0935,
+      "step": 161000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.694066666666667e-05,
+      "loss": 1.0916,
+      "step": 161500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.6930565656565657e-05,
+      "loss": 1.0923,
+      "step": 162000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.6920464646464646e-05,
+      "loss": 1.0892,
+      "step": 162500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.6910363636363638e-05,
+      "loss": 1.0928,
+      "step": 163000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.6900262626262626e-05,
+      "loss": 1.0897,
+      "step": 163500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.689016161616162e-05,
+      "loss": 1.0903,
+      "step": 164000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.688008080808081e-05,
+      "loss": 1.0924,
+      "step": 164500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.6869979797979797e-05,
+      "loss": 1.0893,
+      "step": 165000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.685987878787879e-05,
+      "loss": 1.0897,
+      "step": 165500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.6849777777777778e-05,
+      "loss": 1.0907,
+      "step": 166000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.683967676767677e-05,
+      "loss": 1.0861,
+      "step": 166500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.682957575757576e-05,
+      "loss": 1.09,
+      "step": 167000
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.6819474747474747e-05,
+      "loss": 1.0873,
+      "step": 167500
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.680937373737374e-05,
+      "loss": 1.0889,
+      "step": 168000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.6799272727272728e-05,
+      "loss": 1.0825,
+      "step": 168500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.678917171717172e-05,
+      "loss": 1.0857,
+      "step": 169000
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.677907070707071e-05,
+      "loss": 1.0879,
+      "step": 169500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.67689898989899e-05,
+      "loss": 1.0869,
+      "step": 170000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.675888888888889e-05,
+      "loss": 1.085,
+      "step": 170500
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.674878787878788e-05,
+      "loss": 1.0875,
+      "step": 171000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.673868686868687e-05,
+      "loss": 1.0871,
+      "step": 171500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.672858585858586e-05,
+      "loss": 1.0848,
+      "step": 172000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.6718484848484852e-05,
+      "loss": 1.0838,
+      "step": 172500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.670838383838384e-05,
+      "loss": 1.0866,
+      "step": 173000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.669828282828283e-05,
+      "loss": 1.0851,
+      "step": 173500
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6688202020202023e-05,
+      "loss": 1.0862,
+      "step": 174000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6678101010101012e-05,
+      "loss": 1.0837,
+      "step": 174500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6668e-05,
+      "loss": 1.084,
+      "step": 175000
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6657898989898992e-05,
+      "loss": 1.0833,
+      "step": 175500
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.664779797979798e-05,
+      "loss": 1.0836,
+      "step": 176000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6637717171717175e-05,
+      "loss": 1.0823,
+      "step": 176500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6627636363636365e-05,
+      "loss": 1.0846,
+      "step": 177000
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.6617535353535354e-05,
+      "loss": 1.0834,
+      "step": 177500
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.6607434343434346e-05,
+      "loss": 1.0796,
+      "step": 178000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.6597333333333335e-05,
+      "loss": 1.0812,
+      "step": 178500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.6587232323232327e-05,
+      "loss": 1.083,
+      "step": 179000
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.6577131313131315e-05,
+      "loss": 1.0791,
+      "step": 179500
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.6567030303030304e-05,
+      "loss": 1.0785,
+      "step": 180000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.6556929292929292e-05,
+      "loss": 1.0827,
+      "step": 180500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.6546828282828284e-05,
+      "loss": 1.0786,
+      "step": 181000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.6536727272727273e-05,
+      "loss": 1.0813,
+      "step": 181500
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.7889661479765855,
+      "eval_loss": 0.8995802998542786,
+      "eval_runtime": 56.4423,
+      "eval_samples_per_second": 916.76,
+      "eval_steps_per_second": 14.333,
+      "step": 181836
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.6526646464646467e-05,
+      "loss": 1.0818,
+      "step": 182000
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.6516565656565657e-05,
+      "loss": 1.0785,
+      "step": 182500
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.650646464646465e-05,
+      "loss": 1.0788,
+      "step": 183000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.6496363636363638e-05,
+      "loss": 1.0792,
+      "step": 183500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.6486262626262627e-05,
+      "loss": 1.0796,
+      "step": 184000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.6476161616161615e-05,
+      "loss": 1.0763,
+      "step": 184500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.6466060606060607e-05,
+      "loss": 1.0768,
+      "step": 185000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.6455959595959596e-05,
+      "loss": 1.0795,
+      "step": 185500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.6445858585858588e-05,
+      "loss": 1.0802,
+      "step": 186000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.6435757575757576e-05,
+      "loss": 1.0742,
+      "step": 186500
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.6425676767676767e-05,
+      "loss": 1.0764,
+      "step": 187000
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.641557575757576e-05,
+      "loss": 1.0763,
+      "step": 187500
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.6405474747474747e-05,
+      "loss": 1.0756,
+      "step": 188000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.639537373737374e-05,
+      "loss": 1.0777,
+      "step": 188500
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.6385272727272728e-05,
+      "loss": 1.0742,
+      "step": 189000
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.637519191919192e-05,
+      "loss": 1.0784,
+      "step": 189500
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.636509090909091e-05,
+      "loss": 1.0741,
+      "step": 190000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.6354989898989903e-05,
+      "loss": 1.0746,
+      "step": 190500
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.634488888888889e-05,
+      "loss": 1.0757,
+      "step": 191000
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.633478787878788e-05,
+      "loss": 1.0741,
+      "step": 191500
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.632470707070707e-05,
+      "loss": 1.074,
+      "step": 192000
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.6314606060606062e-05,
+      "loss": 1.0743,
+      "step": 192500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.6304505050505054e-05,
+      "loss": 1.0725,
+      "step": 193000
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.6294404040404043e-05,
+      "loss": 1.0753,
+      "step": 193500
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.628430303030303e-05,
+      "loss": 1.0724,
+      "step": 194000
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.627420202020202e-05,
+      "loss": 1.0751,
+      "step": 194500
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.6264101010101012e-05,
+      "loss": 1.0749,
+      "step": 195000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.6254e-05,
+      "loss": 1.0743,
+      "step": 195500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.6243898989898993e-05,
+      "loss": 1.0736,
+      "step": 196000
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.6233818181818183e-05,
+      "loss": 1.0724,
+      "step": 196500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.6223717171717172e-05,
+      "loss": 1.0711,
+      "step": 197000
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.6213636363636366e-05,
+      "loss": 1.0749,
+      "step": 197500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.6203555555555556e-05,
+      "loss": 1.0735,
+      "step": 198000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.6193454545454548e-05,
+      "loss": 1.0713,
+      "step": 198500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.6183353535353537e-05,
+      "loss": 1.072,
+      "step": 199000
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.617325252525253e-05,
+      "loss": 1.0715,
+      "step": 199500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.6163151515151517e-05,
+      "loss": 1.0702,
+      "step": 200000
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.6153050505050506e-05,
+      "loss": 1.0705,
+      "step": 200500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.6142949494949495e-05,
+      "loss": 1.0745,
+      "step": 201000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6132848484848487e-05,
+      "loss": 1.0707,
+      "step": 201500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.6122747474747475e-05,
+      "loss": 1.0714,
+      "step": 202000
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.611266666666667e-05,
+      "loss": 1.0695,
+      "step": 202500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.6102565656565658e-05,
+      "loss": 1.0718,
+      "step": 203000
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.6092464646464646e-05,
+      "loss": 1.0704,
+      "step": 203500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.6082363636363638e-05,
+      "loss": 1.0684,
+      "step": 204000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.6072262626262627e-05,
+      "loss": 1.0678,
+      "step": 204500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.606216161616162e-05,
+      "loss": 1.0691,
+      "step": 205000
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.6052060606060607e-05,
+      "loss": 1.0676,
+      "step": 205500
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.6041959595959596e-05,
+      "loss": 1.0653,
+      "step": 206000
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.6031858585858585e-05,
+      "loss": 1.0667,
+      "step": 206500
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.6021757575757577e-05,
+      "loss": 1.0695,
+      "step": 207000
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.601167676767677e-05,
+      "loss": 1.0674,
+      "step": 207500
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.600157575757576e-05,
+      "loss": 1.0654,
+      "step": 208000
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.5991474747474748e-05,
+      "loss": 1.0678,
+      "step": 208500
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.598139393939394e-05,
+      "loss": 1.0665,
+      "step": 209000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.597129292929293e-05,
+      "loss": 1.0674,
+      "step": 209500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.5961191919191922e-05,
+      "loss": 1.0654,
+      "step": 210000
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.595109090909091e-05,
+      "loss": 1.0658,
+      "step": 210500
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.59409898989899e-05,
+      "loss": 1.0669,
+      "step": 211000
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.7910602816265144,
+      "eval_loss": 0.8885152339935303,
+      "eval_runtime": 55.1982,
+      "eval_samples_per_second": 937.422,
+      "eval_steps_per_second": 14.656,
+      "step": 211131
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.593088888888889e-05,
+      "loss": 1.0656,
+      "step": 211500
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.592078787878788e-05,
+      "loss": 1.0632,
+      "step": 212000
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.591070707070707e-05,
+      "loss": 1.0627,
+      "step": 212500
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.5900606060606062e-05,
+      "loss": 1.0669,
+      "step": 213000
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.5890525252525253e-05,
+      "loss": 1.0649,
+      "step": 213500
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.5880424242424245e-05,
+      "loss": 1.0643,
+      "step": 214000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.5870323232323234e-05,
+      "loss": 1.0667,
+      "step": 214500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.5860222222222222e-05,
+      "loss": 1.0641,
+      "step": 215000
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.5850141414141416e-05,
+      "loss": 1.0624,
+      "step": 215500
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.5840040404040405e-05,
+      "loss": 1.0641,
+      "step": 216000
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.5829939393939397e-05,
+      "loss": 1.0642,
+      "step": 216500
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.5819838383838385e-05,
+      "loss": 1.0625,
+      "step": 217000
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.5809737373737374e-05,
+      "loss": 1.062,
+      "step": 217500
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.5799636363636366e-05,
+      "loss": 1.0616,
+      "step": 218000
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.5789535353535354e-05,
+      "loss": 1.0609,
+      "step": 218500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.5779434343434346e-05,
+      "loss": 1.0648,
+      "step": 219000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.5769333333333335e-05,
+      "loss": 1.0604,
+      "step": 219500
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.5759232323232324e-05,
+      "loss": 1.0623,
+      "step": 220000
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.5749131313131316e-05,
+      "loss": 1.0619,
+      "step": 220500
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.5739030303030304e-05,
+      "loss": 1.0612,
+      "step": 221000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.5728949494949498e-05,
+      "loss": 1.0631,
+      "step": 221500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.5718848484848487e-05,
+      "loss": 1.0619,
+      "step": 222000
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5708747474747475e-05,
+      "loss": 1.0626,
+      "step": 222500
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.5698646464646464e-05,
+      "loss": 1.0615,
+      "step": 223000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5688545454545456e-05,
+      "loss": 1.0611,
+      "step": 223500
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.567846464646465e-05,
+      "loss": 1.0623,
+      "step": 224000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.566838383838384e-05,
+      "loss": 1.0596,
+      "step": 224500
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.565828282828283e-05,
+      "loss": 1.0625,
+      "step": 225000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.564818181818182e-05,
+      "loss": 1.0597,
+      "step": 225500
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.563808080808081e-05,
+      "loss": 1.059,
+      "step": 226000
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.5628e-05,
+      "loss": 1.0586,
+      "step": 226500
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.5617898989898992e-05,
+      "loss": 1.0588,
+      "step": 227000
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.560779797979798e-05,
+      "loss": 1.0594,
+      "step": 227500
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.5597696969696973e-05,
+      "loss": 1.0576,
+      "step": 228000
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.558759595959596e-05,
+      "loss": 1.0577,
+      "step": 228500
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.557749494949495e-05,
+      "loss": 1.0579,
+      "step": 229000
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.556739393939394e-05,
+      "loss": 1.0581,
+      "step": 229500
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.555729292929293e-05,
+      "loss": 1.0582,
+      "step": 230000
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.5547212121212124e-05,
+      "loss": 1.0615,
+      "step": 230500
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.5537111111111113e-05,
+      "loss": 1.0594,
+      "step": 231000
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.55270101010101e-05,
+      "loss": 1.0568,
+      "step": 231500
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.551690909090909e-05,
+      "loss": 1.0573,
+      "step": 232000
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.5506808080808082e-05,
+      "loss": 1.0569,
+      "step": 232500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.5496727272727276e-05,
+      "loss": 1.0574,
+      "step": 233000
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.5486626262626265e-05,
+      "loss": 1.0588,
+      "step": 233500
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.5476525252525253e-05,
+      "loss": 1.0588,
+      "step": 234000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.5466424242424242e-05,
+      "loss": 1.0568,
+      "step": 234500
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.792497530987494,
+      "eval_loss": 0.8781149387359619,
+      "eval_runtime": 56.3835,
+      "eval_samples_per_second": 917.716,
+      "eval_steps_per_second": 14.348,
+      "step": 234590
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.5456343434343436e-05,
+      "loss": 1.0558,
+      "step": 235000
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.5446242424242424e-05,
+      "loss": 1.0505,
+      "step": 235500
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5436141414141413e-05,
+      "loss": 1.0551,
+      "step": 236000
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.5426040404040405e-05,
+      "loss": 1.0543,
+      "step": 236500
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.5415939393939394e-05,
+      "loss": 1.0547,
+      "step": 237000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.5405838383838386e-05,
+      "loss": 1.0557,
+      "step": 237500
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.5395737373737378e-05,
+      "loss": 1.0538,
+      "step": 238000
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.5385656565656565e-05,
+      "loss": 1.0546,
+      "step": 238500
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.5375555555555557e-05,
+      "loss": 1.056,
+      "step": 239000
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.5365454545454545e-05,
+      "loss": 1.0553,
+      "step": 239500
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5355353535353537e-05,
+      "loss": 1.053,
+      "step": 240000
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5345252525252526e-05,
+      "loss": 1.0553,
+      "step": 240500
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5335151515151518e-05,
+      "loss": 1.0541,
+      "step": 241000
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5325050505050506e-05,
+      "loss": 1.0548,
+      "step": 241500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5314949494949495e-05,
+      "loss": 1.0541,
+      "step": 242000
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5304848484848487e-05,
+      "loss": 1.055,
+      "step": 242500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5294767676767678e-05,
+      "loss": 1.053,
+      "step": 243000
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.5284666666666666e-05,
+      "loss": 1.0515,
+      "step": 243500
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5274565656565658e-05,
+      "loss": 1.0534,
+      "step": 244000
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5264464646464647e-05,
+      "loss": 1.0522,
+      "step": 244500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5254363636363639e-05,
+      "loss": 1.0516,
+      "step": 245000
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5244262626262627e-05,
+      "loss": 1.0539,
+      "step": 245500
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5234161616161618e-05,
+      "loss": 1.0512,
+      "step": 246000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.522408080808081e-05,
+      "loss": 1.0518,
+      "step": 246500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.5213979797979798e-05,
+      "loss": 1.0511,
+      "step": 247000
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.520387878787879e-05,
+      "loss": 1.051,
+      "step": 247500
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.5193777777777779e-05,
+      "loss": 1.0511,
+      "step": 248000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.518367676767677e-05,
+      "loss": 1.0544,
+      "step": 248500
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.5173575757575758e-05,
+      "loss": 1.053,
+      "step": 249000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.516347474747475e-05,
+      "loss": 1.0519,
+      "step": 249500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.5153373737373738e-05,
+      "loss": 1.0513,
+      "step": 250000
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.5143292929292929e-05,
+      "loss": 1.0518,
+      "step": 250500
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.5133191919191921e-05,
+      "loss": 1.0509,
+      "step": 251000
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.512309090909091e-05,
+      "loss": 1.0511,
+      "step": 251500
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.51129898989899e-05,
+      "loss": 1.0502,
+      "step": 252000
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.5102909090909092e-05,
+      "loss": 1.0493,
+      "step": 252500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.509280808080808e-05,
+      "loss": 1.0496,
+      "step": 253000
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.5082707070707073e-05,
+      "loss": 1.0517,
+      "step": 253500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.5072626262626265e-05,
+      "loss": 1.054,
+      "step": 254000
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.5062525252525254e-05,
+      "loss": 1.0517,
+      "step": 254500
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.5052424242424244e-05,
+      "loss": 1.0512,
+      "step": 255000
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.5042323232323232e-05,
+      "loss": 1.0518,
+      "step": 255500
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.5032222222222224e-05,
+      "loss": 1.0499,
+      "step": 256000
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.5022141414141415e-05,
+      "loss": 1.05,
+      "step": 256500
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.5012040404040405e-05,
+      "loss": 1.0465,
+      "step": 257000
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.5001939393939395e-05,
+      "loss": 1.0471,
+      "step": 257500
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4991838383838384e-05,
+      "loss": 1.0506,
+      "step": 258000
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.7936039348617321,
+      "eval_loss": 0.8738781213760376,
+      "eval_runtime": 55.4932,
+      "eval_samples_per_second": 932.438,
+      "eval_steps_per_second": 14.578,
+      "step": 258049
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.4981757575757576e-05,
+      "loss": 1.05,
+      "step": 258500
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.4971656565656567e-05,
+      "loss": 1.0464,
+      "step": 259000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.4961555555555555e-05,
+      "loss": 1.0468,
+      "step": 259500
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.4951454545454547e-05,
+      "loss": 1.0469,
+      "step": 260000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.4941353535353536e-05,
+      "loss": 1.0457,
+      "step": 260500
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.4931252525252526e-05,
+      "loss": 1.0475,
+      "step": 261000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.4921171717171718e-05,
+      "loss": 1.0463,
+      "step": 261500
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.4911070707070707e-05,
+      "loss": 1.0463,
+      "step": 262000
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.4900969696969699e-05,
+      "loss": 1.0451,
+      "step": 262500
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.489086868686869e-05,
+      "loss": 1.0451,
+      "step": 263000
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.4880767676767678e-05,
+      "loss": 1.0452,
+      "step": 263500
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.4870666666666668e-05,
+      "loss": 1.0478,
+      "step": 264000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.4860565656565658e-05,
+      "loss": 1.045,
+      "step": 264500
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.4850484848484849e-05,
+      "loss": 1.0457,
+      "step": 265000
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.4840383838383841e-05,
+      "loss": 1.0473,
+      "step": 265500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.483028282828283e-05,
+      "loss": 1.0469,
+      "step": 266000
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.482018181818182e-05,
+      "loss": 1.0456,
+      "step": 266500
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.4810080808080808e-05,
+      "loss": 1.0459,
+      "step": 267000
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.47999797979798e-05,
+      "loss": 1.0462,
+      "step": 267500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.4789898989898993e-05,
+      "loss": 1.0448,
+      "step": 268000
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.4779797979797981e-05,
+      "loss": 1.0443,
+      "step": 268500
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.4769696969696971e-05,
+      "loss": 1.0444,
+      "step": 269000
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.475959595959596e-05,
+      "loss": 1.0448,
+      "step": 269500
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.4749494949494952e-05,
+      "loss": 1.0443,
+      "step": 270000
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.473939393939394e-05,
+      "loss": 1.0443,
+      "step": 270500
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.4729292929292931e-05,
+      "loss": 1.0438,
+      "step": 271000
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.471919191919192e-05,
+      "loss": 1.0423,
+      "step": 271500
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.4709090909090911e-05,
+      "loss": 1.0447,
+      "step": 272000
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.46989898989899e-05,
+      "loss": 1.0456,
+      "step": 272500
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.468888888888889e-05,
+      "loss": 1.0425,
+      "step": 273000
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.4678808080808083e-05,
+      "loss": 1.0459,
+      "step": 273500
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.4668707070707071e-05,
+      "loss": 1.0427,
+      "step": 274000
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.4658606060606061e-05,
+      "loss": 1.0457,
+      "step": 274500
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.464850505050505e-05,
+      "loss": 1.042,
+      "step": 275000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.4638424242424242e-05,
+      "loss": 1.0422,
+      "step": 275500
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.4628323232323234e-05,
+      "loss": 1.0435,
+      "step": 276000
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.4618222222222223e-05,
+      "loss": 1.0411,
+      "step": 276500
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.4608121212121213e-05,
+      "loss": 1.0417,
+      "step": 277000
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.4598020202020202e-05,
+      "loss": 1.0417,
+      "step": 277500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.4587939393939394e-05,
+      "loss": 1.0401,
+      "step": 278000
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.4577838383838386e-05,
+      "loss": 1.0435,
+      "step": 278500
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.4567737373737375e-05,
+      "loss": 1.0426,
+      "step": 279000
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.4557636363636365e-05,
+      "loss": 1.043,
+      "step": 279500
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.4547535353535353e-05,
+      "loss": 1.0397,
+      "step": 280000
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.4537454545454546e-05,
+      "loss": 1.0392,
+      "step": 280500
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.4527353535353536e-05,
+      "loss": 1.0396,
+      "step": 281000
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.4517252525252525e-05,
+      "loss": 1.0418,
+      "step": 281500
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.7950182658039489,
+      "eval_loss": 0.8657459020614624,
+      "eval_runtime": 55.7371,
+      "eval_samples_per_second": 928.358,
+      "eval_steps_per_second": 14.515,
+      "step": 281508
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.4507151515151517e-05,
+      "loss": 1.0399,
+      "step": 282000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.4497090909090909e-05,
+      "loss": 1.0405,
+      "step": 282500
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.4486989898989901e-05,
+      "loss": 1.0385,
+      "step": 283000
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.447688888888889e-05,
+      "loss": 1.0398,
+      "step": 283500
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.446678787878788e-05,
+      "loss": 1.04,
+      "step": 284000
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.4456686868686869e-05,
+      "loss": 1.0409,
+      "step": 284500
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.444660606060606e-05,
+      "loss": 1.0361,
+      "step": 285000
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.4436505050505051e-05,
+      "loss": 1.0395,
+      "step": 285500
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.4426404040404041e-05,
+      "loss": 1.0348,
+      "step": 286000
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.4416303030303032e-05,
+      "loss": 1.038,
+      "step": 286500
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.440620202020202e-05,
+      "loss": 1.0374,
+      "step": 287000
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.439610101010101e-05,
+      "loss": 1.0388,
+      "step": 287500
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.4386000000000003e-05,
+      "loss": 1.0389,
+      "step": 288000
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.4375898989898991e-05,
+      "loss": 1.0392,
+      "step": 288500
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.4365797979797981e-05,
+      "loss": 1.0402,
+      "step": 289000
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.4355717171717174e-05,
+      "loss": 1.0397,
+      "step": 289500
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.4345616161616162e-05,
+      "loss": 1.0414,
+      "step": 290000
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.4335515151515154e-05,
+      "loss": 1.0377,
+      "step": 290500
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.4325414141414143e-05,
+      "loss": 1.0372,
+      "step": 291000
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.4315313131313133e-05,
+      "loss": 1.0391,
+      "step": 291500
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.4305212121212122e-05,
+      "loss": 1.0353,
+      "step": 292000
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.4295111111111114e-05,
+      "loss": 1.0382,
+      "step": 292500
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.4285030303030304e-05,
+      "loss": 1.0382,
+      "step": 293000
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.4274929292929295e-05,
+      "loss": 1.0375,
+      "step": 293500
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.4264828282828285e-05,
+      "loss": 1.038,
+      "step": 294000
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.4254727272727273e-05,
+      "loss": 1.0384,
+      "step": 294500
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.4244626262626264e-05,
+      "loss": 1.0371,
+      "step": 295000
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.4234545454545456e-05,
+      "loss": 1.0357,
+      "step": 295500
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 1.0359,
+      "step": 296000
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.4214363636363637e-05,
+      "loss": 1.0357,
+      "step": 296500
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.4204262626262629e-05,
+      "loss": 1.0373,
+      "step": 297000
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.4194161616161617e-05,
+      "loss": 1.0358,
+      "step": 297500
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.4184060606060608e-05,
+      "loss": 1.0391,
+      "step": 298000
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.4173959595959596e-05,
+      "loss": 1.0375,
+      "step": 298500
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.4163858585858588e-05,
+      "loss": 1.0372,
+      "step": 299000
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.4153757575757577e-05,
+      "loss": 1.0337,
+      "step": 299500
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.4143656565656567e-05,
+      "loss": 1.0338,
+      "step": 300000
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.413357575757576e-05,
+      "loss": 1.0357,
+      "step": 300500
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.4123474747474748e-05,
+      "loss": 1.0351,
+      "step": 301000
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.4113373737373738e-05,
+      "loss": 1.0344,
+      "step": 301500
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.4103272727272729e-05,
+      "loss": 1.0353,
+      "step": 302000
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.4093171717171719e-05,
+      "loss": 1.0364,
+      "step": 302500
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.4083070707070707e-05,
+      "loss": 1.0339,
+      "step": 303000
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.4072969696969698e-05,
+      "loss": 1.0343,
+      "step": 303500
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.406288888888889e-05,
+      "loss": 1.0342,
+      "step": 304000
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.4052787878787879e-05,
+      "loss": 1.0342,
+      "step": 304500
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.7953241575126613,
+      "eval_loss": 0.8643110990524292,
+      "eval_runtime": 55.5899,
+      "eval_samples_per_second": 930.817,
+      "eval_steps_per_second": 14.553,
+      "step": 304967
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.404268686868687e-05,
+      "loss": 1.0349,
+      "step": 305000
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.4032585858585859e-05,
+      "loss": 1.0342,
+      "step": 305500
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.402248484848485e-05,
+      "loss": 1.0372,
+      "step": 306000
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.4012383838383838e-05,
+      "loss": 1.0327,
+      "step": 306500
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.400230303030303e-05,
+      "loss": 1.0319,
+      "step": 307000
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.3992202020202022e-05,
+      "loss": 1.0311,
+      "step": 307500
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.398210101010101e-05,
+      "loss": 1.0347,
+      "step": 308000
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.3972000000000001e-05,
+      "loss": 1.0326,
+      "step": 308500
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.3961898989898991e-05,
+      "loss": 1.032,
+      "step": 309000
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.395179797979798e-05,
+      "loss": 1.0301,
+      "step": 309500
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.3941696969696972e-05,
+      "loss": 1.0335,
+      "step": 310000
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.393159595959596e-05,
+      "loss": 1.0341,
+      "step": 310500
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.392149494949495e-05,
+      "loss": 1.0319,
+      "step": 311000
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.3911414141414143e-05,
+      "loss": 1.0304,
+      "step": 311500
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.3901333333333335e-05,
+      "loss": 1.0307,
+      "step": 312000
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.3891232323232324e-05,
+      "loss": 1.0326,
+      "step": 312500
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.3881131313131316e-05,
+      "loss": 1.0309,
+      "step": 313000
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.3871030303030304e-05,
+      "loss": 1.0308,
+      "step": 313500
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.3860929292929295e-05,
+      "loss": 1.0296,
+      "step": 314000
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.3850828282828283e-05,
+      "loss": 1.0315,
+      "step": 314500
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.3840727272727275e-05,
+      "loss": 1.0309,
+      "step": 315000
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.3830626262626264e-05,
+      "loss": 1.0285,
+      "step": 315500
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.3820525252525254e-05,
+      "loss": 1.028,
+      "step": 316000
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.3810424242424243e-05,
+      "loss": 1.0299,
+      "step": 316500
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.3800343434343435e-05,
+      "loss": 1.0309,
+      "step": 317000
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.3790242424242425e-05,
+      "loss": 1.03,
+      "step": 317500
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.3780141414141414e-05,
+      "loss": 1.0303,
+      "step": 318000
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.3770040404040406e-05,
+      "loss": 1.0321,
+      "step": 318500
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.3759959595959598e-05,
+      "loss": 1.03,
+      "step": 319000
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.374987878787879e-05,
+      "loss": 1.0291,
+      "step": 319500
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.3739777777777779e-05,
+      "loss": 1.0298,
+      "step": 320000
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.372967676767677e-05,
+      "loss": 1.0319,
+      "step": 320500
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.3719575757575758e-05,
+      "loss": 1.0312,
+      "step": 321000
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.370947474747475e-05,
+      "loss": 1.0281,
+      "step": 321500
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.3699373737373738e-05,
+      "loss": 1.0289,
+      "step": 322000
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.3689272727272729e-05,
+      "loss": 1.0286,
+      "step": 322500
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.3679171717171717e-05,
+      "loss": 1.0276,
+      "step": 323000
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.366909090909091e-05,
+      "loss": 1.0276,
+      "step": 323500
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.36589898989899e-05,
+      "loss": 1.0285,
+      "step": 324000
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.3648888888888888e-05,
+      "loss": 1.0319,
+      "step": 324500
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.363878787878788e-05,
+      "loss": 1.0303,
+      "step": 325000
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.3628686868686869e-05,
+      "loss": 1.0293,
+      "step": 325500
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.361858585858586e-05,
+      "loss": 1.0275,
+      "step": 326000
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.3608484848484848e-05,
+      "loss": 1.0286,
+      "step": 326500
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.359838383838384e-05,
+      "loss": 1.0293,
+      "step": 327000
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.3588282828282828e-05,
+      "loss": 1.0291,
+      "step": 327500
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.3578222222222224e-05,
+      "loss": 1.0286,
+      "step": 328000
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.7967256149708635,
+      "eval_loss": 0.8558052182197571,
+      "eval_runtime": 55.2402,
+      "eval_samples_per_second": 936.709,
+      "eval_steps_per_second": 14.645,
+      "step": 328426
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.3568121212121213e-05,
+      "loss": 1.0285,
+      "step": 328500
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.3558040404040405e-05,
+      "loss": 1.0263,
+      "step": 329000
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.3547939393939396e-05,
+      "loss": 1.0277,
+      "step": 329500
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.3537838383838384e-05,
+      "loss": 1.027,
+      "step": 330000
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.3527737373737374e-05,
+      "loss": 1.0258,
+      "step": 330500
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.3517636363636365e-05,
+      "loss": 1.0269,
+      "step": 331000
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.3507535353535355e-05,
+      "loss": 1.0271,
+      "step": 331500
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.3497434343434344e-05,
+      "loss": 1.0289,
+      "step": 332000
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.3487333333333334e-05,
+      "loss": 1.028,
+      "step": 332500
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.3477232323232326e-05,
+      "loss": 1.0281,
+      "step": 333000
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.3467131313131314e-05,
+      "loss": 1.0253,
+      "step": 333500
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.3457030303030305e-05,
+      "loss": 1.0246,
+      "step": 334000
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.3446929292929293e-05,
+      "loss": 1.0281,
+      "step": 334500
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.3436828282828285e-05,
+      "loss": 1.0275,
+      "step": 335000
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.3426747474747478e-05,
+      "loss": 1.0257,
+      "step": 335500
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.3416646464646466e-05,
+      "loss": 1.0271,
+      "step": 336000
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.3406545454545456e-05,
+      "loss": 1.0253,
+      "step": 336500
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.3396484848484849e-05,
+      "loss": 1.0237,
+      "step": 337000
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.338638383838384e-05,
+      "loss": 1.0253,
+      "step": 337500
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.337628282828283e-05,
+      "loss": 1.0267,
+      "step": 338000
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.336618181818182e-05,
+      "loss": 1.0251,
+      "step": 338500
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.3356080808080808e-05,
+      "loss": 1.0238,
+      "step": 339000
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.33459797979798e-05,
+      "loss": 1.0253,
+      "step": 339500
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.3335878787878789e-05,
+      "loss": 1.0264,
+      "step": 340000
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.332577777777778e-05,
+      "loss": 1.0259,
+      "step": 340500
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.3315676767676768e-05,
+      "loss": 1.0235,
+      "step": 341000
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.330557575757576e-05,
+      "loss": 1.0238,
+      "step": 341500
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.3295474747474748e-05,
+      "loss": 1.0234,
+      "step": 342000
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.328539393939394e-05,
+      "loss": 1.0267,
+      "step": 342500
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.3275292929292931e-05,
+      "loss": 1.0272,
+      "step": 343000
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.326519191919192e-05,
+      "loss": 1.0236,
+      "step": 343500
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.3255090909090912e-05,
+      "loss": 1.0255,
+      "step": 344000
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.3245010101010102e-05,
+      "loss": 1.0227,
+      "step": 344500
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.3234909090909092e-05,
+      "loss": 1.0236,
+      "step": 345000
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.3224808080808083e-05,
+      "loss": 1.0234,
+      "step": 345500
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.3214707070707071e-05,
+      "loss": 1.0209,
+      "step": 346000
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.3204606060606062e-05,
+      "loss": 1.0239,
+      "step": 346500
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.319450505050505e-05,
+      "loss": 1.0234,
+      "step": 347000
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.3184424242424242e-05,
+      "loss": 1.0238,
+      "step": 347500
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.3174323232323234e-05,
+      "loss": 1.0268,
+      "step": 348000
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.3164222222222223e-05,
+      "loss": 1.0252,
+      "step": 348500
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.3154121212121213e-05,
+      "loss": 1.0255,
+      "step": 349000
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.3144020202020202e-05,
+      "loss": 1.0239,
+      "step": 349500
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.3133919191919194e-05,
+      "loss": 1.0246,
+      "step": 350000
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.3123818181818182e-05,
+      "loss": 1.0244,
+      "step": 350500
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.3113717171717173e-05,
+      "loss": 1.0245,
+      "step": 351000
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.3103656565656567e-05,
+      "loss": 1.0223,
+      "step": 351500
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.7972933232004333,
+      "eval_loss": 0.8523717522621155,
+      "eval_runtime": 56.3728,
+      "eval_samples_per_second": 917.889,
+      "eval_steps_per_second": 14.351,
+      "step": 351885
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.3093555555555557e-05,
+      "loss": 1.0208,
+      "step": 352000
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.3083454545454546e-05,
+      "loss": 1.022,
+      "step": 352500
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 1.3073353535353536e-05,
+      "loss": 1.0223,
+      "step": 353000
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.3063252525252526e-05,
+      "loss": 1.0209,
+      "step": 353500
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.305319191919192e-05,
+      "loss": 1.0214,
+      "step": 354000
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.3043090909090909e-05,
+      "loss": 1.0206,
+      "step": 354500
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.3032989898989901e-05,
+      "loss": 1.0216,
+      "step": 355000
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.302288888888889e-05,
+      "loss": 1.0211,
+      "step": 355500
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.301278787878788e-05,
+      "loss": 1.0202,
+      "step": 356000
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.3002686868686869e-05,
+      "loss": 1.0204,
+      "step": 356500
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.299260606060606e-05,
+      "loss": 1.0214,
+      "step": 357000
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.2982505050505053e-05,
+      "loss": 1.02,
+      "step": 357500
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.2972404040404041e-05,
+      "loss": 1.0226,
+      "step": 358000
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.2962303030303032e-05,
+      "loss": 1.0229,
+      "step": 358500
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.295220202020202e-05,
+      "loss": 1.0197,
+      "step": 359000
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.294210101010101e-05,
+      "loss": 1.0185,
+      "step": 359500
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.2932020202020203e-05,
+      "loss": 1.0184,
+      "step": 360000
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.2921919191919191e-05,
+      "loss": 1.0221,
+      "step": 360500
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.2911818181818183e-05,
+      "loss": 1.0221,
+      "step": 361000
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.2901717171717172e-05,
+      "loss": 1.0184,
+      "step": 361500
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.2891616161616162e-05,
+      "loss": 1.0183,
+      "step": 362000
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.2881515151515151e-05,
+      "loss": 1.0196,
+      "step": 362500
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.2871414141414143e-05,
+      "loss": 1.0201,
+      "step": 363000
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 1.2861313131313133e-05,
+      "loss": 1.0196,
+      "step": 363500
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.2851232323232324e-05,
+      "loss": 1.0191,
+      "step": 364000
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.2841131313131314e-05,
+      "loss": 1.0203,
+      "step": 364500
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.2831030303030306e-05,
+      "loss": 1.0181,
+      "step": 365000
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.2820929292929295e-05,
+      "loss": 1.0183,
+      "step": 365500
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.2810848484848487e-05,
+      "loss": 1.0218,
+      "step": 366000
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.2800747474747475e-05,
+      "loss": 1.0191,
+      "step": 366500
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.2790646464646466e-05,
+      "loss": 1.0191,
+      "step": 367000
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.2780545454545456e-05,
+      "loss": 1.0192,
+      "step": 367500
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.2770444444444445e-05,
+      "loss": 1.0202,
+      "step": 368000
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.2760343434343437e-05,
+      "loss": 1.0182,
+      "step": 368500
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.2750242424242425e-05,
+      "loss": 1.0194,
+      "step": 369000
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.2740141414141415e-05,
+      "loss": 1.0196,
+      "step": 369500
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.2730040404040404e-05,
+      "loss": 1.0168,
+      "step": 370000
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.2719939393939396e-05,
+      "loss": 1.0185,
+      "step": 370500
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.2709858585858588e-05,
+      "loss": 1.0166,
+      "step": 371000
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.2699757575757577e-05,
+      "loss": 1.0197,
+      "step": 371500
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.2689676767676769e-05,
+      "loss": 1.0172,
+      "step": 372000
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.267957575757576e-05,
+      "loss": 1.0179,
+      "step": 372500
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.2669474747474748e-05,
+      "loss": 1.0174,
+      "step": 373000
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.2659373737373738e-05,
+      "loss": 1.0197,
+      "step": 373500
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.2649272727272729e-05,
+      "loss": 1.0155,
+      "step": 374000
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.2639171717171719e-05,
+      "loss": 1.0178,
+      "step": 374500
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.2629090909090911e-05,
+      "loss": 1.0189,
+      "step": 375000
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.7982388277111986,
+      "eval_loss": 0.8482341170310974,
+      "eval_runtime": 55.8333,
+      "eval_samples_per_second": 926.759,
+      "eval_steps_per_second": 14.49,
+      "step": 375344
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 1.26189898989899e-05,
+      "loss": 1.0191,
+      "step": 375500
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 1.260888888888889e-05,
+      "loss": 1.0175,
+      "step": 376000
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 1.2598787878787879e-05,
+      "loss": 1.0156,
+      "step": 376500
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 1.258868686868687e-05,
+      "loss": 1.0147,
+      "step": 377000
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 1.2578585858585859e-05,
+      "loss": 1.0176,
+      "step": 377500
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 1.256848484848485e-05,
+      "loss": 1.0191,
+      "step": 378000
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 1.2558383838383838e-05,
+      "loss": 1.017,
+      "step": 378500
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 1.254828282828283e-05,
+      "loss": 1.0169,
+      "step": 379000
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 1.2538181818181819e-05,
+      "loss": 1.0166,
+      "step": 379500
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 1.2528080808080809e-05,
+      "loss": 1.0149,
+      "step": 380000
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 1.2517979797979797e-05,
+      "loss": 1.0146,
+      "step": 380500
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.250789898989899e-05,
+      "loss": 1.0169,
+      "step": 381000
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.2497818181818182e-05,
+      "loss": 1.0184,
+      "step": 381500
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.2487737373737374e-05,
+      "loss": 1.0132,
+      "step": 382000
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.2477636363636364e-05,
+      "loss": 1.014,
+      "step": 382500
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.2467535353535353e-05,
+      "loss": 1.0151,
+      "step": 383000
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.2457434343434345e-05,
+      "loss": 1.0158,
+      "step": 383500
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.2447333333333334e-05,
+      "loss": 1.0157,
+      "step": 384000
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.2437232323232324e-05,
+      "loss": 1.0158,
+      "step": 384500
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.2427131313131313e-05,
+      "loss": 1.0158,
+      "step": 385000
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 1.2417050505050505e-05,
+      "loss": 1.0175,
+      "step": 385500
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.2406949494949497e-05,
+      "loss": 1.014,
+      "step": 386000
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.2396848484848485e-05,
+      "loss": 1.0139,
+      "step": 386500
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.2386747474747476e-05,
+      "loss": 1.0141,
+      "step": 387000
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.2376666666666668e-05,
+      "loss": 1.0154,
+      "step": 387500
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 1.2366565656565656e-05,
+      "loss": 1.0151,
+      "step": 388000
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.2356464646464647e-05,
+      "loss": 1.0153,
+      "step": 388500
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.2346363636363637e-05,
+      "loss": 1.0126,
+      "step": 389000
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.2336262626262627e-05,
+      "loss": 1.0155,
+      "step": 389500
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.2326161616161618e-05,
+      "loss": 1.0132,
+      "step": 390000
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 1.2316060606060606e-05,
+      "loss": 1.0153,
+      "step": 390500
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.2305959595959598e-05,
+      "loss": 1.0147,
+      "step": 391000
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.2295858585858587e-05,
+      "loss": 1.017,
+      "step": 391500
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.2285757575757577e-05,
+      "loss": 1.0127,
+      "step": 392000
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.2275656565656566e-05,
+      "loss": 1.0172,
+      "step": 392500
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.2265555555555558e-05,
+      "loss": 1.012,
+      "step": 393000
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.225547474747475e-05,
+      "loss": 1.0133,
+      "step": 393500
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.2245393939393942e-05,
+      "loss": 1.0132,
+      "step": 394000
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.223529292929293e-05,
+      "loss": 1.0132,
+      "step": 394500
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.2225191919191921e-05,
+      "loss": 1.0158,
+      "step": 395000
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.221509090909091e-05,
+      "loss": 1.015,
+      "step": 395500
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.22049898989899e-05,
+      "loss": 1.0099,
+      "step": 396000
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 1.219488888888889e-05,
+      "loss": 1.0159,
+      "step": 396500
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.218478787878788e-05,
+      "loss": 1.0138,
+      "step": 397000
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.2174686868686869e-05,
+      "loss": 1.0121,
+      "step": 397500
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.2164606060606061e-05,
+      "loss": 1.0117,
+      "step": 398000
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 1.2154505050505052e-05,
+      "loss": 1.0108,
+      "step": 398500
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.7986988582872165,
+      "eval_loss": 0.8448593020439148,
+      "eval_runtime": 55.9245,
+      "eval_samples_per_second": 925.248,
+      "eval_steps_per_second": 14.466,
+      "step": 398803
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 1.214440404040404e-05,
+      "loss": 1.0112,
+      "step": 399000
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.2134303030303032e-05,
+      "loss": 1.0113,
+      "step": 399500
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.212420202020202e-05,
+      "loss": 1.0097,
+      "step": 400000
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.2114101010101011e-05,
+      "loss": 1.0117,
+      "step": 400500
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.2104e-05,
+      "loss": 1.0093,
+      "step": 401000
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.2093898989898992e-05,
+      "loss": 1.0133,
+      "step": 401500
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.2083818181818184e-05,
+      "loss": 1.0139,
+      "step": 402000
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.2073717171717172e-05,
+      "loss": 1.0112,
+      "step": 402500
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.2063616161616163e-05,
+      "loss": 1.014,
+      "step": 403000
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.2053515151515151e-05,
+      "loss": 1.0126,
+      "step": 403500
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.2043434343434344e-05,
+      "loss": 1.0113,
+      "step": 404000
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 1.2033333333333334e-05,
+      "loss": 1.0117,
+      "step": 404500
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 1.2023232323232324e-05,
+      "loss": 1.0095,
+      "step": 405000
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 1.2013131313131314e-05,
+      "loss": 1.0125,
+      "step": 405500
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.2003030303030303e-05,
+      "loss": 1.0095,
+      "step": 406000
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.1992929292929293e-05,
+      "loss": 1.0092,
+      "step": 406500
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 1.1982828282828282e-05,
+      "loss": 1.01,
+      "step": 407000
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 1.1972727272727274e-05,
+      "loss": 1.0127,
+      "step": 407500
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.1962646464646466e-05,
+      "loss": 1.0117,
+      "step": 408000
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.1952545454545455e-05,
+      "loss": 1.01,
+      "step": 408500
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.1942444444444445e-05,
+      "loss": 1.01,
+      "step": 409000
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 1.1932363636363637e-05,
+      "loss": 1.011,
+      "step": 409500
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.1922262626262626e-05,
+      "loss": 1.0087,
+      "step": 410000
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.1912161616161618e-05,
+      "loss": 1.0086,
+      "step": 410500
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.1902060606060608e-05,
+      "loss": 1.0098,
+      "step": 411000
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.1891959595959597e-05,
+      "loss": 1.0111,
+      "step": 411500
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.1881858585858587e-05,
+      "loss": 1.0059,
+      "step": 412000
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.1871757575757577e-05,
+      "loss": 1.0098,
+      "step": 412500
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.1861676767676768e-05,
+      "loss": 1.0108,
+      "step": 413000
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.185157575757576e-05,
+      "loss": 1.0102,
+      "step": 413500
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.1841474747474748e-05,
+      "loss": 1.0087,
+      "step": 414000
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.183139393939394e-05,
+      "loss": 1.0103,
+      "step": 414500
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.1821292929292931e-05,
+      "loss": 1.0067,
+      "step": 415000
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.181119191919192e-05,
+      "loss": 1.0111,
+      "step": 415500
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.1801090909090912e-05,
+      "loss": 1.0092,
+      "step": 416000
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.17909898989899e-05,
+      "loss": 1.0083,
+      "step": 416500
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.178088888888889e-05,
+      "loss": 1.0114,
+      "step": 417000
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.1770787878787879e-05,
+      "loss": 1.0088,
+      "step": 417500
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.1760686868686871e-05,
+      "loss": 1.0108,
+      "step": 418000
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.175058585858586e-05,
+      "loss": 1.0089,
+      "step": 418500
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.1740505050505052e-05,
+      "loss": 1.01,
+      "step": 419000
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.1730404040404042e-05,
+      "loss": 1.0104,
+      "step": 419500
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.172030303030303e-05,
+      "loss": 1.0084,
+      "step": 420000
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.1710202020202021e-05,
+      "loss": 1.0103,
+      "step": 420500
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.170010101010101e-05,
+      "loss": 1.0088,
+      "step": 421000
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.1690020202020202e-05,
+      "loss": 1.0079,
+      "step": 421500
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.1679919191919194e-05,
+      "loss": 1.0107,
+      "step": 422000
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.7993050643486012,
+      "eval_loss": 0.8416274189949036,
+      "eval_runtime": 55.186,
+      "eval_samples_per_second": 937.629,
+      "eval_steps_per_second": 14.66,
+      "step": 422262
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.1669818181818182e-05,
+      "loss": 1.01,
+      "step": 422500
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.1659717171717173e-05,
+      "loss": 1.0059,
+      "step": 423000
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.1649616161616161e-05,
+      "loss": 1.0052,
+      "step": 423500
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.1639515151515153e-05,
+      "loss": 1.0102,
+      "step": 424000
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.1629414141414142e-05,
+      "loss": 1.0107,
+      "step": 424500
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.1619333333333334e-05,
+      "loss": 1.0075,
+      "step": 425000
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 1.1609252525252526e-05,
+      "loss": 1.0068,
+      "step": 425500
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 1.1599151515151517e-05,
+      "loss": 1.0076,
+      "step": 426000
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 1.1589050505050505e-05,
+      "loss": 1.0095,
+      "step": 426500
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 1.1578949494949496e-05,
+      "loss": 1.0071,
+      "step": 427000
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 1.1568848484848486e-05,
+      "loss": 1.0062,
+      "step": 427500
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 1.1558747474747476e-05,
+      "loss": 1.0068,
+      "step": 428000
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 1.1548646464646465e-05,
+      "loss": 1.0085,
+      "step": 428500
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 1.1538545454545455e-05,
+      "loss": 1.0087,
+      "step": 429000
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 1.1528444444444444e-05,
+      "loss": 1.0035,
+      "step": 429500
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 1.1518343434343436e-05,
+      "loss": 1.0041,
+      "step": 430000
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 1.1508262626262628e-05,
+      "loss": 1.0041,
+      "step": 430500
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 1.1498161616161616e-05,
+      "loss": 1.0064,
+      "step": 431000
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 1.1488060606060607e-05,
+      "loss": 1.0061,
+      "step": 431500
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 1.1477959595959599e-05,
+      "loss": 1.002,
+      "step": 432000
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 1.1467898989898991e-05,
+      "loss": 1.0086,
+      "step": 432500
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 1.145779797979798e-05,
+      "loss": 1.0076,
+      "step": 433000
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 1.144769696969697e-05,
+      "loss": 1.0075,
+      "step": 433500
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 1.143759595959596e-05,
+      "loss": 1.0069,
+      "step": 434000
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 1.142749494949495e-05,
+      "loss": 1.0044,
+      "step": 434500
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 1.141739393939394e-05,
+      "loss": 1.0053,
+      "step": 435000
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 1.140729292929293e-05,
+      "loss": 1.0075,
+      "step": 435500
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 1.1397191919191921e-05,
+      "loss": 1.0038,
+      "step": 436000
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 1.138711111111111e-05,
+      "loss": 1.007,
+      "step": 436500
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 1.1377010101010102e-05,
+      "loss": 1.0048,
+      "step": 437000
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 1.1366909090909093e-05,
+      "loss": 1.0061,
+      "step": 437500
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 1.1356808080808081e-05,
+      "loss": 1.0065,
+      "step": 438000
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 1.1346727272727273e-05,
+      "loss": 1.0043,
+      "step": 438500
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 1.1336626262626265e-05,
+      "loss": 1.0048,
+      "step": 439000
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 1.1326525252525254e-05,
+      "loss": 1.0043,
+      "step": 439500
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 1.1316424242424244e-05,
+      "loss": 1.0078,
+      "step": 440000
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 1.1306323232323233e-05,
+      "loss": 1.0058,
+      "step": 440500
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 1.1296222222222223e-05,
+      "loss": 1.0053,
+      "step": 441000
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 1.1286121212121213e-05,
+      "loss": 1.0054,
+      "step": 441500
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 1.1276020202020204e-05,
+      "loss": 1.0047,
+      "step": 442000
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 1.1265939393939396e-05,
+      "loss": 1.0058,
+      "step": 442500
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 1.1255838383838385e-05,
+      "loss": 1.0073,
+      "step": 443000
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 1.1245737373737375e-05,
+      "loss": 1.0035,
+      "step": 443500
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 1.1235636363636363e-05,
+      "loss": 1.0029,
+      "step": 444000
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 1.1225555555555556e-05,
+      "loss": 1.0056,
+      "step": 444500
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 1.1215454545454548e-05,
+      "loss": 1.0058,
+      "step": 445000
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 1.1205353535353536e-05,
+      "loss": 1.003,
+      "step": 445500
+    },
+    {
+      "epoch": 19.0,
+      "eval_accuracy": 0.7996911332179277,
+      "eval_loss": 0.839302122592926,
+      "eval_runtime": 55.348,
+      "eval_samples_per_second": 934.885,
+      "eval_steps_per_second": 14.617,
+      "step": 445721
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 1.1195252525252527e-05,
+      "loss": 1.0028,
+      "step": 446000
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.1185171717171719e-05,
+      "loss": 1.0039,
+      "step": 446500
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 1.1175070707070707e-05,
+      "loss": 1.0028,
+      "step": 447000
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 1.1164969696969698e-05,
+      "loss": 1.0063,
+      "step": 447500
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 1.1154868686868688e-05,
+      "loss": 1.0027,
+      "step": 448000
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 1.1144767676767678e-05,
+      "loss": 1.0024,
+      "step": 448500
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 1.1134666666666667e-05,
+      "loss": 1.0017,
+      "step": 449000
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 1.1124565656565657e-05,
+      "loss": 1.0044,
+      "step": 449500
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 1.1114464646464646e-05,
+      "loss": 1.0034,
+      "step": 450000
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 1.1104383838383838e-05,
+      "loss": 1.0033,
+      "step": 450500
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 1.109428282828283e-05,
+      "loss": 1.003,
+      "step": 451000
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 1.1084181818181819e-05,
+      "loss": 1.0045,
+      "step": 451500
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 1.1074080808080809e-05,
+      "loss": 1.0015,
+      "step": 452000
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 1.1063979797979797e-05,
+      "loss": 1.0032,
+      "step": 452500
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 1.105387878787879e-05,
+      "loss": 1.0042,
+      "step": 453000
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 1.1043797979797982e-05,
+      "loss": 1.0013,
+      "step": 453500
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.103369696969697e-05,
+      "loss": 1.0024,
+      "step": 454000
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 1.102359595959596e-05,
+      "loss": 1.0043,
+      "step": 454500
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 1.101349494949495e-05,
+      "loss": 1.0017,
+      "step": 455000
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 1.1003414141414141e-05,
+      "loss": 1.0022,
+      "step": 455500
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.0993313131313132e-05,
+      "loss": 1.004,
+      "step": 456000
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.0983212121212122e-05,
+      "loss": 1.002,
+      "step": 456500
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 1.0973111111111112e-05,
+      "loss": 1.003,
+      "step": 457000
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 1.0963010101010101e-05,
+      "loss": 1.0013,
+      "step": 457500
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 1.0952909090909091e-05,
+      "loss": 1.002,
+      "step": 458000
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 1.0942808080808083e-05,
+      "loss": 1.001,
+      "step": 458500
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 1.0932707070707072e-05,
+      "loss": 1.0019,
+      "step": 459000
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 1.0922626262626264e-05,
+      "loss": 1.0034,
+      "step": 459500
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 1.0912545454545456e-05,
+      "loss": 1.0007,
+      "step": 460000
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 1.0902444444444445e-05,
+      "loss": 1.0018,
+      "step": 460500
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 1.0892343434343435e-05,
+      "loss": 1.0,
+      "step": 461000
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.0882262626262627e-05,
+      "loss": 1.0018,
+      "step": 461500
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.0872161616161616e-05,
+      "loss": 1.0026,
+      "step": 462000
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.0862060606060606e-05,
+      "loss": 1.002,
+      "step": 462500
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 1.0851959595959597e-05,
+      "loss": 1.001,
+      "step": 463000
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 1.0841858585858587e-05,
+      "loss": 1.0005,
+      "step": 463500
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.0831757575757577e-05,
+      "loss": 1.0038,
+      "step": 464000
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.0821656565656566e-05,
+      "loss": 1.0027,
+      "step": 464500
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.0811555555555558e-05,
+      "loss": 1.0008,
+      "step": 465000
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.0801454545454546e-05,
+      "loss": 1.0015,
+      "step": 465500
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 1.0791353535353537e-05,
+      "loss": 1.0004,
+      "step": 466000
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 1.0781252525252525e-05,
+      "loss": 1.0018,
+      "step": 466500
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 1.0771171717171717e-05,
+      "loss": 1.0025,
+      "step": 467000
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 1.076107070707071e-05,
+      "loss": 1.0013,
+      "step": 467500
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.0750969696969698e-05,
+      "loss": 1.0011,
+      "step": 468000
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.0740868686868688e-05,
+      "loss": 1.0021,
+      "step": 468500
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 1.073078787878788e-05,
+      "loss": 1.0013,
+      "step": 469000
+    },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 0.8004582095158734,
+      "eval_loss": 0.8350439667701721,
+      "eval_runtime": 55.8457,
+      "eval_samples_per_second": 926.553,
+      "eval_steps_per_second": 14.486,
+      "step": 469180
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 1.0720686868686869e-05,
+      "loss": 0.9998,
+      "step": 469500
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 1.071058585858586e-05,
+      "loss": 0.9982,
+      "step": 470000
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 1.070048484848485e-05,
+      "loss": 0.9999,
+      "step": 470500
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 1.069038383838384e-05,
+      "loss": 0.9975,
+      "step": 471000
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 1.0680282828282829e-05,
+      "loss": 1.001,
+      "step": 471500
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 1.067020202020202e-05,
+      "loss": 0.9995,
+      "step": 472000
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 1.0660101010101011e-05,
+      "loss": 1.0025,
+      "step": 472500
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 1.065e-05,
+      "loss": 0.9976,
+      "step": 473000
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 1.0639898989898992e-05,
+      "loss": 1.0008,
+      "step": 473500
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 1.062979797979798e-05,
+      "loss": 0.9997,
+      "step": 474000
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 1.061969696969697e-05,
+      "loss": 0.9997,
+      "step": 474500
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 1.0609595959595959e-05,
+      "loss": 1.0011,
+      "step": 475000
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 1.0599494949494951e-05,
+      "loss": 1.0001,
+      "step": 475500
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 1.058939393939394e-05,
+      "loss": 0.9992,
+      "step": 476000
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 1.0579313131313132e-05,
+      "loss": 0.9995,
+      "step": 476500
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 1.0569212121212122e-05,
+      "loss": 1.0016,
+      "step": 477000
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 1.0559131313131314e-05,
+      "loss": 0.9996,
+      "step": 477500
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 1.0549030303030303e-05,
+      "loss": 0.9963,
+      "step": 478000
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 1.0538949494949495e-05,
+      "loss": 1.0001,
+      "step": 478500
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 1.0528848484848486e-05,
+      "loss": 1.0007,
+      "step": 479000
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 1.0518747474747474e-05,
+      "loss": 1.0007,
+      "step": 479500
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 1.0508646464646466e-05,
+      "loss": 1.0001,
+      "step": 480000
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 1.0498545454545455e-05,
+      "loss": 0.9998,
+      "step": 480500
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 1.0488444444444445e-05,
+      "loss": 0.9978,
+      "step": 481000
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 1.0478363636363637e-05,
+      "loss": 0.999,
+      "step": 481500
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 1.0468262626262626e-05,
+      "loss": 0.9964,
+      "step": 482000
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 1.0458161616161618e-05,
+      "loss": 0.9987,
+      "step": 482500
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 1.0448060606060606e-05,
+      "loss": 0.9986,
+      "step": 483000
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 1.0437959595959597e-05,
+      "loss": 0.9975,
+      "step": 483500
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 1.0427858585858585e-05,
+      "loss": 0.9992,
+      "step": 484000
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 1.0417757575757577e-05,
+      "loss": 0.998,
+      "step": 484500
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 1.0407656565656568e-05,
+      "loss": 0.9999,
+      "step": 485000
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 1.0397555555555556e-05,
+      "loss": 0.9962,
+      "step": 485500
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 1.0387454545454547e-05,
+      "loss": 0.9965,
+      "step": 486000
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 1.0377353535353537e-05,
+      "loss": 0.999,
+      "step": 486500
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 1.0367252525252527e-05,
+      "loss": 0.998,
+      "step": 487000
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 1.0357151515151516e-05,
+      "loss": 0.9966,
+      "step": 487500
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 1.0347070707070708e-05,
+      "loss": 0.9989,
+      "step": 488000
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 1.03369898989899e-05,
+      "loss": 0.9989,
+      "step": 488500
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 1.032688888888889e-05,
+      "loss": 0.9984,
+      "step": 489000
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 1.0316787878787879e-05,
+      "loss": 0.9978,
+      "step": 489500
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 1.0306686868686871e-05,
+      "loss": 0.9967,
+      "step": 490000
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 1.0296606060606063e-05,
+      "loss": 0.9967,
+      "step": 490500
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 1.0286505050505052e-05,
+      "loss": 1.0011,
+      "step": 491000
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 1.0276404040404042e-05,
+      "loss": 0.9987,
+      "step": 491500
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 1.026630303030303e-05,
+      "loss": 0.9957,
+      "step": 492000
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 1.0256202020202021e-05,
+      "loss": 0.9968,
+      "step": 492500
+    },
+    {
+      "epoch": 21.0,
+      "eval_accuracy": 0.8014651902027975,
+      "eval_loss": 0.8311926126480103,
+      "eval_runtime": 55.7332,
+      "eval_samples_per_second": 928.424,
+      "eval_steps_per_second": 14.516,
+      "step": 492639
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 1.0246101010101011e-05,
+      "loss": 0.9966,
+      "step": 493000
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 1.0236020202020202e-05,
+      "loss": 0.9974,
+      "step": 493500
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 1.0225919191919194e-05,
+      "loss": 0.9945,
+      "step": 494000
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 1.0215818181818182e-05,
+      "loss": 0.996,
+      "step": 494500
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 1.0205737373737375e-05,
+      "loss": 0.9977,
+      "step": 495000
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 1.0195636363636365e-05,
+      "loss": 0.9942,
+      "step": 495500
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 1.0185535353535354e-05,
+      "loss": 0.9952,
+      "step": 496000
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 1.0175434343434346e-05,
+      "loss": 0.9938,
+      "step": 496500
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 1.0165333333333334e-05,
+      "loss": 0.9954,
+      "step": 497000
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 1.0155232323232324e-05,
+      "loss": 0.996,
+      "step": 497500
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 1.0145131313131313e-05,
+      "loss": 0.9982,
+      "step": 498000
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 1.0135030303030305e-05,
+      "loss": 0.9982,
+      "step": 498500
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 1.0124929292929294e-05,
+      "loss": 0.9973,
+      "step": 499000
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 1.0114828282828284e-05,
+      "loss": 0.9944,
+      "step": 499500
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 1.0104747474747476e-05,
+      "loss": 0.9962,
+      "step": 500000
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 1.0094646464646465e-05,
+      "loss": 0.9948,
+      "step": 500500
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 1.0084545454545455e-05,
+      "loss": 0.9953,
+      "step": 501000
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 1.0074444444444445e-05,
+      "loss": 0.9972,
+      "step": 501500
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 1.0064343434343436e-05,
+      "loss": 0.9925,
+      "step": 502000
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 1.0054242424242424e-05,
+      "loss": 0.9952,
+      "step": 502500
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 1.0044141414141414e-05,
+      "loss": 0.9942,
+      "step": 503000
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 1.0034040404040403e-05,
+      "loss": 0.9936,
+      "step": 503500
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 1.0023959595959595e-05,
+      "loss": 0.9957,
+      "step": 504000
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 1.0013858585858587e-05,
+      "loss": 0.9936,
+      "step": 504500
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 1.0003757575757576e-05,
+      "loss": 0.9935,
+      "step": 505000
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 9.993656565656566e-06,
+      "loss": 0.9965,
+      "step": 505500
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 9.98359595959596e-06,
+      "loss": 0.9972,
+      "step": 506000
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 9.97349494949495e-06,
+      "loss": 0.9949,
+      "step": 506500
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 9.963393939393941e-06,
+      "loss": 0.9956,
+      "step": 507000
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 9.95329292929293e-06,
+      "loss": 0.9955,
+      "step": 507500
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 9.94319191919192e-06,
+      "loss": 0.9933,
+      "step": 508000
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 9.933111111111112e-06,
+      "loss": 0.9952,
+      "step": 508500
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 9.923010101010102e-06,
+      "loss": 0.9968,
+      "step": 509000
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 9.912909090909093e-06,
+      "loss": 0.9963,
+      "step": 509500
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 9.902828282828283e-06,
+      "loss": 0.9948,
+      "step": 510000
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 9.892727272727273e-06,
+      "loss": 0.9942,
+      "step": 510500
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 9.882626262626264e-06,
+      "loss": 0.9929,
+      "step": 511000
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 9.872525252525254e-06,
+      "loss": 0.995,
+      "step": 511500
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 9.862424242424243e-06,
+      "loss": 0.995,
+      "step": 512000
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 9.852323232323233e-06,
+      "loss": 0.9948,
+      "step": 512500
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 9.842222222222223e-06,
+      "loss": 0.9949,
+      "step": 513000
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 9.832121212121214e-06,
+      "loss": 0.9927,
+      "step": 513500
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 9.822020202020202e-06,
+      "loss": 0.9966,
+      "step": 514000
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 9.811919191919192e-06,
+      "loss": 0.9946,
+      "step": 514500
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 9.801818181818183e-06,
+      "loss": 0.9976,
+      "step": 515000
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 9.791717171717173e-06,
+      "loss": 0.9931,
+      "step": 515500
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 9.781616161616162e-06,
+      "loss": 0.9945,
+      "step": 516000
+    },
+    {
+      "epoch": 22.0,
+      "eval_accuracy": 0.8020606570385413,
+      "eval_loss": 0.8275821208953857,
+      "eval_runtime": 55.6689,
+      "eval_samples_per_second": 929.495,
+      "eval_steps_per_second": 14.532,
+      "step": 516098
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 9.771515151515152e-06,
+      "loss": 0.9944,
+      "step": 516500
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 9.761414141414142e-06,
+      "loss": 0.9923,
+      "step": 517000
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 9.751333333333334e-06,
+      "loss": 0.9928,
+      "step": 517500
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 9.741232323232323e-06,
+      "loss": 0.9926,
+      "step": 518000
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 9.731151515151515e-06,
+      "loss": 0.9923,
+      "step": 518500
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 9.721050505050506e-06,
+      "loss": 0.9971,
+      "step": 519000
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 9.710949494949496e-06,
+      "loss": 0.9932,
+      "step": 519500
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 9.700848484848484e-06,
+      "loss": 0.9936,
+      "step": 520000
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 9.690747474747475e-06,
+      "loss": 0.9927,
+      "step": 520500
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 9.680646464646467e-06,
+      "loss": 0.9917,
+      "step": 521000
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 9.670545454545455e-06,
+      "loss": 0.9939,
+      "step": 521500
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 9.660444444444446e-06,
+      "loss": 0.9915,
+      "step": 522000
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 9.650363636363636e-06,
+      "loss": 0.9931,
+      "step": 522500
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 9.640262626262628e-06,
+      "loss": 0.9929,
+      "step": 523000
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 9.630161616161617e-06,
+      "loss": 0.9928,
+      "step": 523500
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 9.620060606060607e-06,
+      "loss": 0.9913,
+      "step": 524000
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 9.6099797979798e-06,
+      "loss": 0.9911,
+      "step": 524500
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 9.59987878787879e-06,
+      "loss": 0.9947,
+      "step": 525000
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 9.58977777777778e-06,
+      "loss": 0.9934,
+      "step": 525500
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 9.57969696969697e-06,
+      "loss": 0.9937,
+      "step": 526000
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 9.56959595959596e-06,
+      "loss": 0.9919,
+      "step": 526500
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 9.559494949494951e-06,
+      "loss": 0.9928,
+      "step": 527000
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 9.549393939393941e-06,
+      "loss": 0.9918,
+      "step": 527500
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 9.53929292929293e-06,
+      "loss": 0.9933,
+      "step": 528000
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 9.52919191919192e-06,
+      "loss": 0.9944,
+      "step": 528500
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 9.51909090909091e-06,
+      "loss": 0.9929,
+      "step": 529000
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 9.509010101010103e-06,
+      "loss": 0.9904,
+      "step": 529500
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 9.498909090909091e-06,
+      "loss": 0.991,
+      "step": 530000
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 9.488808080808081e-06,
+      "loss": 0.9913,
+      "step": 530500
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 9.478707070707072e-06,
+      "loss": 0.9935,
+      "step": 531000
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 9.468606060606062e-06,
+      "loss": 0.9917,
+      "step": 531500
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 9.45850505050505e-06,
+      "loss": 0.991,
+      "step": 532000
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 9.448404040404041e-06,
+      "loss": 0.9917,
+      "step": 532500
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 9.438303030303031e-06,
+      "loss": 0.9927,
+      "step": 533000
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 9.428202020202022e-06,
+      "loss": 0.9929,
+      "step": 533500
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 9.418121212121212e-06,
+      "loss": 0.9945,
+      "step": 534000
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 9.408040404040404e-06,
+      "loss": 0.9918,
+      "step": 534500
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 9.397939393939395e-06,
+      "loss": 0.9876,
+      "step": 535000
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 9.387838383838385e-06,
+      "loss": 0.9934,
+      "step": 535500
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 9.377737373737375e-06,
+      "loss": 0.9911,
+      "step": 536000
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 9.367636363636364e-06,
+      "loss": 0.9916,
+      "step": 536500
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 9.357535353535354e-06,
+      "loss": 0.9914,
+      "step": 537000
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 9.347434343434344e-06,
+      "loss": 0.992,
+      "step": 537500
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 9.337333333333335e-06,
+      "loss": 0.9909,
+      "step": 538000
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 9.327232323232323e-06,
+      "loss": 0.9918,
+      "step": 538500
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 9.317131313131314e-06,
+      "loss": 0.992,
+      "step": 539000
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 9.307030303030304e-06,
+      "loss": 0.991,
+      "step": 539500
+    },
+    {
+      "epoch": 23.0,
+      "eval_accuracy": 0.8018969414661152,
+      "eval_loss": 0.8275611400604248,
+      "eval_runtime": 55.1567,
+      "eval_samples_per_second": 938.127,
+      "eval_steps_per_second": 14.667,
+      "step": 539557
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 9.296949494949496e-06,
+      "loss": 0.989,
+      "step": 540000
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 9.286848484848485e-06,
+      "loss": 0.9919,
+      "step": 540500
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 9.276747474747475e-06,
+      "loss": 0.9901,
+      "step": 541000
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 9.266646464646465e-06,
+      "loss": 0.9906,
+      "step": 541500
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 9.256545454545455e-06,
+      "loss": 0.9917,
+      "step": 542000
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 9.246444444444446e-06,
+      "loss": 0.9884,
+      "step": 542500
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 9.236363636363636e-06,
+      "loss": 0.9899,
+      "step": 543000
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 9.226262626262627e-06,
+      "loss": 0.9883,
+      "step": 543500
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 9.216161616161617e-06,
+      "loss": 0.9883,
+      "step": 544000
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 9.206060606060607e-06,
+      "loss": 0.9919,
+      "step": 544500
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 9.195959595959597e-06,
+      "loss": 0.9869,
+      "step": 545000
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 9.185858585858588e-06,
+      "loss": 0.9907,
+      "step": 545500
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 9.175777777777778e-06,
+      "loss": 0.9903,
+      "step": 546000
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 9.165676767676769e-06,
+      "loss": 0.9894,
+      "step": 546500
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 9.15559595959596e-06,
+      "loss": 0.9886,
+      "step": 547000
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 9.145494949494951e-06,
+      "loss": 0.9882,
+      "step": 547500
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 9.135393939393941e-06,
+      "loss": 0.989,
+      "step": 548000
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 9.12529292929293e-06,
+      "loss": 0.9889,
+      "step": 548500
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 9.11519191919192e-06,
+      "loss": 0.9874,
+      "step": 549000
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 9.10509090909091e-06,
+      "loss": 0.9894,
+      "step": 549500
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 9.0949898989899e-06,
+      "loss": 0.9896,
+      "step": 550000
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 9.08488888888889e-06,
+      "loss": 0.9905,
+      "step": 550500
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 9.074808080808082e-06,
+      "loss": 0.9904,
+      "step": 551000
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 9.064707070707072e-06,
+      "loss": 0.9909,
+      "step": 551500
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 9.054606060606062e-06,
+      "loss": 0.9893,
+      "step": 552000
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 9.044505050505051e-06,
+      "loss": 0.9897,
+      "step": 552500
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 9.034404040404041e-06,
+      "loss": 0.9856,
+      "step": 553000
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 9.024303030303031e-06,
+      "loss": 0.9887,
+      "step": 553500
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 9.01420202020202e-06,
+      "loss": 0.9889,
+      "step": 554000
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 9.00410101010101e-06,
+      "loss": 0.99,
+      "step": 554500
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 8.994020202020203e-06,
+      "loss": 0.9892,
+      "step": 555000
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 8.983919191919193e-06,
+      "loss": 0.9889,
+      "step": 555500
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 8.973818181818183e-06,
+      "loss": 0.9898,
+      "step": 556000
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 8.963717171717172e-06,
+      "loss": 0.9892,
+      "step": 556500
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 8.953616161616162e-06,
+      "loss": 0.9898,
+      "step": 557000
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 8.943535353535354e-06,
+      "loss": 0.9873,
+      "step": 557500
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 8.933434343434345e-06,
+      "loss": 0.9892,
+      "step": 558000
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 8.923333333333333e-06,
+      "loss": 0.9903,
+      "step": 558500
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 8.913232323232323e-06,
+      "loss": 0.9906,
+      "step": 559000
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 8.903131313131314e-06,
+      "loss": 0.9858,
+      "step": 559500
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 8.893050505050506e-06,
+      "loss": 0.9897,
+      "step": 560000
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 8.882969696969698e-06,
+      "loss": 0.9869,
+      "step": 560500
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 8.872868686868687e-06,
+      "loss": 0.9877,
+      "step": 561000
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 8.862767676767677e-06,
+      "loss": 0.9877,
+      "step": 561500
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 8.852666666666667e-06,
+      "loss": 0.9897,
+      "step": 562000
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 8.842565656565658e-06,
+      "loss": 0.989,
+      "step": 562500
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 8.832464646464646e-06,
+      "loss": 0.9862,
+      "step": 563000
+    },
+    {
+      "epoch": 24.0,
+      "eval_accuracy": 0.802191879946925,
+      "eval_loss": 0.8242408037185669,
+      "eval_runtime": 55.2894,
+      "eval_samples_per_second": 935.876,
+      "eval_steps_per_second": 14.632,
+      "step": 563016
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 8.822363636363637e-06,
+      "loss": 0.988,
+      "step": 563500
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 8.812262626262627e-06,
+      "loss": 0.9873,
+      "step": 564000
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 8.802161616161617e-06,
+      "loss": 0.9868,
+      "step": 564500
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 8.792060606060607e-06,
+      "loss": 0.9884,
+      "step": 565000
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 8.781959595959598e-06,
+      "loss": 0.9866,
+      "step": 565500
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 8.771878787878788e-06,
+      "loss": 0.9867,
+      "step": 566000
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 8.76179797979798e-06,
+      "loss": 0.9881,
+      "step": 566500
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 8.75169696969697e-06,
+      "loss": 0.9882,
+      "step": 567000
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 8.74159595959596e-06,
+      "loss": 0.9878,
+      "step": 567500
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 8.73149494949495e-06,
+      "loss": 0.9887,
+      "step": 568000
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 8.72139393939394e-06,
+      "loss": 0.9847,
+      "step": 568500
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 8.71129292929293e-06,
+      "loss": 0.9884,
+      "step": 569000
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 8.70119191919192e-06,
+      "loss": 0.9885,
+      "step": 569500
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 8.69109090909091e-06,
+      "loss": 0.987,
+      "step": 570000
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 8.6809898989899e-06,
+      "loss": 0.9868,
+      "step": 570500
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 8.67088888888889e-06,
+      "loss": 0.9875,
+      "step": 571000
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 8.66078787878788e-06,
+      "loss": 0.9866,
+      "step": 571500
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 8.65068686868687e-06,
+      "loss": 0.986,
+      "step": 572000
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 8.640626262626264e-06,
+      "loss": 0.9874,
+      "step": 572500
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 8.630525252525253e-06,
+      "loss": 0.9874,
+      "step": 573000
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 8.620424242424243e-06,
+      "loss": 0.9886,
+      "step": 573500
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 8.610323232323234e-06,
+      "loss": 0.9869,
+      "step": 574000
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 8.600222222222224e-06,
+      "loss": 0.986,
+      "step": 574500
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 8.590121212121213e-06,
+      "loss": 0.9869,
+      "step": 575000
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 8.580020202020203e-06,
+      "loss": 0.9876,
+      "step": 575500
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 8.569919191919193e-06,
+      "loss": 0.9867,
+      "step": 576000
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 8.559818181818182e-06,
+      "loss": 0.9868,
+      "step": 576500
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 8.549737373737374e-06,
+      "loss": 0.9862,
+      "step": 577000
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 8.539636363636364e-06,
+      "loss": 0.9845,
+      "step": 577500
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 8.529535353535355e-06,
+      "loss": 0.9847,
+      "step": 578000
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 8.519434343434345e-06,
+      "loss": 0.9858,
+      "step": 578500
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 8.509333333333333e-06,
+      "loss": 0.9844,
+      "step": 579000
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 8.499232323232324e-06,
+      "loss": 0.9853,
+      "step": 579500
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 8.489131313131314e-06,
+      "loss": 0.9857,
+      "step": 580000
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 8.479050505050506e-06,
+      "loss": 0.9869,
+      "step": 580500
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 8.468949494949495e-06,
+      "loss": 0.9853,
+      "step": 581000
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 8.458848484848485e-06,
+      "loss": 0.9831,
+      "step": 581500
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 8.448747474747475e-06,
+      "loss": 0.9856,
+      "step": 582000
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 8.438646464646466e-06,
+      "loss": 0.9883,
+      "step": 582500
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 8.428565656565656e-06,
+      "loss": 0.9851,
+      "step": 583000
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 8.418464646464647e-06,
+      "loss": 0.9851,
+      "step": 583500
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 8.408363636363637e-06,
+      "loss": 0.9866,
+      "step": 584000
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 8.398262626262627e-06,
+      "loss": 0.9861,
+      "step": 584500
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 8.38818181818182e-06,
+      "loss": 0.9845,
+      "step": 585000
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 8.378080808080808e-06,
+      "loss": 0.9882,
+      "step": 585500
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 8.367979797979798e-06,
+      "loss": 0.983,
+      "step": 586000
+    },
+    {
+      "epoch": 25.0,
+      "eval_accuracy": 0.8029419404119512,
+      "eval_loss": 0.8218053579330444,
+      "eval_runtime": 55.9508,
+      "eval_samples_per_second": 924.813,
+      "eval_steps_per_second": 14.459,
+      "step": 586475
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 8.357878787878789e-06,
+      "loss": 0.984,
+      "step": 586500
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 8.347777777777779e-06,
+      "loss": 0.9821,
+      "step": 587000
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 8.33769696969697e-06,
+      "loss": 0.9842,
+      "step": 587500
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 8.32759595959596e-06,
+      "loss": 0.9849,
+      "step": 588000
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 8.31749494949495e-06,
+      "loss": 0.9827,
+      "step": 588500
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 8.30739393939394e-06,
+      "loss": 0.9861,
+      "step": 589000
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 8.29729292929293e-06,
+      "loss": 0.9845,
+      "step": 589500
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 8.28719191919192e-06,
+      "loss": 0.9826,
+      "step": 590000
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 8.277111111111111e-06,
+      "loss": 0.984,
+      "step": 590500
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 8.267010101010102e-06,
+      "loss": 0.9831,
+      "step": 591000
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 8.256909090909092e-06,
+      "loss": 0.9861,
+      "step": 591500
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 8.246808080808082e-06,
+      "loss": 0.985,
+      "step": 592000
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 8.236707070707072e-06,
+      "loss": 0.9864,
+      "step": 592500
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 8.226626262626263e-06,
+      "loss": 0.9852,
+      "step": 593000
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 8.216525252525253e-06,
+      "loss": 0.9844,
+      "step": 593500
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 8.206424242424244e-06,
+      "loss": 0.983,
+      "step": 594000
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 8.196323232323234e-06,
+      "loss": 0.9844,
+      "step": 594500
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 8.186222222222222e-06,
+      "loss": 0.9827,
+      "step": 595000
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 8.176121212121213e-06,
+      "loss": 0.9842,
+      "step": 595500
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 8.166020202020203e-06,
+      "loss": 0.9849,
+      "step": 596000
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 8.155919191919193e-06,
+      "loss": 0.9849,
+      "step": 596500
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 8.145818181818182e-06,
+      "loss": 0.9841,
+      "step": 597000
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 8.135737373737374e-06,
+      "loss": 0.9832,
+      "step": 597500
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 8.125636363636364e-06,
+      "loss": 0.9856,
+      "step": 598000
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 8.115535353535355e-06,
+      "loss": 0.9845,
+      "step": 598500
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 8.105434343434343e-06,
+      "loss": 0.9841,
+      "step": 599000
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 8.095333333333334e-06,
+      "loss": 0.9833,
+      "step": 599500
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 8.085232323232324e-06,
+      "loss": 0.9831,
+      "step": 600000
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 8.075131313131314e-06,
+      "loss": 0.983,
+      "step": 600500
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 8.065050505050506e-06,
+      "loss": 0.9845,
+      "step": 601000
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 8.054949494949495e-06,
+      "loss": 0.9827,
+      "step": 601500
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 8.044848484848485e-06,
+      "loss": 0.9849,
+      "step": 602000
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 8.034747474747476e-06,
+      "loss": 0.9831,
+      "step": 602500
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 8.024646464646464e-06,
+      "loss": 0.9829,
+      "step": 603000
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 8.014545454545455e-06,
+      "loss": 0.9856,
+      "step": 603500
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 8.004444444444445e-06,
+      "loss": 0.984,
+      "step": 604000
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 7.994343434343435e-06,
+      "loss": 0.9859,
+      "step": 604500
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 7.984262626262627e-06,
+      "loss": 0.9851,
+      "step": 605000
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 7.974161616161616e-06,
+      "loss": 0.9845,
+      "step": 605500
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 7.964060606060606e-06,
+      "loss": 0.9822,
+      "step": 606000
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 7.953979797979798e-06,
+      "loss": 0.9822,
+      "step": 606500
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 7.94389898989899e-06,
+      "loss": 0.985,
+      "step": 607000
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 7.933797979797981e-06,
+      "loss": 0.9839,
+      "step": 607500
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 7.92369696969697e-06,
+      "loss": 0.981,
+      "step": 608000
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 7.91359595959596e-06,
+      "loss": 0.9809,
+      "step": 608500
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 7.90349494949495e-06,
+      "loss": 0.9822,
+      "step": 609000
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 7.89339393939394e-06,
+      "loss": 0.9831,
+      "step": 609500
+    },
+    {
+      "epoch": 26.0,
+      "eval_accuracy": 0.8031982303159179,
+      "eval_loss": 0.8216830492019653,
+      "eval_runtime": 55.7951,
+      "eval_samples_per_second": 927.393,
+      "eval_steps_per_second": 14.499,
+      "step": 609934
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 7.883313131313131e-06,
+      "loss": 0.9833,
+      "step": 610000
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 7.873212121212121e-06,
+      "loss": 0.9827,
+      "step": 610500
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 7.863111111111112e-06,
+      "loss": 0.9827,
+      "step": 611000
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 7.853010101010102e-06,
+      "loss": 0.9829,
+      "step": 611500
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 7.842929292929294e-06,
+      "loss": 0.9812,
+      "step": 612000
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 7.832828282828283e-06,
+      "loss": 0.9811,
+      "step": 612500
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 7.822727272727273e-06,
+      "loss": 0.9835,
+      "step": 613000
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 7.812626262626263e-06,
+      "loss": 0.9809,
+      "step": 613500
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 7.802525252525252e-06,
+      "loss": 0.9824,
+      "step": 614000
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 7.792424242424244e-06,
+      "loss": 0.9841,
+      "step": 614500
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 7.782343434343434e-06,
+      "loss": 0.9833,
+      "step": 615000
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 7.772242424242425e-06,
+      "loss": 0.9805,
+      "step": 615500
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 7.762141414141415e-06,
+      "loss": 0.982,
+      "step": 616000
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 7.752040404040405e-06,
+      "loss": 0.9818,
+      "step": 616500
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 7.741939393939396e-06,
+      "loss": 0.9821,
+      "step": 617000
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 7.731858585858586e-06,
+      "loss": 0.9832,
+      "step": 617500
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 7.721757575757576e-06,
+      "loss": 0.9847,
+      "step": 618000
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 7.711656565656567e-06,
+      "loss": 0.9816,
+      "step": 618500
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 7.701575757575757e-06,
+      "loss": 0.9789,
+      "step": 619000
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 7.69147474747475e-06,
+      "loss": 0.9838,
+      "step": 619500
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 7.681373737373738e-06,
+      "loss": 0.9823,
+      "step": 620000
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 7.671272727272728e-06,
+      "loss": 0.9813,
+      "step": 620500
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 7.661171717171718e-06,
+      "loss": 0.9808,
+      "step": 621000
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 7.651070707070709e-06,
+      "loss": 0.9813,
+      "step": 621500
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 7.640969696969697e-06,
+      "loss": 0.9806,
+      "step": 622000
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 7.630868686868688e-06,
+      "loss": 0.9822,
+      "step": 622500
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 7.620767676767678e-06,
+      "loss": 0.9818,
+      "step": 623000
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 7.610666666666667e-06,
+      "loss": 0.9791,
+      "step": 623500
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 7.6005656565656576e-06,
+      "loss": 0.9826,
+      "step": 624000
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 7.590464646464647e-06,
+      "loss": 0.9817,
+      "step": 624500
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 7.5803636363636364e-06,
+      "loss": 0.9803,
+      "step": 625000
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 7.570282828282829e-06,
+      "loss": 0.9799,
+      "step": 625500
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 7.560181818181819e-06,
+      "loss": 0.9822,
+      "step": 626000
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 7.550080808080808e-06,
+      "loss": 0.9802,
+      "step": 626500
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 0.981,
+      "step": 627000
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 7.52989898989899e-06,
+      "loss": 0.981,
+      "step": 627500
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 7.51979797979798e-06,
+      "loss": 0.9822,
+      "step": 628000
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 7.50969696969697e-06,
+      "loss": 0.9838,
+      "step": 628500
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.49959595959596e-06,
+      "loss": 0.9812,
+      "step": 629000
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.4894949494949495e-06,
+      "loss": 0.9828,
+      "step": 629500
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.47939393939394e-06,
+      "loss": 0.9796,
+      "step": 630000
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.469292929292929e-06,
+      "loss": 0.9802,
+      "step": 630500
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.4591919191919196e-06,
+      "loss": 0.9804,
+      "step": 631000
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.449090909090909e-06,
+      "loss": 0.9798,
+      "step": 631500
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.439010101010101e-06,
+      "loss": 0.981,
+      "step": 632000
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.428909090909091e-06,
+      "loss": 0.9816,
+      "step": 632500
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.418808080808081e-06,
+      "loss": 0.9797,
+      "step": 633000
+    },
+    {
+      "epoch": 27.0,
+      "eval_accuracy": 0.8033411252212916,
+      "eval_loss": 0.8195702433586121,
+      "eval_runtime": 55.9315,
+      "eval_samples_per_second": 925.131,
+      "eval_steps_per_second": 14.464,
+      "step": 633393
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.408707070707072e-06,
+      "loss": 0.9806,
+      "step": 633500
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.3986060606060615e-06,
+      "loss": 0.9801,
+      "step": 634000
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.388525252525253e-06,
+      "loss": 0.9804,
+      "step": 634500
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.378424242424242e-06,
+      "loss": 0.9774,
+      "step": 635000
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.3683232323232335e-06,
+      "loss": 0.9795,
+      "step": 635500
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.358222222222223e-06,
+      "loss": 0.9765,
+      "step": 636000
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.348121212121213e-06,
+      "loss": 0.9802,
+      "step": 636500
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.3380404040404055e-06,
+      "loss": 0.9792,
+      "step": 637000
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.327939393939395e-06,
+      "loss": 0.9817,
+      "step": 637500
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.317838383838385e-06,
+      "loss": 0.98,
+      "step": 638000
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 7.307737373737375e-06,
+      "loss": 0.9791,
+      "step": 638500
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 7.297636363636364e-06,
+      "loss": 0.9806,
+      "step": 639000
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 7.287535353535354e-06,
+      "loss": 0.9813,
+      "step": 639500
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 7.277434343434344e-06,
+      "loss": 0.9802,
+      "step": 640000
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 7.267333333333334e-06,
+      "loss": 0.9808,
+      "step": 640500
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 7.2572323232323236e-06,
+      "loss": 0.9788,
+      "step": 641000
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 7.247151515151516e-06,
+      "loss": 0.9796,
+      "step": 641500
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 7.237050505050506e-06,
+      "loss": 0.9803,
+      "step": 642000
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 7.2269696969696975e-06,
+      "loss": 0.9791,
+      "step": 642500
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 7.216868686868688e-06,
+      "loss": 0.979,
+      "step": 643000
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 7.206767676767677e-06,
+      "loss": 0.9787,
+      "step": 643500
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 7.1966666666666675e-06,
+      "loss": 0.9785,
+      "step": 644000
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 7.186565656565657e-06,
+      "loss": 0.98,
+      "step": 644500
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 7.176464646464647e-06,
+      "loss": 0.98,
+      "step": 645000
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 7.166363636363637e-06,
+      "loss": 0.9797,
+      "step": 645500
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 7.156282828282829e-06,
+      "loss": 0.9806,
+      "step": 646000
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 7.146181818181818e-06,
+      "loss": 0.9791,
+      "step": 646500
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 7.136080808080809e-06,
+      "loss": 0.9831,
+      "step": 647000
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 7.125979797979798e-06,
+      "loss": 0.9815,
+      "step": 647500
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 7.115878787878788e-06,
+      "loss": 0.9778,
+      "step": 648000
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 7.105777777777778e-06,
+      "loss": 0.9777,
+      "step": 648500
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 7.095676767676768e-06,
+      "loss": 0.9797,
+      "step": 649000
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 7.0855757575757575e-06,
+      "loss": 0.9789,
+      "step": 649500
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 7.075474747474748e-06,
+      "loss": 0.9795,
+      "step": 650000
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 7.06539393939394e-06,
+      "loss": 0.9779,
+      "step": 650500
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 7.0552929292929295e-06,
+      "loss": 0.9797,
+      "step": 651000
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 7.045191919191919e-06,
+      "loss": 0.9825,
+      "step": 651500
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 7.035111111111111e-06,
+      "loss": 0.9773,
+      "step": 652000
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 7.0250101010101015e-06,
+      "loss": 0.9764,
+      "step": 652500
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 7.014909090909091e-06,
+      "loss": 0.9791,
+      "step": 653000
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 7.004808080808081e-06,
+      "loss": 0.9801,
+      "step": 653500
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 6.994707070707071e-06,
+      "loss": 0.9787,
+      "step": 654000
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 6.984606060606062e-06,
+      "loss": 0.9813,
+      "step": 654500
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 6.974525252525252e-06,
+      "loss": 0.981,
+      "step": 655000
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 6.964424242424243e-06,
+      "loss": 0.9785,
+      "step": 655500
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 6.954343434343435e-06,
+      "loss": 0.9791,
+      "step": 656000
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 6.944242424242424e-06,
+      "loss": 0.9777,
+      "step": 656500
+    },
+    {
+      "epoch": 28.0,
+      "eval_accuracy": 0.8037420817792058,
+      "eval_loss": 0.8172222971916199,
+      "eval_runtime": 55.3873,
+      "eval_samples_per_second": 934.221,
+      "eval_steps_per_second": 14.606,
+      "step": 656852
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 6.934141414141415e-06,
+      "loss": 0.9793,
+      "step": 657000
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 6.924040404040404e-06,
+      "loss": 0.9763,
+      "step": 657500
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 6.913939393939394e-06,
+      "loss": 0.9771,
+      "step": 658000
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 6.903838383838385e-06,
+      "loss": 0.9784,
+      "step": 658500
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 6.893737373737375e-06,
+      "loss": 0.9759,
+      "step": 659000
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 6.883636363636364e-06,
+      "loss": 0.9808,
+      "step": 659500
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 6.873535353535355e-06,
+      "loss": 0.9803,
+      "step": 660000
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 6.863434343434344e-06,
+      "loss": 0.9778,
+      "step": 660500
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 6.853333333333334e-06,
+      "loss": 0.9778,
+      "step": 661000
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 6.843252525252526e-06,
+      "loss": 0.9777,
+      "step": 661500
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 6.833151515151516e-06,
+      "loss": 0.9778,
+      "step": 662000
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 6.8230505050505055e-06,
+      "loss": 0.9754,
+      "step": 662500
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 6.812949494949496e-06,
+      "loss": 0.9781,
+      "step": 663000
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 6.802868686868688e-06,
+      "loss": 0.9779,
+      "step": 663500
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 6.7927676767676775e-06,
+      "loss": 0.9783,
+      "step": 664000
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 6.782666666666668e-06,
+      "loss": 0.977,
+      "step": 664500
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 6.772565656565657e-06,
+      "loss": 0.9778,
+      "step": 665000
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 6.7624646464646475e-06,
+      "loss": 0.9758,
+      "step": 665500
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 6.752363636363637e-06,
+      "loss": 0.9779,
+      "step": 666000
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 6.742282828282829e-06,
+      "loss": 0.9749,
+      "step": 666500
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 6.732181818181819e-06,
+      "loss": 0.978,
+      "step": 667000
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 6.722080808080809e-06,
+      "loss": 0.9761,
+      "step": 667500
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 6.711979797979798e-06,
+      "loss": 0.9755,
+      "step": 668000
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 6.701878787878789e-06,
+      "loss": 0.9784,
+      "step": 668500
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 6.691777777777778e-06,
+      "loss": 0.9744,
+      "step": 669000
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 6.681676767676768e-06,
+      "loss": 0.9756,
+      "step": 669500
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 6.671575757575758e-06,
+      "loss": 0.9783,
+      "step": 670000
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 6.661474747474748e-06,
+      "loss": 0.9752,
+      "step": 670500
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 6.6513737373737375e-06,
+      "loss": 0.9771,
+      "step": 671000
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 6.64129292929293e-06,
+      "loss": 0.9774,
+      "step": 671500
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 6.631191919191919e-06,
+      "loss": 0.9793,
+      "step": 672000
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 6.6210909090909095e-06,
+      "loss": 0.9757,
+      "step": 672500
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 6.610989898989899e-06,
+      "loss": 0.9769,
+      "step": 673000
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 6.600888888888889e-06,
+      "loss": 0.9796,
+      "step": 673500
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 6.590808080808081e-06,
+      "loss": 0.9739,
+      "step": 674000
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 6.580707070707071e-06,
+      "loss": 0.9787,
+      "step": 674500
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 6.57060606060606e-06,
+      "loss": 0.975,
+      "step": 675000
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 6.560505050505051e-06,
+      "loss": 0.9772,
+      "step": 675500
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 6.550424242424243e-06,
+      "loss": 0.9804,
+      "step": 676000
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 6.540343434343434e-06,
+      "loss": 0.9777,
+      "step": 676500
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 6.5302424242424245e-06,
+      "loss": 0.9779,
+      "step": 677000
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 6.520141414141414e-06,
+      "loss": 0.9769,
+      "step": 677500
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 6.510040404040404e-06,
+      "loss": 0.9781,
+      "step": 678000
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 6.499939393939394e-06,
+      "loss": 0.9743,
+      "step": 678500
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 6.489838383838384e-06,
+      "loss": 0.9775,
+      "step": 679000
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 6.479737373737375e-06,
+      "loss": 0.9761,
+      "step": 679500
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 6.4696363636363646e-06,
+      "loss": 0.9763,
+      "step": 680000
+    },
+    {
+      "epoch": 29.0,
+      "eval_accuracy": 0.8039930156734253,
+      "eval_loss": 0.8149685263633728,
+      "eval_runtime": 55.6261,
+      "eval_samples_per_second": 930.211,
+      "eval_steps_per_second": 14.544,
+      "step": 680311
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 6.459555555555556e-06,
+      "loss": 0.9737,
+      "step": 680500
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 6.449454545454546e-06,
+      "loss": 0.9767,
+      "step": 681000
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 6.4393535353535365e-06,
+      "loss": 0.9742,
+      "step": 681500
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 6.429252525252526e-06,
+      "loss": 0.9747,
+      "step": 682000
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 6.419151515151516e-06,
+      "loss": 0.9778,
+      "step": 682500
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 6.4090707070707085e-06,
+      "loss": 0.9743,
+      "step": 683000
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 6.398969696969698e-06,
+      "loss": 0.977,
+      "step": 683500
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 6.388868686868687e-06,
+      "loss": 0.9756,
+      "step": 684000
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 6.378767676767678e-06,
+      "loss": 0.9747,
+      "step": 684500
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 6.368666666666667e-06,
+      "loss": 0.9787,
+      "step": 685000
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 6.358565656565657e-06,
+      "loss": 0.977,
+      "step": 685500
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 6.34848484848485e-06,
+      "loss": 0.9738,
+      "step": 686000
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 6.338404040404041e-06,
+      "loss": 0.9755,
+      "step": 686500
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 6.328303030303031e-06,
+      "loss": 0.9764,
+      "step": 687000
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 6.318202020202021e-06,
+      "loss": 0.9767,
+      "step": 687500
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 6.308101010101011e-06,
+      "loss": 0.9753,
+      "step": 688000
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 6.298020202020203e-06,
+      "loss": 0.9773,
+      "step": 688500
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 6.287919191919193e-06,
+      "loss": 0.9772,
+      "step": 689000
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 6.277818181818183e-06,
+      "loss": 0.9718,
+      "step": 689500
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 6.2677171717171725e-06,
+      "loss": 0.974,
+      "step": 690000
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 6.257616161616163e-06,
+      "loss": 0.9743,
+      "step": 690500
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 6.247515151515152e-06,
+      "loss": 0.9742,
+      "step": 691000
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 6.2374343434343445e-06,
+      "loss": 0.9758,
+      "step": 691500
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 6.227333333333334e-06,
+      "loss": 0.9751,
+      "step": 692000
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 6.217232323232324e-06,
+      "loss": 0.9762,
+      "step": 692500
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 6.207131313131314e-06,
+      "loss": 0.9748,
+      "step": 693000
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 6.197030303030304e-06,
+      "loss": 0.9761,
+      "step": 693500
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 6.186929292929293e-06,
+      "loss": 0.976,
+      "step": 694000
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 6.176848484848486e-06,
+      "loss": 0.9753,
+      "step": 694500
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 6.166747474747475e-06,
+      "loss": 0.9752,
+      "step": 695000
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 6.156646464646465e-06,
+      "loss": 0.9762,
+      "step": 695500
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 6.146545454545455e-06,
+      "loss": 0.9757,
+      "step": 696000
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 6.136444444444445e-06,
+      "loss": 0.9727,
+      "step": 696500
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 6.126363636363637e-06,
+      "loss": 0.9753,
+      "step": 697000
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.116262626262627e-06,
+      "loss": 0.9743,
+      "step": 697500
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.106161616161617e-06,
+      "loss": 0.9755,
+      "step": 698000
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 6.0960606060606065e-06,
+      "loss": 0.9745,
+      "step": 698500
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 6.085959595959596e-06,
+      "loss": 0.9731,
+      "step": 699000
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 6.075858585858586e-06,
+      "loss": 0.9775,
+      "step": 699500
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 6.065757575757576e-06,
+      "loss": 0.9751,
+      "step": 700000
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 6.055656565656566e-06,
+      "loss": 0.9762,
+      "step": 700500
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 6.045575757575758e-06,
+      "loss": 0.9728,
+      "step": 701000
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 6.035474747474748e-06,
+      "loss": 0.9732,
+      "step": 701500
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 6.025373737373738e-06,
+      "loss": 0.9768,
+      "step": 702000
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 6.015272727272727e-06,
+      "loss": 0.973,
+      "step": 702500
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 6.005171717171718e-06,
+      "loss": 0.9757,
+      "step": 703000
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 5.995070707070707e-06,
+      "loss": 0.9756,
+      "step": 703500
+    },
+    {
+      "epoch": 30.0,
+      "eval_accuracy": 0.8041670667258376,
+      "eval_loss": 0.8154268264770508,
+      "eval_runtime": 55.767,
+      "eval_samples_per_second": 927.86,
+      "eval_steps_per_second": 14.507,
+      "step": 703770
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 5.984969696969698e-06,
+      "loss": 0.9718,
+      "step": 704000
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 5.974868686868688e-06,
+      "loss": 0.9756,
+      "step": 704500
+    },
+    {
+      "epoch": 30.05,
+      "learning_rate": 5.964787878787879e-06,
+      "loss": 0.9744,
+      "step": 705000
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 5.9546868686868685e-06,
+      "loss": 0.9746,
+      "step": 705500
+    },
+    {
+      "epoch": 30.1,
+      "learning_rate": 5.94458585858586e-06,
+      "loss": 0.9752,
+      "step": 706000
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 5.934484848484849e-06,
+      "loss": 0.9755,
+      "step": 706500
+    },
+    {
+      "epoch": 30.14,
+      "learning_rate": 5.9244040404040405e-06,
+      "loss": 0.973,
+      "step": 707000
+    },
+    {
+      "epoch": 30.16,
+      "learning_rate": 5.914303030303032e-06,
+      "loss": 0.9724,
+      "step": 707500
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 5.904202020202021e-06,
+      "loss": 0.9735,
+      "step": 708000
+    },
+    {
+      "epoch": 30.2,
+      "learning_rate": 5.894101010101011e-06,
+      "loss": 0.9724,
+      "step": 708500
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 5.884020202020202e-06,
+      "loss": 0.9747,
+      "step": 709000
+    },
+    {
+      "epoch": 30.24,
+      "learning_rate": 5.873919191919193e-06,
+      "loss": 0.9753,
+      "step": 709500
+    },
+    {
+      "epoch": 30.27,
+      "learning_rate": 5.8638181818181824e-06,
+      "loss": 0.9732,
+      "step": 710000
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 5.853717171717173e-06,
+      "loss": 0.9729,
+      "step": 710500
+    },
+    {
+      "epoch": 30.31,
+      "learning_rate": 5.843616161616162e-06,
+      "loss": 0.9757,
+      "step": 711000
+    },
+    {
+      "epoch": 30.33,
+      "learning_rate": 5.833535353535354e-06,
+      "loss": 0.9715,
+      "step": 711500
+    },
+    {
+      "epoch": 30.35,
+      "learning_rate": 5.823434343434345e-06,
+      "loss": 0.974,
+      "step": 712000
+    },
+    {
+      "epoch": 30.37,
+      "learning_rate": 5.813333333333334e-06,
+      "loss": 0.9743,
+      "step": 712500
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 5.803232323232324e-06,
+      "loss": 0.974,
+      "step": 713000
+    },
+    {
+      "epoch": 30.41,
+      "learning_rate": 5.793131313131314e-06,
+      "loss": 0.9752,
+      "step": 713500
+    },
+    {
+      "epoch": 30.44,
+      "learning_rate": 5.783030303030303e-06,
+      "loss": 0.9716,
+      "step": 714000
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 5.772929292929294e-06,
+      "loss": 0.9757,
+      "step": 714500
+    },
+    {
+      "epoch": 30.48,
+      "learning_rate": 5.762828282828283e-06,
+      "loss": 0.9725,
+      "step": 715000
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 5.752747474747475e-06,
+      "loss": 0.974,
+      "step": 715500
+    },
+    {
+      "epoch": 30.52,
+      "learning_rate": 5.7426666666666675e-06,
+      "loss": 0.9718,
+      "step": 716000
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 5.732565656565657e-06,
+      "loss": 0.9735,
+      "step": 716500
+    },
+    {
+      "epoch": 30.56,
+      "learning_rate": 5.722464646464647e-06,
+      "loss": 0.9729,
+      "step": 717000
+    },
+    {
+      "epoch": 30.59,
+      "learning_rate": 5.712363636363637e-06,
+      "loss": 0.9733,
+      "step": 717500
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 5.702262626262627e-06,
+      "loss": 0.9741,
+      "step": 718000
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 5.6921616161616164e-06,
+      "loss": 0.9769,
+      "step": 718500
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 5.682080808080809e-06,
+      "loss": 0.9735,
+      "step": 719000
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 5.671979797979799e-06,
+      "loss": 0.9722,
+      "step": 719500
+    },
+    {
+      "epoch": 30.69,
+      "learning_rate": 5.661878787878788e-06,
+      "loss": 0.9747,
+      "step": 720000
+    },
+    {
+      "epoch": 30.71,
+      "learning_rate": 5.651777777777778e-06,
+      "loss": 0.9723,
+      "step": 720500
+    },
+    {
+      "epoch": 30.73,
+      "learning_rate": 5.641676767676768e-06,
+      "loss": 0.9732,
+      "step": 721000
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 5.63159595959596e-06,
+      "loss": 0.9736,
+      "step": 721500
+    },
+    {
+      "epoch": 30.78,
+      "learning_rate": 5.62149494949495e-06,
+      "loss": 0.9731,
+      "step": 722000
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 5.61139393939394e-06,
+      "loss": 0.9736,
+      "step": 722500
+    },
+    {
+      "epoch": 30.82,
+      "learning_rate": 5.6012929292929295e-06,
+      "loss": 0.973,
+      "step": 723000
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 5.591212121212122e-06,
+      "loss": 0.9744,
+      "step": 723500
+    },
+    {
+      "epoch": 30.86,
+      "learning_rate": 5.581111111111111e-06,
+      "loss": 0.9747,
+      "step": 724000
+    },
+    {
+      "epoch": 30.88,
+      "learning_rate": 5.5710101010101015e-06,
+      "loss": 0.9714,
+      "step": 724500
+    },
+    {
+      "epoch": 30.9,
+      "learning_rate": 5.560909090909091e-06,
+      "loss": 0.974,
+      "step": 725000
+    },
+    {
+      "epoch": 30.93,
+      "learning_rate": 5.550808080808081e-06,
+      "loss": 0.9736,
+      "step": 725500
+    },
+    {
+      "epoch": 30.95,
+      "learning_rate": 5.5407272727272735e-06,
+      "loss": 0.9707,
+      "step": 726000
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 5.530626262626263e-06,
+      "loss": 0.9721,
+      "step": 726500
+    },
+    {
+      "epoch": 30.99,
+      "learning_rate": 5.520525252525253e-06,
+      "loss": 0.9759,
+      "step": 727000
+    },
+    {
+      "epoch": 31.0,
+      "eval_accuracy": 0.8045686859011433,
+      "eval_loss": 0.812103807926178,
+      "eval_runtime": 55.9635,
+      "eval_samples_per_second": 924.603,
+      "eval_steps_per_second": 14.456,
+      "step": 727229
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 5.510424242424243e-06,
+      "loss": 0.9718,
+      "step": 727500
+    },
+    {
+      "epoch": 31.03,
+      "learning_rate": 5.500343434343435e-06,
+      "loss": 0.9702,
+      "step": 728000
+    },
+    {
+      "epoch": 31.05,
+      "learning_rate": 5.490242424242424e-06,
+      "loss": 0.9735,
+      "step": 728500
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 5.480141414141415e-06,
+      "loss": 0.972,
+      "step": 729000
+    },
+    {
+      "epoch": 31.1,
+      "learning_rate": 5.470040404040404e-06,
+      "loss": 0.9728,
+      "step": 729500
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 5.459939393939394e-06,
+      "loss": 0.9702,
+      "step": 730000
+    },
+    {
+      "epoch": 31.14,
+      "learning_rate": 5.449838383838384e-06,
+      "loss": 0.9727,
+      "step": 730500
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 5.439737373737374e-06,
+      "loss": 0.9721,
+      "step": 731000
+    },
+    {
+      "epoch": 31.18,
+      "learning_rate": 5.4296363636363635e-06,
+      "loss": 0.972,
+      "step": 731500
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 5.419535353535354e-06,
+      "loss": 0.9683,
+      "step": 732000
+    },
+    {
+      "epoch": 31.22,
+      "learning_rate": 5.409434343434344e-06,
+      "loss": 0.9701,
+      "step": 732500
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 5.3993535353535355e-06,
+      "loss": 0.9746,
+      "step": 733000
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 5.389252525252525e-06,
+      "loss": 0.9706,
+      "step": 733500
+    },
+    {
+      "epoch": 31.29,
+      "learning_rate": 5.379171717171717e-06,
+      "loss": 0.9694,
+      "step": 734000
+    },
+    {
+      "epoch": 31.31,
+      "learning_rate": 5.3690909090909094e-06,
+      "loss": 0.9704,
+      "step": 734500
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 5.358989898989899e-06,
+      "loss": 0.972,
+      "step": 735000
+    },
+    {
+      "epoch": 31.35,
+      "learning_rate": 5.348888888888889e-06,
+      "loss": 0.971,
+      "step": 735500
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 5.338787878787879e-06,
+      "loss": 0.973,
+      "step": 736000
+    },
+    {
+      "epoch": 31.4,
+      "learning_rate": 5.328686868686869e-06,
+      "loss": 0.9741,
+      "step": 736500
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 5.318606060606061e-06,
+      "loss": 0.9714,
+      "step": 737000
+    },
+    {
+      "epoch": 31.44,
+      "learning_rate": 5.3085050505050506e-06,
+      "loss": 0.973,
+      "step": 737500
+    },
+    {
+      "epoch": 31.46,
+      "learning_rate": 5.298404040404041e-06,
+      "loss": 0.9724,
+      "step": 738000
+    },
+    {
+      "epoch": 31.48,
+      "learning_rate": 5.28830303030303e-06,
+      "loss": 0.9726,
+      "step": 738500
+    },
+    {
+      "epoch": 31.5,
+      "learning_rate": 5.27820202020202e-06,
+      "loss": 0.9729,
+      "step": 739000
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 5.268101010101011e-06,
+      "loss": 0.9727,
+      "step": 739500
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 5.258000000000001e-06,
+      "loss": 0.9759,
+      "step": 740000
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 5.247898989898991e-06,
+      "loss": 0.9723,
+      "step": 740500
+    },
+    {
+      "epoch": 31.59,
+      "learning_rate": 5.237797979797981e-06,
+      "loss": 0.9699,
+      "step": 741000
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 5.227717171717172e-06,
+      "loss": 0.9718,
+      "step": 741500
+    },
+    {
+      "epoch": 31.63,
+      "learning_rate": 5.2176161616161626e-06,
+      "loss": 0.9727,
+      "step": 742000
+    },
+    {
+      "epoch": 31.65,
+      "learning_rate": 5.207515151515152e-06,
+      "loss": 0.9718,
+      "step": 742500
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 5.197414141414142e-06,
+      "loss": 0.9733,
+      "step": 743000
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 5.187313131313132e-06,
+      "loss": 0.9739,
+      "step": 743500
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 5.177232323232324e-06,
+      "loss": 0.973,
+      "step": 744000
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 5.167151515151515e-06,
+      "loss": 0.9697,
+      "step": 744500
+    },
+    {
+      "epoch": 31.76,
+      "learning_rate": 5.157050505050506e-06,
+      "loss": 0.9714,
+      "step": 745000
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 5.146949494949496e-06,
+      "loss": 0.9702,
+      "step": 745500
+    },
+    {
+      "epoch": 31.8,
+      "learning_rate": 5.136848484848485e-06,
+      "loss": 0.9707,
+      "step": 746000
+    },
+    {
+      "epoch": 31.82,
+      "learning_rate": 5.126747474747476e-06,
+      "loss": 0.9715,
+      "step": 746500
+    },
+    {
+      "epoch": 31.84,
+      "learning_rate": 5.116646464646465e-06,
+      "loss": 0.9709,
+      "step": 747000
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 5.106545454545455e-06,
+      "loss": 0.9694,
+      "step": 747500
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 5.096444444444445e-06,
+      "loss": 0.9703,
+      "step": 748000
+    },
+    {
+      "epoch": 31.91,
+      "learning_rate": 5.086383838383839e-06,
+      "loss": 0.9687,
+      "step": 748500
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 5.076282828282829e-06,
+      "loss": 0.9721,
+      "step": 749000
+    },
+    {
+      "epoch": 31.95,
+      "learning_rate": 5.066181818181819e-06,
+      "loss": 0.972,
+      "step": 749500
+    },
+    {
+      "epoch": 31.97,
+      "learning_rate": 5.056080808080809e-06,
+      "loss": 0.9711,
+      "step": 750000
+    },
+    {
+      "epoch": 31.99,
+      "learning_rate": 5.0459797979797985e-06,
+      "loss": 0.9711,
+      "step": 750500
+    },
+    {
+      "epoch": 32.0,
+      "eval_accuracy": 0.804346371104709,
+      "eval_loss": 0.8124815225601196,
+      "eval_runtime": 54.8342,
+      "eval_samples_per_second": 943.645,
+      "eval_steps_per_second": 14.754,
+      "step": 750688
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 5.035878787878789e-06,
+      "loss": 0.9714,
+      "step": 751000
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 5.025777777777778e-06,
+      "loss": 0.9697,
+      "step": 751500
+    },
+    {
+      "epoch": 32.06,
+      "learning_rate": 5.0156767676767685e-06,
+      "loss": 0.9721,
+      "step": 752000
+    },
+    {
+      "epoch": 32.08,
+      "learning_rate": 5.00559595959596e-06,
+      "loss": 0.9705,
+      "step": 752500
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 4.99549494949495e-06,
+      "loss": 0.97,
+      "step": 753000
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 4.98539393939394e-06,
+      "loss": 0.9712,
+      "step": 753500
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 4.97529292929293e-06,
+      "loss": 0.9703,
+      "step": 754000
+    },
+    {
+      "epoch": 32.16,
+      "learning_rate": 4.965191919191919e-06,
+      "loss": 0.9689,
+      "step": 754500
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 4.95509090909091e-06,
+      "loss": 0.9694,
+      "step": 755000
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 4.944989898989899e-06,
+      "loss": 0.9724,
+      "step": 755500
+    },
+    {
+      "epoch": 32.23,
+      "learning_rate": 4.934909090909091e-06,
+      "loss": 0.9717,
+      "step": 756000
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 4.924808080808081e-06,
+      "loss": 0.9705,
+      "step": 756500
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 4.914707070707071e-06,
+      "loss": 0.9714,
+      "step": 757000
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 4.9046060606060605e-06,
+      "loss": 0.9691,
+      "step": 757500
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 4.894505050505051e-06,
+      "loss": 0.9704,
+      "step": 758000
+    },
+    {
+      "epoch": 32.33,
+      "learning_rate": 4.884404040404041e-06,
+      "loss": 0.9712,
+      "step": 758500
+    },
+    {
+      "epoch": 32.35,
+      "learning_rate": 4.8743030303030305e-06,
+      "loss": 0.968,
+      "step": 759000
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 4.864202020202021e-06,
+      "loss": 0.9723,
+      "step": 759500
+    },
+    {
+      "epoch": 32.4,
+      "learning_rate": 4.85410101010101e-06,
+      "loss": 0.9719,
+      "step": 760000
+    },
+    {
+      "epoch": 32.42,
+      "learning_rate": 4.8440000000000005e-06,
+      "loss": 0.9696,
+      "step": 760500
+    },
+    {
+      "epoch": 32.44,
+      "learning_rate": 4.83389898989899e-06,
+      "loss": 0.9689,
+      "step": 761000
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 4.82379797979798e-06,
+      "loss": 0.9727,
+      "step": 761500
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 4.8137171717171725e-06,
+      "loss": 0.9704,
+      "step": 762000
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 4.803616161616162e-06,
+      "loss": 0.9701,
+      "step": 762500
+    },
+    {
+      "epoch": 32.52,
+      "learning_rate": 4.793515151515151e-06,
+      "loss": 0.9676,
+      "step": 763000
+    },
+    {
+      "epoch": 32.55,
+      "learning_rate": 4.783414141414142e-06,
+      "loss": 0.9701,
+      "step": 763500
+    },
+    {
+      "epoch": 32.57,
+      "learning_rate": 4.773333333333334e-06,
+      "loss": 0.9707,
+      "step": 764000
+    },
+    {
+      "epoch": 32.59,
+      "learning_rate": 4.763232323232323e-06,
+      "loss": 0.9709,
+      "step": 764500
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 4.753151515151516e-06,
+      "loss": 0.9701,
+      "step": 765000
+    },
+    {
+      "epoch": 32.63,
+      "learning_rate": 4.743050505050505e-06,
+      "loss": 0.9684,
+      "step": 765500
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 4.732949494949495e-06,
+      "loss": 0.9697,
+      "step": 766000
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 4.722848484848485e-06,
+      "loss": 0.9682,
+      "step": 766500
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 4.712747474747475e-06,
+      "loss": 0.9692,
+      "step": 767000
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 4.7026464646464645e-06,
+      "loss": 0.9678,
+      "step": 767500
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 4.692545454545455e-06,
+      "loss": 0.9693,
+      "step": 768000
+    },
+    {
+      "epoch": 32.76,
+      "learning_rate": 4.682444444444445e-06,
+      "loss": 0.9706,
+      "step": 768500
+    },
+    {
+      "epoch": 32.78,
+      "learning_rate": 4.6723434343434345e-06,
+      "loss": 0.9674,
+      "step": 769000
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 4.662262626262627e-06,
+      "loss": 0.9688,
+      "step": 769500
+    },
+    {
+      "epoch": 32.82,
+      "learning_rate": 4.652161616161617e-06,
+      "loss": 0.9683,
+      "step": 770000
+    },
+    {
+      "epoch": 32.84,
+      "learning_rate": 4.6420808080808085e-06,
+      "loss": 0.969,
+      "step": 770500
+    },
+    {
+      "epoch": 32.87,
+      "learning_rate": 4.631979797979798e-06,
+      "loss": 0.9714,
+      "step": 771000
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 4.621878787878788e-06,
+      "loss": 0.9714,
+      "step": 771500
+    },
+    {
+      "epoch": 32.91,
+      "learning_rate": 4.6117777777777785e-06,
+      "loss": 0.9689,
+      "step": 772000
+    },
+    {
+      "epoch": 32.93,
+      "learning_rate": 4.60169696969697e-06,
+      "loss": 0.9678,
+      "step": 772500
+    },
+    {
+      "epoch": 32.95,
+      "learning_rate": 4.591595959595959e-06,
+      "loss": 0.9704,
+      "step": 773000
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 4.5814949494949505e-06,
+      "loss": 0.9695,
+      "step": 773500
+    },
+    {
+      "epoch": 32.99,
+      "learning_rate": 4.57139393939394e-06,
+      "loss": 0.969,
+      "step": 774000
+    },
+    {
+      "epoch": 33.0,
+      "eval_accuracy": 0.8049899130630866,
+      "eval_loss": 0.8093631863594055,
+      "eval_runtime": 56.1917,
+      "eval_samples_per_second": 920.847,
+      "eval_steps_per_second": 14.397,
+      "step": 774147
+    },
+    {
+      "epoch": 33.02,
+      "learning_rate": 4.56129292929293e-06,
+      "loss": 0.9706,
+      "step": 774500
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 4.551212121212122e-06,
+      "loss": 0.9687,
+      "step": 775000
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 4.541111111111112e-06,
+      "loss": 0.9685,
+      "step": 775500
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 4.531010101010101e-06,
+      "loss": 0.9694,
+      "step": 776000
+    },
+    {
+      "epoch": 33.1,
+      "learning_rate": 4.520929292929293e-06,
+      "loss": 0.9696,
+      "step": 776500
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 4.510828282828284e-06,
+      "loss": 0.9706,
+      "step": 777000
+    },
+    {
+      "epoch": 33.14,
+      "learning_rate": 4.500727272727273e-06,
+      "loss": 0.9665,
+      "step": 777500
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 4.490626262626263e-06,
+      "loss": 0.9688,
+      "step": 778000
+    },
+    {
+      "epoch": 33.19,
+      "learning_rate": 4.480525252525253e-06,
+      "loss": 0.9699,
+      "step": 778500
+    },
+    {
+      "epoch": 33.21,
+      "learning_rate": 4.4704242424242424e-06,
+      "loss": 0.9704,
+      "step": 779000
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 4.460323232323233e-06,
+      "loss": 0.9693,
+      "step": 779500
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 4.450222222222222e-06,
+      "loss": 0.9703,
+      "step": 780000
+    },
+    {
+      "epoch": 33.27,
+      "learning_rate": 4.440141414141414e-06,
+      "loss": 0.9687,
+      "step": 780500
+    },
+    {
+      "epoch": 33.29,
+      "learning_rate": 4.430040404040405e-06,
+      "loss": 0.9716,
+      "step": 781000
+    },
+    {
+      "epoch": 33.31,
+      "learning_rate": 4.419939393939394e-06,
+      "loss": 0.9701,
+      "step": 781500
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 4.4098383838383844e-06,
+      "loss": 0.9694,
+      "step": 782000
+    },
+    {
+      "epoch": 33.36,
+      "learning_rate": 4.399737373737374e-06,
+      "loss": 0.9679,
+      "step": 782500
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 4.389656565656566e-06,
+      "loss": 0.9695,
+      "step": 783000
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 4.3795555555555556e-06,
+      "loss": 0.9684,
+      "step": 783500
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 4.369454545454546e-06,
+      "loss": 0.969,
+      "step": 784000
+    },
+    {
+      "epoch": 33.44,
+      "learning_rate": 4.359353535353535e-06,
+      "loss": 0.9682,
+      "step": 784500
+    },
+    {
+      "epoch": 33.46,
+      "learning_rate": 4.3492727272727275e-06,
+      "loss": 0.9694,
+      "step": 785000
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 4.339171717171717e-06,
+      "loss": 0.9704,
+      "step": 785500
+    },
+    {
+      "epoch": 33.51,
+      "learning_rate": 4.329090909090909e-06,
+      "loss": 0.9694,
+      "step": 786000
+    },
+    {
+      "epoch": 33.53,
+      "learning_rate": 4.3189898989898995e-06,
+      "loss": 0.9697,
+      "step": 786500
+    },
+    {
+      "epoch": 33.55,
+      "learning_rate": 4.308888888888889e-06,
+      "loss": 0.9687,
+      "step": 787000
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 4.298787878787879e-06,
+      "loss": 0.9723,
+      "step": 787500
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 4.288686868686869e-06,
+      "loss": 0.9696,
+      "step": 788000
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 4.278585858585859e-06,
+      "loss": 0.9679,
+      "step": 788500
+    },
+    {
+      "epoch": 33.63,
+      "learning_rate": 4.268484848484849e-06,
+      "loss": 0.9691,
+      "step": 789000
+    },
+    {
+      "epoch": 33.65,
+      "learning_rate": 4.258383838383839e-06,
+      "loss": 0.9677,
+      "step": 789500
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 4.248282828282829e-06,
+      "loss": 0.9678,
+      "step": 790000
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 4.238181818181818e-06,
+      "loss": 0.9708,
+      "step": 790500
+    },
+    {
+      "epoch": 33.72,
+      "learning_rate": 4.228080808080809e-06,
+      "loss": 0.9681,
+      "step": 791000
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 4.218e-06,
+      "loss": 0.9704,
+      "step": 791500
+    },
+    {
+      "epoch": 33.76,
+      "learning_rate": 4.20789898989899e-06,
+      "loss": 0.9667,
+      "step": 792000
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 4.19779797979798e-06,
+      "loss": 0.9675,
+      "step": 792500
+    },
+    {
+      "epoch": 33.8,
+      "learning_rate": 4.18769696969697e-06,
+      "loss": 0.9664,
+      "step": 793000
+    },
+    {
+      "epoch": 33.82,
+      "learning_rate": 4.1775959595959596e-06,
+      "loss": 0.9664,
+      "step": 793500
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 4.167515151515152e-06,
+      "loss": 0.9686,
+      "step": 794000
+    },
+    {
+      "epoch": 33.87,
+      "learning_rate": 4.157434343434344e-06,
+      "loss": 0.9664,
+      "step": 794500
+    },
+    {
+      "epoch": 33.89,
+      "learning_rate": 4.1473333333333335e-06,
+      "loss": 0.97,
+      "step": 795000
+    },
+    {
+      "epoch": 33.91,
+      "learning_rate": 4.137232323232324e-06,
+      "loss": 0.9694,
+      "step": 795500
+    },
+    {
+      "epoch": 33.93,
+      "learning_rate": 4.127131313131313e-06,
+      "loss": 0.9688,
+      "step": 796000
+    },
+    {
+      "epoch": 33.95,
+      "learning_rate": 4.1170303030303035e-06,
+      "loss": 0.9699,
+      "step": 796500
+    },
+    {
+      "epoch": 33.97,
+      "learning_rate": 4.106929292929293e-06,
+      "loss": 0.9655,
+      "step": 797000
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 4.096848484848485e-06,
+      "loss": 0.9671,
+      "step": 797500
+    },
+    {
+      "epoch": 34.0,
+      "eval_accuracy": 0.8052881810426409,
+      "eval_loss": 0.8087067008018494,
+      "eval_runtime": 56.5603,
+      "eval_samples_per_second": 914.847,
+      "eval_steps_per_second": 14.303,
+      "step": 797606
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 4.086747474747475e-06,
+      "loss": 0.9659,
+      "step": 798000
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 4.076646464646465e-06,
+      "loss": 0.9693,
+      "step": 798500
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 4.066545454545454e-06,
+      "loss": 0.9673,
+      "step": 799000
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 4.056444444444445e-06,
+      "loss": 0.9667,
+      "step": 799500
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 4.046343434343435e-06,
+      "loss": 0.9664,
+      "step": 800000
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 4.036242424242424e-06,
+      "loss": 0.9691,
+      "step": 800500
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 4.026161616161617e-06,
+      "loss": 0.9671,
+      "step": 801000
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 4.016060606060607e-06,
+      "loss": 0.9675,
+      "step": 801500
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 4.005959595959596e-06,
+      "loss": 0.968,
+      "step": 802000
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 3.995858585858587e-06,
+      "loss": 0.9684,
+      "step": 802500
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 3.985757575757576e-06,
+      "loss": 0.9677,
+      "step": 803000
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 3.975656565656566e-06,
+      "loss": 0.9674,
+      "step": 803500
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 3.965555555555556e-06,
+      "loss": 0.967,
+      "step": 804000
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 3.955454545454545e-06,
+      "loss": 0.9658,
+      "step": 804500
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 3.9453535353535355e-06,
+      "loss": 0.9705,
+      "step": 805000
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 3.935252525252525e-06,
+      "loss": 0.9689,
+      "step": 805500
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 3.925171717171717e-06,
+      "loss": 0.9659,
+      "step": 806000
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 3.9150707070707075e-06,
+      "loss": 0.9662,
+      "step": 806500
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 3.904969696969697e-06,
+      "loss": 0.9714,
+      "step": 807000
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 3.894868686868687e-06,
+      "loss": 0.97,
+      "step": 807500
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 3.884787878787879e-06,
+      "loss": 0.9674,
+      "step": 808000
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 3.874686868686869e-06,
+      "loss": 0.9683,
+      "step": 808500
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 3.864606060606061e-06,
+      "loss": 0.9687,
+      "step": 809000
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 3.854505050505051e-06,
+      "loss": 0.9699,
+      "step": 809500
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 3.844404040404041e-06,
+      "loss": 0.968,
+      "step": 810000
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 3.83430303030303e-06,
+      "loss": 0.9661,
+      "step": 810500
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 3.8242222222222226e-06,
+      "loss": 0.9671,
+      "step": 811000
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 3.8141212121212124e-06,
+      "loss": 0.9687,
+      "step": 811500
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 3.8040202020202023e-06,
+      "loss": 0.9673,
+      "step": 812000
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 3.793919191919192e-06,
+      "loss": 0.9657,
+      "step": 812500
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 3.7838181818181825e-06,
+      "loss": 0.9691,
+      "step": 813000
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 3.7737171717171723e-06,
+      "loss": 0.9688,
+      "step": 813500
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 3.763616161616162e-06,
+      "loss": 0.9656,
+      "step": 814000
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 3.753515151515152e-06,
+      "loss": 0.9696,
+      "step": 814500
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 3.743414141414142e-06,
+      "loss": 0.9664,
+      "step": 815000
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 3.7333131313131313e-06,
+      "loss": 0.9682,
+      "step": 815500
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 3.723212121212121e-06,
+      "loss": 0.9653,
+      "step": 816000
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 3.713111111111111e-06,
+      "loss": 0.9661,
+      "step": 816500
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 3.703010101010101e-06,
+      "loss": 0.9669,
+      "step": 817000
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 3.692929292929293e-06,
+      "loss": 0.9652,
+      "step": 817500
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 3.682828282828283e-06,
+      "loss": 0.9633,
+      "step": 818000
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 3.672727272727273e-06,
+      "loss": 0.9661,
+      "step": 818500
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 3.662626262626263e-06,
+      "loss": 0.9641,
+      "step": 819000
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 3.6525454545454546e-06,
+      "loss": 0.9685,
+      "step": 819500
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 3.6424444444444445e-06,
+      "loss": 0.9684,
+      "step": 820000
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 3.6323636363636367e-06,
+      "loss": 0.9652,
+      "step": 820500
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3.6222626262626266e-06,
+      "loss": 0.9674,
+      "step": 821000
+    },
+    {
+      "epoch": 35.0,
+      "eval_accuracy": 0.8053584306118093,
+      "eval_loss": 0.8091870546340942,
+      "eval_runtime": 56.1385,
+      "eval_samples_per_second": 921.72,
+      "eval_steps_per_second": 14.411,
+      "step": 821065
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 3.6121616161616164e-06,
+      "loss": 0.9666,
+      "step": 821500
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 3.6020808080808083e-06,
+      "loss": 0.9643,
+      "step": 822000
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 3.591979797979798e-06,
+      "loss": 0.9665,
+      "step": 822500
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 3.581878787878788e-06,
+      "loss": 0.9664,
+      "step": 823000
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 3.571777777777778e-06,
+      "loss": 0.9674,
+      "step": 823500
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 3.5616767676767677e-06,
+      "loss": 0.9642,
+      "step": 824000
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 3.551575757575758e-06,
+      "loss": 0.9686,
+      "step": 824500
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 3.541474747474748e-06,
+      "loss": 0.9653,
+      "step": 825000
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 3.5313737373737377e-06,
+      "loss": 0.9646,
+      "step": 825500
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 3.5212727272727276e-06,
+      "loss": 0.9679,
+      "step": 826000
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 3.5112121212121214e-06,
+      "loss": 0.9656,
+      "step": 826500
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 3.5011111111111112e-06,
+      "loss": 0.9641,
+      "step": 827000
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 3.491010101010101e-06,
+      "loss": 0.963,
+      "step": 827500
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 3.4809090909090914e-06,
+      "loss": 0.9683,
+      "step": 828000
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 3.4708080808080812e-06,
+      "loss": 0.9676,
+      "step": 828500
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 3.460707070707071e-06,
+      "loss": 0.9685,
+      "step": 829000
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 3.450606060606061e-06,
+      "loss": 0.9672,
+      "step": 829500
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 3.440505050505051e-06,
+      "loss": 0.9654,
+      "step": 830000
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 3.4304040404040407e-06,
+      "loss": 0.9686,
+      "step": 830500
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 3.4203030303030306e-06,
+      "loss": 0.9672,
+      "step": 831000
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 3.4102020202020204e-06,
+      "loss": 0.9658,
+      "step": 831500
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 3.4001010101010103e-06,
+      "loss": 0.9686,
+      "step": 832000
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 3.390020202020202e-06,
+      "loss": 0.9648,
+      "step": 832500
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 3.3799393939393944e-06,
+      "loss": 0.9673,
+      "step": 833000
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 3.3698383838383842e-06,
+      "loss": 0.9678,
+      "step": 833500
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 3.359737373737374e-06,
+      "loss": 0.9652,
+      "step": 834000
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 3.349636363636364e-06,
+      "loss": 0.9655,
+      "step": 834500
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 3.339535353535354e-06,
+      "loss": 0.9665,
+      "step": 835000
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 3.3294343434343433e-06,
+      "loss": 0.9645,
+      "step": 835500
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 3.319333333333334e-06,
+      "loss": 0.964,
+      "step": 836000
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 3.3092525252525254e-06,
+      "loss": 0.9636,
+      "step": 836500
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 3.2991515151515152e-06,
+      "loss": 0.9648,
+      "step": 837000
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 3.289050505050505e-06,
+      "loss": 0.9659,
+      "step": 837500
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 3.2789494949494954e-06,
+      "loss": 0.9657,
+      "step": 838000
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 3.2688686868686868e-06,
+      "loss": 0.9636,
+      "step": 838500
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 3.2587676767676766e-06,
+      "loss": 0.9675,
+      "step": 839000
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 3.2486666666666674e-06,
+      "loss": 0.9656,
+      "step": 839500
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 3.2385656565656572e-06,
+      "loss": 0.9652,
+      "step": 840000
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 3.2284646464646467e-06,
+      "loss": 0.9662,
+      "step": 840500
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 3.2183838383838385e-06,
+      "loss": 0.967,
+      "step": 841000
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 3.2082828282828288e-06,
+      "loss": 0.9668,
+      "step": 841500
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 3.1981818181818186e-06,
+      "loss": 0.9663,
+      "step": 842000
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 3.1880808080808085e-06,
+      "loss": 0.9673,
+      "step": 842500
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 3.1780000000000003e-06,
+      "loss": 0.9673,
+      "step": 843000
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 3.16789898989899e-06,
+      "loss": 0.9653,
+      "step": 843500
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 3.15779797979798e-06,
+      "loss": 0.9694,
+      "step": 844000
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 3.14769696969697e-06,
+      "loss": 0.965,
+      "step": 844500
+    },
+    {
+      "epoch": 36.0,
+      "eval_accuracy": 0.8058653824826437,
+      "eval_loss": 0.8053589463233948,
+      "eval_runtime": 56.5157,
+      "eval_samples_per_second": 915.568,
+      "eval_steps_per_second": 14.315,
+      "step": 844524
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 3.1375959595959598e-06,
+      "loss": 0.9638,
+      "step": 845000
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 3.1274949494949496e-06,
+      "loss": 0.9643,
+      "step": 845500
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 3.1173939393939395e-06,
+      "loss": 0.9668,
+      "step": 846000
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 3.1072929292929294e-06,
+      "loss": 0.966,
+      "step": 846500
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 3.0971919191919192e-06,
+      "loss": 0.9654,
+      "step": 847000
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 3.0871111111111115e-06,
+      "loss": 0.9654,
+      "step": 847500
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 3.077010101010101e-06,
+      "loss": 0.9648,
+      "step": 848000
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 3.066929292929293e-06,
+      "loss": 0.9646,
+      "step": 848500
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 3.056828282828283e-06,
+      "loss": 0.9629,
+      "step": 849000
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 3.046727272727273e-06,
+      "loss": 0.9658,
+      "step": 849500
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 3.0366262626262628e-06,
+      "loss": 0.9662,
+      "step": 850000
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 3.0265252525252526e-06,
+      "loss": 0.9644,
+      "step": 850500
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 3.016424242424243e-06,
+      "loss": 0.965,
+      "step": 851000
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 3.0063232323232328e-06,
+      "loss": 0.9661,
+      "step": 851500
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 2.9962222222222226e-06,
+      "loss": 0.9648,
+      "step": 852000
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 2.9861212121212125e-06,
+      "loss": 0.9642,
+      "step": 852500
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 2.9760202020202024e-06,
+      "loss": 0.9657,
+      "step": 853000
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 2.965939393939394e-06,
+      "loss": 0.9662,
+      "step": 853500
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 2.955838383838384e-06,
+      "loss": 0.9658,
+      "step": 854000
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 2.945737373737374e-06,
+      "loss": 0.9649,
+      "step": 854500
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 2.9356363636363638e-06,
+      "loss": 0.9649,
+      "step": 855000
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 2.9255353535353536e-06,
+      "loss": 0.9654,
+      "step": 855500
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 2.9154343434343435e-06,
+      "loss": 0.9626,
+      "step": 856000
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 2.9053535353535357e-06,
+      "loss": 0.9629,
+      "step": 856500
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 2.8952525252525256e-06,
+      "loss": 0.966,
+      "step": 857000
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 2.8851515151515155e-06,
+      "loss": 0.9662,
+      "step": 857500
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 2.8750707070707073e-06,
+      "loss": 0.9646,
+      "step": 858000
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 2.864969696969697e-06,
+      "loss": 0.9634,
+      "step": 858500
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 2.854868686868687e-06,
+      "loss": 0.9679,
+      "step": 859000
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 2.844767676767677e-06,
+      "loss": 0.9643,
+      "step": 859500
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 2.8346666666666667e-06,
+      "loss": 0.964,
+      "step": 860000
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 2.824565656565657e-06,
+      "loss": 0.9669,
+      "step": 860500
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 2.814464646464647e-06,
+      "loss": 0.965,
+      "step": 861000
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 2.8043636363636368e-06,
+      "loss": 0.9656,
+      "step": 861500
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 2.7942626262626266e-06,
+      "loss": 0.9636,
+      "step": 862000
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 2.7841818181818184e-06,
+      "loss": 0.9645,
+      "step": 862500
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 2.7740808080808083e-06,
+      "loss": 0.9648,
+      "step": 863000
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 2.763979797979798e-06,
+      "loss": 0.965,
+      "step": 863500
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 2.753878787878788e-06,
+      "loss": 0.9648,
+      "step": 864000
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 2.743777777777778e-06,
+      "loss": 0.965,
+      "step": 864500
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 2.7336767676767678e-06,
+      "loss": 0.9664,
+      "step": 865000
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 2.7235757575757576e-06,
+      "loss": 0.9638,
+      "step": 865500
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 2.7134747474747475e-06,
+      "loss": 0.9665,
+      "step": 866000
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 2.7033737373737378e-06,
+      "loss": 0.9631,
+      "step": 866500
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 2.6932929292929296e-06,
+      "loss": 0.9641,
+      "step": 867000
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 2.683191919191919e-06,
+      "loss": 0.9657,
+      "step": 867500
+    },
+    {
+      "epoch": 37.0,
+      "eval_accuracy": 0.8055344760931228,
+      "eval_loss": 0.8070488572120667,
+      "eval_runtime": 55.2192,
+      "eval_samples_per_second": 937.065,
+      "eval_steps_per_second": 14.651,
+      "step": 867983
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 2.673090909090909e-06,
+      "loss": 0.9638,
+      "step": 868000
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 2.663010101010101e-06,
+      "loss": 0.9631,
+      "step": 868500
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 2.652909090909091e-06,
+      "loss": 0.9634,
+      "step": 869000
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 2.6428282828282833e-06,
+      "loss": 0.9648,
+      "step": 869500
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 2.6327272727272727e-06,
+      "loss": 0.9652,
+      "step": 870000
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 2.6226262626262626e-06,
+      "loss": 0.9628,
+      "step": 870500
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 2.6125252525252524e-06,
+      "loss": 0.9637,
+      "step": 871000
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 2.6024242424242423e-06,
+      "loss": 0.9633,
+      "step": 871500
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 2.592323232323233e-06,
+      "loss": 0.9636,
+      "step": 872000
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 2.5822424242424244e-06,
+      "loss": 0.9613,
+      "step": 872500
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 2.5721414141414143e-06,
+      "loss": 0.9656,
+      "step": 873000
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 2.562040404040404e-06,
+      "loss": 0.9632,
+      "step": 873500
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 2.5519393939393944e-06,
+      "loss": 0.9612,
+      "step": 874000
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 2.5418383838383843e-06,
+      "loss": 0.9634,
+      "step": 874500
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 2.531737373737374e-06,
+      "loss": 0.9632,
+      "step": 875000
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 2.521636363636364e-06,
+      "loss": 0.9633,
+      "step": 875500
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 2.511535353535354e-06,
+      "loss": 0.9645,
+      "step": 876000
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 2.5014343434343437e-06,
+      "loss": 0.963,
+      "step": 876500
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 2.4913535353535356e-06,
+      "loss": 0.9627,
+      "step": 877000
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 2.4812525252525254e-06,
+      "loss": 0.9636,
+      "step": 877500
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 2.4711515151515153e-06,
+      "loss": 0.9645,
+      "step": 878000
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 2.461050505050505e-06,
+      "loss": 0.965,
+      "step": 878500
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 2.4509494949494954e-06,
+      "loss": 0.9637,
+      "step": 879000
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 2.440848484848485e-06,
+      "loss": 0.9645,
+      "step": 879500
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 2.4307474747474747e-06,
+      "loss": 0.9643,
+      "step": 880000
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 2.4206464646464646e-06,
+      "loss": 0.9639,
+      "step": 880500
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 2.410545454545455e-06,
+      "loss": 0.9648,
+      "step": 881000
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 2.4004444444444447e-06,
+      "loss": 0.9638,
+      "step": 881500
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 2.3903636363636366e-06,
+      "loss": 0.9647,
+      "step": 882000
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 2.3802626262626264e-06,
+      "loss": 0.9649,
+      "step": 882500
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 2.3701616161616163e-06,
+      "loss": 0.9639,
+      "step": 883000
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 2.360060606060606e-06,
+      "loss": 0.9648,
+      "step": 883500
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 2.349979797979798e-06,
+      "loss": 0.9615,
+      "step": 884000
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 2.3398989898989902e-06,
+      "loss": 0.965,
+      "step": 884500
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 2.32979797979798e-06,
+      "loss": 0.9656,
+      "step": 885000
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 2.31969696969697e-06,
+      "loss": 0.9633,
+      "step": 885500
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 2.30959595959596e-06,
+      "loss": 0.9634,
+      "step": 886000
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 2.2994949494949497e-06,
+      "loss": 0.9659,
+      "step": 886500
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 2.2893939393939395e-06,
+      "loss": 0.9632,
+      "step": 887000
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 2.2793131313131314e-06,
+      "loss": 0.9635,
+      "step": 887500
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 2.2692121212121217e-06,
+      "loss": 0.9641,
+      "step": 888000
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 2.2591111111111115e-06,
+      "loss": 0.9642,
+      "step": 888500
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 2.2490101010101014e-06,
+      "loss": 0.9639,
+      "step": 889000
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 2.238929292929293e-06,
+      "loss": 0.9637,
+      "step": 889500
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 2.228828282828283e-06,
+      "loss": 0.9633,
+      "step": 890000
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 2.218727272727273e-06,
+      "loss": 0.9638,
+      "step": 890500
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 2.208626262626263e-06,
+      "loss": 0.964,
+      "step": 891000
+    },
+    {
+      "epoch": 38.0,
+      "eval_accuracy": 0.8058275368916192,
+      "eval_loss": 0.8062019944190979,
+      "eval_runtime": 55.7349,
+      "eval_samples_per_second": 928.395,
+      "eval_steps_per_second": 14.515,
+      "step": 891442
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 2.1985252525252527e-06,
+      "loss": 0.9647,
+      "step": 891500
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 2.1884242424242425e-06,
+      "loss": 0.9649,
+      "step": 892000
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 2.1783232323232324e-06,
+      "loss": 0.9639,
+      "step": 892500
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 2.1682222222222223e-06,
+      "loss": 0.9625,
+      "step": 893000
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 2.158141414141414e-06,
+      "loss": 0.9633,
+      "step": 893500
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 2.1480404040404044e-06,
+      "loss": 0.963,
+      "step": 894000
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 2.1379393939393942e-06,
+      "loss": 0.963,
+      "step": 894500
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 2.127838383838384e-06,
+      "loss": 0.9632,
+      "step": 895000
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 2.117737373737374e-06,
+      "loss": 0.9638,
+      "step": 895500
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 2.107636363636364e-06,
+      "loss": 0.9663,
+      "step": 896000
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 2.0975353535353537e-06,
+      "loss": 0.9623,
+      "step": 896500
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 2.0874545454545455e-06,
+      "loss": 0.9629,
+      "step": 897000
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 2.0773535353535354e-06,
+      "loss": 0.9635,
+      "step": 897500
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 2.0672525252525257e-06,
+      "loss": 0.9633,
+      "step": 898000
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 2.0571515151515155e-06,
+      "loss": 0.965,
+      "step": 898500
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 2.0470505050505054e-06,
+      "loss": 0.963,
+      "step": 899000
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 2.036949494949495e-06,
+      "loss": 0.965,
+      "step": 899500
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 2.026868686868687e-06,
+      "loss": 0.9613,
+      "step": 900000
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 2.016767676767677e-06,
+      "loss": 0.9622,
+      "step": 900500
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 2.006666666666667e-06,
+      "loss": 0.9625,
+      "step": 901000
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 1.9965656565656567e-06,
+      "loss": 0.9613,
+      "step": 901500
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 1.9864848484848485e-06,
+      "loss": 0.9614,
+      "step": 902000
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 1.9763838383838383e-06,
+      "loss": 0.9631,
+      "step": 902500
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 1.966282828282828e-06,
+      "loss": 0.9629,
+      "step": 903000
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 1.9561818181818185e-06,
+      "loss": 0.9639,
+      "step": 903500
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 1.9460808080808084e-06,
+      "loss": 0.9624,
+      "step": 904000
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 1.9359797979797982e-06,
+      "loss": 0.96,
+      "step": 904500
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 1.925878787878788e-06,
+      "loss": 0.96,
+      "step": 905000
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 1.91579797979798e-06,
+      "loss": 0.961,
+      "step": 905500
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 1.9056969696969698e-06,
+      "loss": 0.9639,
+      "step": 906000
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 1.8955959595959596e-06,
+      "loss": 0.9627,
+      "step": 906500
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 1.8854949494949495e-06,
+      "loss": 0.9633,
+      "step": 907000
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 1.8754141414141415e-06,
+      "loss": 0.962,
+      "step": 907500
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 1.8653333333333334e-06,
+      "loss": 0.9632,
+      "step": 908000
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 1.8552323232323232e-06,
+      "loss": 0.9617,
+      "step": 908500
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 1.8451313131313133e-06,
+      "loss": 0.9647,
+      "step": 909000
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 1.8350303030303032e-06,
+      "loss": 0.9642,
+      "step": 909500
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 1.824929292929293e-06,
+      "loss": 0.9613,
+      "step": 910000
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 1.8148282828282829e-06,
+      "loss": 0.9599,
+      "step": 910500
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 1.804727272727273e-06,
+      "loss": 0.961,
+      "step": 911000
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 1.7946262626262628e-06,
+      "loss": 0.9637,
+      "step": 911500
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 1.7845252525252527e-06,
+      "loss": 0.9601,
+      "step": 912000
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 1.7744444444444445e-06,
+      "loss": 0.9644,
+      "step": 912500
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 1.7643434343434346e-06,
+      "loss": 0.9618,
+      "step": 913000
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 1.7542424242424245e-06,
+      "loss": 0.962,
+      "step": 913500
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 1.7441414141414143e-06,
+      "loss": 0.965,
+      "step": 914000
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 1.7340404040404042e-06,
+      "loss": 0.9614,
+      "step": 914500
+    },
+    {
+      "epoch": 39.0,
+      "eval_accuracy": 0.8056050030390269,
+      "eval_loss": 0.8058241009712219,
+      "eval_runtime": 55.9643,
+      "eval_samples_per_second": 924.59,
+      "eval_steps_per_second": 14.456,
+      "step": 914901
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 1.7239393939393943e-06,
+      "loss": 0.9632,
+      "step": 915000
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 1.7138383838383841e-06,
+      "loss": 0.9652,
+      "step": 915500
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 1.7037373737373738e-06,
+      "loss": 0.9623,
+      "step": 916000
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 1.6936363636363636e-06,
+      "loss": 0.9614,
+      "step": 916500
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 1.6835353535353537e-06,
+      "loss": 0.9612,
+      "step": 917000
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 1.6734545454545455e-06,
+      "loss": 0.9646,
+      "step": 917500
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 1.6633535353535354e-06,
+      "loss": 0.9623,
+      "step": 918000
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 1.6532727272727274e-06,
+      "loss": 0.9611,
+      "step": 918500
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 1.6431919191919193e-06,
+      "loss": 0.9623,
+      "step": 919000
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 1.6330909090909091e-06,
+      "loss": 0.9628,
+      "step": 919500
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 1.622989898989899e-06,
+      "loss": 0.9616,
+      "step": 920000
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 1.612888888888889e-06,
+      "loss": 0.9641,
+      "step": 920500
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 1.6028080808080809e-06,
+      "loss": 0.9607,
+      "step": 921000
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 1.5927070707070707e-06,
+      "loss": 0.9628,
+      "step": 921500
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 1.5826060606060608e-06,
+      "loss": 0.965,
+      "step": 922000
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 1.5725050505050507e-06,
+      "loss": 0.9625,
+      "step": 922500
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 1.5624040404040405e-06,
+      "loss": 0.9605,
+      "step": 923000
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 1.5523030303030304e-06,
+      "loss": 0.9626,
+      "step": 923500
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 1.5422020202020203e-06,
+      "loss": 0.9635,
+      "step": 924000
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 1.5321010101010103e-06,
+      "loss": 0.9628,
+      "step": 924500
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 1.5220000000000002e-06,
+      "loss": 0.9629,
+      "step": 925000
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 1.51189898989899e-06,
+      "loss": 0.9615,
+      "step": 925500
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 1.5017979797979797e-06,
+      "loss": 0.9616,
+      "step": 926000
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 1.49169696969697e-06,
+      "loss": 0.9622,
+      "step": 926500
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 1.4816161616161618e-06,
+      "loss": 0.9615,
+      "step": 927000
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 1.4715151515151515e-06,
+      "loss": 0.9619,
+      "step": 927500
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 1.4614343434343437e-06,
+      "loss": 0.9615,
+      "step": 928000
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 1.4513333333333334e-06,
+      "loss": 0.9623,
+      "step": 928500
+    },
+    {
+      "epoch": 39.6,
+      "learning_rate": 1.4412525252525252e-06,
+      "loss": 0.9616,
+      "step": 929000
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 1.4311515151515155e-06,
+      "loss": 0.9605,
+      "step": 929500
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 1.4210505050505051e-06,
+      "loss": 0.9638,
+      "step": 930000
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 1.410949494949495e-06,
+      "loss": 0.9612,
+      "step": 930500
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 1.4008686868686868e-06,
+      "loss": 0.9631,
+      "step": 931000
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 1.390767676767677e-06,
+      "loss": 0.9585,
+      "step": 931500
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 1.3806666666666668e-06,
+      "loss": 0.9616,
+      "step": 932000
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 1.3705656565656566e-06,
+      "loss": 0.9606,
+      "step": 932500
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 1.3604646464646465e-06,
+      "loss": 0.9626,
+      "step": 933000
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 1.3503636363636366e-06,
+      "loss": 0.9629,
+      "step": 933500
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 1.3402626262626264e-06,
+      "loss": 0.9627,
+      "step": 934000
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 1.3301616161616163e-06,
+      "loss": 0.9609,
+      "step": 934500
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 1.3200606060606062e-06,
+      "loss": 0.9605,
+      "step": 935000
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 1.309959595959596e-06,
+      "loss": 0.9613,
+      "step": 935500
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 1.299858585858586e-06,
+      "loss": 0.9616,
+      "step": 936000
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 1.289757575757576e-06,
+      "loss": 0.9606,
+      "step": 936500
+    },
+    {
+      "epoch": 39.94,
+      "learning_rate": 1.2796565656565656e-06,
+      "loss": 0.962,
+      "step": 937000
+    },
+    {
+      "epoch": 39.96,
+      "learning_rate": 1.2695555555555555e-06,
+      "loss": 0.9614,
+      "step": 937500
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 1.2594545454545456e-06,
+      "loss": 0.9615,
+      "step": 938000
+    },
+    {
+      "epoch": 40.0,
+      "eval_accuracy": 0.8062926496749798,
+      "eval_loss": 0.8033291101455688,
+      "eval_runtime": 56.4926,
+      "eval_samples_per_second": 915.944,
+      "eval_steps_per_second": 14.32,
+      "step": 938360
+    },
+    {
+      "epoch": 40.01,
+      "learning_rate": 1.2493535353535354e-06,
+      "loss": 0.964,
+      "step": 938500
+    },
+    {
+      "epoch": 40.03,
+      "learning_rate": 1.2392727272727275e-06,
+      "loss": 0.9632,
+      "step": 939000
+    },
+    {
+      "epoch": 40.05,
+      "learning_rate": 1.2291717171717173e-06,
+      "loss": 0.9612,
+      "step": 939500
+    },
+    {
+      "epoch": 40.07,
+      "learning_rate": 1.2190707070707072e-06,
+      "loss": 0.961,
+      "step": 940000
+    },
+    {
+      "epoch": 40.09,
+      "learning_rate": 1.208969696969697e-06,
+      "loss": 0.962,
+      "step": 940500
+    },
+    {
+      "epoch": 40.11,
+      "learning_rate": 1.198868686868687e-06,
+      "loss": 0.9598,
+      "step": 941000
+    },
+    {
+      "epoch": 40.13,
+      "learning_rate": 1.188787878787879e-06,
+      "loss": 0.9581,
+      "step": 941500
+    },
+    {
+      "epoch": 40.16,
+      "learning_rate": 1.1787070707070708e-06,
+      "loss": 0.9647,
+      "step": 942000
+    },
+    {
+      "epoch": 40.18,
+      "learning_rate": 1.1686060606060606e-06,
+      "loss": 0.9607,
+      "step": 942500
+    },
+    {
+      "epoch": 40.2,
+      "learning_rate": 1.1585050505050507e-06,
+      "loss": 0.9603,
+      "step": 943000
+    },
+    {
+      "epoch": 40.22,
+      "learning_rate": 1.1484040404040404e-06,
+      "loss": 0.9633,
+      "step": 943500
+    },
+    {
+      "epoch": 40.24,
+      "learning_rate": 1.1383232323232324e-06,
+      "loss": 0.9632,
+      "step": 944000
+    },
+    {
+      "epoch": 40.26,
+      "learning_rate": 1.1282222222222223e-06,
+      "loss": 0.9639,
+      "step": 944500
+    },
+    {
+      "epoch": 40.28,
+      "learning_rate": 1.1181212121212121e-06,
+      "loss": 0.9605,
+      "step": 945000
+    },
+    {
+      "epoch": 40.3,
+      "learning_rate": 1.1080202020202022e-06,
+      "loss": 0.9621,
+      "step": 945500
+    },
+    {
+      "epoch": 40.33,
+      "learning_rate": 1.097919191919192e-06,
+      "loss": 0.962,
+      "step": 946000
+    },
+    {
+      "epoch": 40.35,
+      "learning_rate": 1.087818181818182e-06,
+      "loss": 0.9634,
+      "step": 946500
+    },
+    {
+      "epoch": 40.37,
+      "learning_rate": 1.0777171717171718e-06,
+      "loss": 0.961,
+      "step": 947000
+    },
+    {
+      "epoch": 40.39,
+      "learning_rate": 1.0676161616161619e-06,
+      "loss": 0.9592,
+      "step": 947500
+    },
+    {
+      "epoch": 40.41,
+      "learning_rate": 1.0575151515151515e-06,
+      "loss": 0.9605,
+      "step": 948000
+    },
+    {
+      "epoch": 40.43,
+      "learning_rate": 1.0474141414141414e-06,
+      "loss": 0.9634,
+      "step": 948500
+    },
+    {
+      "epoch": 40.45,
+      "learning_rate": 1.0373333333333334e-06,
+      "loss": 0.9596,
+      "step": 949000
+    },
+    {
+      "epoch": 40.47,
+      "learning_rate": 1.0272323232323233e-06,
+      "loss": 0.9636,
+      "step": 949500
+    },
+    {
+      "epoch": 40.5,
+      "learning_rate": 1.0171313131313131e-06,
+      "loss": 0.9626,
+      "step": 950000
+    },
+    {
+      "epoch": 40.52,
+      "learning_rate": 1.0070505050505052e-06,
+      "loss": 0.9594,
+      "step": 950500
+    },
+    {
+      "epoch": 40.54,
+      "learning_rate": 9.96949494949495e-07,
+      "loss": 0.9593,
+      "step": 951000
+    },
+    {
+      "epoch": 40.56,
+      "learning_rate": 9.86848484848485e-07,
+      "loss": 0.9615,
+      "step": 951500
+    },
+    {
+      "epoch": 40.58,
+      "learning_rate": 9.767474747474748e-07,
+      "loss": 0.9602,
+      "step": 952000
+    },
+    {
+      "epoch": 40.6,
+      "learning_rate": 9.666666666666668e-07,
+      "loss": 0.9623,
+      "step": 952500
+    },
+    {
+      "epoch": 40.62,
+      "learning_rate": 9.565656565656567e-07,
+      "loss": 0.9585,
+      "step": 953000
+    },
+    {
+      "epoch": 40.65,
+      "learning_rate": 9.464646464646465e-07,
+      "loss": 0.9619,
+      "step": 953500
+    },
+    {
+      "epoch": 40.67,
+      "learning_rate": 9.363636363636365e-07,
+      "loss": 0.9603,
+      "step": 954000
+    },
+    {
+      "epoch": 40.69,
+      "learning_rate": 9.262626262626264e-07,
+      "loss": 0.9618,
+      "step": 954500
+    },
+    {
+      "epoch": 40.71,
+      "learning_rate": 9.161616161616162e-07,
+      "loss": 0.9598,
+      "step": 955000
+    },
+    {
+      "epoch": 40.73,
+      "learning_rate": 9.060606060606062e-07,
+      "loss": 0.9633,
+      "step": 955500
+    },
+    {
+      "epoch": 40.75,
+      "learning_rate": 8.959595959595959e-07,
+      "loss": 0.9603,
+      "step": 956000
+    },
+    {
+      "epoch": 40.77,
+      "learning_rate": 8.858585858585859e-07,
+      "loss": 0.9611,
+      "step": 956500
+    },
+    {
+      "epoch": 40.79,
+      "learning_rate": 8.757777777777778e-07,
+      "loss": 0.9605,
+      "step": 957000
+    },
+    {
+      "epoch": 40.82,
+      "learning_rate": 8.656969696969697e-07,
+      "loss": 0.9633,
+      "step": 957500
+    },
+    {
+      "epoch": 40.84,
+      "learning_rate": 8.555959595959596e-07,
+      "loss": 0.9603,
+      "step": 958000
+    },
+    {
+      "epoch": 40.86,
+      "learning_rate": 8.454949494949495e-07,
+      "loss": 0.9607,
+      "step": 958500
+    },
+    {
+      "epoch": 40.88,
+      "learning_rate": 8.354141414141414e-07,
+      "loss": 0.9604,
+      "step": 959000
+    },
+    {
+      "epoch": 40.9,
+      "learning_rate": 8.253131313131314e-07,
+      "loss": 0.9626,
+      "step": 959500
+    },
+    {
+      "epoch": 40.92,
+      "learning_rate": 8.152121212121213e-07,
+      "loss": 0.962,
+      "step": 960000
+    },
+    {
+      "epoch": 40.94,
+      "learning_rate": 8.051111111111112e-07,
+      "loss": 0.9592,
+      "step": 960500
+    },
+    {
+      "epoch": 40.97,
+      "learning_rate": 7.950101010101011e-07,
+      "loss": 0.9602,
+      "step": 961000
+    },
+    {
+      "epoch": 40.99,
+      "learning_rate": 7.849090909090911e-07,
+      "loss": 0.9614,
+      "step": 961500
+    },
+    {
+      "epoch": 41.0,
+      "eval_accuracy": 0.8060622863370541,
+      "eval_loss": 0.8039979338645935,
+      "eval_runtime": 56.2871,
+      "eval_samples_per_second": 919.287,
+      "eval_steps_per_second": 14.373,
+      "step": 961819
+    },
+    {
+      "epoch": 41.01,
+      "learning_rate": 7.748080808080808e-07,
+      "loss": 0.9642,
+      "step": 962000
+    },
+    {
+      "epoch": 41.03,
+      "learning_rate": 7.647070707070707e-07,
+      "loss": 0.962,
+      "step": 962500
+    },
+    {
+      "epoch": 41.05,
+      "learning_rate": 7.546060606060607e-07,
+      "loss": 0.9606,
+      "step": 963000
+    },
+    {
+      "epoch": 41.07,
+      "learning_rate": 7.445050505050505e-07,
+      "loss": 0.9606,
+      "step": 963500
+    },
+    {
+      "epoch": 41.09,
+      "learning_rate": 7.344040404040405e-07,
+      "loss": 0.9608,
+      "step": 964000
+    },
+    {
+      "epoch": 41.11,
+      "learning_rate": 7.243232323232324e-07,
+      "loss": 0.9612,
+      "step": 964500
+    },
+    {
+      "epoch": 41.14,
+      "learning_rate": 7.142222222222223e-07,
+      "loss": 0.9607,
+      "step": 965000
+    },
+    {
+      "epoch": 41.16,
+      "learning_rate": 7.041212121212123e-07,
+      "loss": 0.9593,
+      "step": 965500
+    },
+    {
+      "epoch": 41.18,
+      "learning_rate": 6.940202020202021e-07,
+      "loss": 0.9603,
+      "step": 966000
+    },
+    {
+      "epoch": 41.2,
+      "learning_rate": 6.839191919191921e-07,
+      "loss": 0.9603,
+      "step": 966500
+    },
+    {
+      "epoch": 41.22,
+      "learning_rate": 6.738181818181818e-07,
+      "loss": 0.9618,
+      "step": 967000
+    },
+    {
+      "epoch": 41.24,
+      "learning_rate": 6.637171717171717e-07,
+      "loss": 0.9599,
+      "step": 967500
+    },
+    {
+      "epoch": 41.26,
+      "learning_rate": 6.536161616161617e-07,
+      "loss": 0.9617,
+      "step": 968000
+    },
+    {
+      "epoch": 41.28,
+      "learning_rate": 6.435151515151515e-07,
+      "loss": 0.9601,
+      "step": 968500
+    },
+    {
+      "epoch": 41.31,
+      "learning_rate": 6.334343434343435e-07,
+      "loss": 0.9586,
+      "step": 969000
+    },
+    {
+      "epoch": 41.33,
+      "learning_rate": 6.233333333333333e-07,
+      "loss": 0.9605,
+      "step": 969500
+    },
+    {
+      "epoch": 41.35,
+      "learning_rate": 6.132525252525253e-07,
+      "loss": 0.9596,
+      "step": 970000
+    },
+    {
+      "epoch": 41.37,
+      "learning_rate": 6.031515151515152e-07,
+      "loss": 0.9599,
+      "step": 970500
+    },
+    {
+      "epoch": 41.39,
+      "learning_rate": 5.930505050505051e-07,
+      "loss": 0.9618,
+      "step": 971000
+    },
+    {
+      "epoch": 41.41,
+      "learning_rate": 5.829494949494951e-07,
+      "loss": 0.9635,
+      "step": 971500
+    },
+    {
+      "epoch": 41.43,
+      "learning_rate": 5.728484848484849e-07,
+      "loss": 0.9622,
+      "step": 972000
+    },
+    {
+      "epoch": 41.46,
+      "learning_rate": 5.627474747474748e-07,
+      "loss": 0.9614,
+      "step": 972500
+    },
+    {
+      "epoch": 41.48,
+      "learning_rate": 5.526464646464648e-07,
+      "loss": 0.9611,
+      "step": 973000
+    },
+    {
+      "epoch": 41.5,
+      "learning_rate": 5.425454545454545e-07,
+      "loss": 0.9584,
+      "step": 973500
+    },
+    {
+      "epoch": 41.52,
+      "learning_rate": 5.324646464646466e-07,
+      "loss": 0.9617,
+      "step": 974000
+    },
+    {
+      "epoch": 41.54,
+      "learning_rate": 5.223636363636364e-07,
+      "loss": 0.9606,
+      "step": 974500
+    },
+    {
+      "epoch": 41.56,
+      "learning_rate": 5.122626262626263e-07,
+      "loss": 0.9611,
+      "step": 975000
+    },
+    {
+      "epoch": 41.58,
+      "learning_rate": 5.021616161616162e-07,
+      "loss": 0.9607,
+      "step": 975500
+    },
+    {
+      "epoch": 41.6,
+      "learning_rate": 4.920606060606061e-07,
+      "loss": 0.9627,
+      "step": 976000
+    },
+    {
+      "epoch": 41.63,
+      "learning_rate": 4.82e-07,
+      "loss": 0.9629,
+      "step": 976500
+    },
+    {
+      "epoch": 41.65,
+      "learning_rate": 4.718989898989899e-07,
+      "loss": 0.9639,
+      "step": 977000
+    },
+    {
+      "epoch": 41.67,
+      "learning_rate": 4.6179797979797983e-07,
+      "loss": 0.9615,
+      "step": 977500
+    },
+    {
+      "epoch": 41.69,
+      "learning_rate": 4.5169696969696975e-07,
+      "loss": 0.9602,
+      "step": 978000
+    },
+    {
+      "epoch": 41.71,
+      "learning_rate": 4.4159595959595967e-07,
+      "loss": 0.9605,
+      "step": 978500
+    },
+    {
+      "epoch": 41.73,
+      "learning_rate": 4.3149494949494953e-07,
+      "loss": 0.9597,
+      "step": 979000
+    },
+    {
+      "epoch": 41.75,
+      "learning_rate": 4.2139393939393944e-07,
+      "loss": 0.9607,
+      "step": 979500
+    },
+    {
+      "epoch": 41.78,
+      "learning_rate": 4.1129292929292936e-07,
+      "loss": 0.96,
+      "step": 980000
+    },
+    {
+      "epoch": 41.8,
+      "learning_rate": 4.011919191919193e-07,
+      "loss": 0.9618,
+      "step": 980500
+    },
+    {
+      "epoch": 41.82,
+      "learning_rate": 3.9111111111111115e-07,
+      "loss": 0.9604,
+      "step": 981000
+    },
+    {
+      "epoch": 41.84,
+      "learning_rate": 3.81010101010101e-07,
+      "loss": 0.9586,
+      "step": 981500
+    },
+    {
+      "epoch": 41.86,
+      "learning_rate": 3.7090909090909093e-07,
+      "loss": 0.9619,
+      "step": 982000
+    },
+    {
+      "epoch": 41.88,
+      "learning_rate": 3.6082828282828286e-07,
+      "loss": 0.9599,
+      "step": 982500
+    },
+    {
+      "epoch": 41.9,
+      "learning_rate": 3.5072727272727273e-07,
+      "loss": 0.9607,
+      "step": 983000
+    },
+    {
+      "epoch": 41.92,
+      "learning_rate": 3.4062626262626264e-07,
+      "loss": 0.9586,
+      "step": 983500
+    },
+    {
+      "epoch": 41.95,
+      "learning_rate": 3.305252525252525e-07,
+      "loss": 0.9601,
+      "step": 984000
+    },
+    {
+      "epoch": 41.97,
+      "learning_rate": 3.204242424242424e-07,
+      "loss": 0.9604,
+      "step": 984500
+    },
+    {
+      "epoch": 41.99,
+      "learning_rate": 3.1034343434343435e-07,
+      "loss": 0.9631,
+      "step": 985000
+    },
+    {
+      "epoch": 42.0,
+      "eval_accuracy": 0.8062162078442503,
+      "eval_loss": 0.8020749688148499,
+      "eval_runtime": 56.8036,
+      "eval_samples_per_second": 910.929,
+      "eval_steps_per_second": 14.242,
+      "step": 985278
+    },
+    {
+      "epoch": 42.01,
+      "learning_rate": 3.0024242424242427e-07,
+      "loss": 0.9596,
+      "step": 985500
+    },
+    {
+      "epoch": 42.03,
+      "learning_rate": 2.901414141414142e-07,
+      "loss": 0.9596,
+      "step": 986000
+    },
+    {
+      "epoch": 42.05,
+      "learning_rate": 2.8004040404040405e-07,
+      "loss": 0.9598,
+      "step": 986500
+    },
+    {
+      "epoch": 42.07,
+      "learning_rate": 2.69959595959596e-07,
+      "loss": 0.9614,
+      "step": 987000
+    },
+    {
+      "epoch": 42.09,
+      "learning_rate": 2.5985858585858584e-07,
+      "loss": 0.9591,
+      "step": 987500
+    },
+    {
+      "epoch": 42.12,
+      "learning_rate": 2.4975757575757576e-07,
+      "loss": 0.9618,
+      "step": 988000
+    },
+    {
+      "epoch": 42.14,
+      "learning_rate": 2.396565656565657e-07,
+      "loss": 0.9617,
+      "step": 988500
+    },
+    {
+      "epoch": 42.16,
+      "learning_rate": 2.2955555555555556e-07,
+      "loss": 0.9605,
+      "step": 989000
+    },
+    {
+      "epoch": 42.18,
+      "learning_rate": 2.1945454545454548e-07,
+      "loss": 0.9624,
+      "step": 989500
+    },
+    {
+      "epoch": 42.2,
+      "learning_rate": 2.0935353535353537e-07,
+      "loss": 0.9607,
+      "step": 990000
+    },
+    {
+      "epoch": 42.22,
+      "learning_rate": 1.9925252525252529e-07,
+      "loss": 0.96,
+      "step": 990500
+    },
+    {
+      "epoch": 42.24,
+      "learning_rate": 1.891717171717172e-07,
+      "loss": 0.96,
+      "step": 991000
+    },
+    {
+      "epoch": 42.27,
+      "learning_rate": 1.790707070707071e-07,
+      "loss": 0.9589,
+      "step": 991500
+    },
+    {
+      "epoch": 42.29,
+      "learning_rate": 1.68969696969697e-07,
+      "loss": 0.9631,
+      "step": 992000
+    },
+    {
+      "epoch": 42.31,
+      "learning_rate": 1.5886868686868686e-07,
+      "loss": 0.9604,
+      "step": 992500
+    },
+    {
+      "epoch": 42.33,
+      "learning_rate": 1.4876767676767677e-07,
+      "loss": 0.9591,
+      "step": 993000
+    },
+    {
+      "epoch": 42.35,
+      "learning_rate": 1.386868686868687e-07,
+      "loss": 0.9608,
+      "step": 993500
+    },
+    {
+      "epoch": 42.37,
+      "learning_rate": 1.285858585858586e-07,
+      "loss": 0.9604,
+      "step": 994000
+    },
+    {
+      "epoch": 42.39,
+      "learning_rate": 1.1850505050505051e-07,
+      "loss": 0.9611,
+      "step": 994500
+    },
+    {
+      "epoch": 42.41,
+      "learning_rate": 1.0840404040404042e-07,
+      "loss": 0.9582,
+      "step": 995000
+    },
+    {
+      "epoch": 42.44,
+      "learning_rate": 9.83030303030303e-08,
+      "loss": 0.9607,
+      "step": 995500
+    },
+    {
+      "epoch": 42.46,
+      "learning_rate": 8.820202020202021e-08,
+      "loss": 0.9589,
+      "step": 996000
+    },
+    {
+      "epoch": 42.48,
+      "learning_rate": 7.810101010101011e-08,
+      "loss": 0.9614,
+      "step": 996500
+    },
+    {
+      "epoch": 42.5,
+      "learning_rate": 6.8e-08,
+      "loss": 0.9607,
+      "step": 997000
+    },
+    {
+      "epoch": 42.52,
+      "learning_rate": 5.78989898989899e-08,
+      "loss": 0.9609,
+      "step": 997500
+    },
+    {
+      "epoch": 42.54,
+      "learning_rate": 4.781818181818183e-08,
+      "loss": 0.9631,
+      "step": 998000
+    },
+    {
+      "epoch": 42.56,
+      "learning_rate": 3.7717171717171717e-08,
+      "loss": 0.961,
+      "step": 998500
+    },
+    {
+      "epoch": 42.58,
+      "learning_rate": 2.761616161616162e-08,
+      "loss": 0.9614,
+      "step": 999000
+    },
+    {
+      "epoch": 42.61,
+      "learning_rate": 1.7515151515151516e-08,
+      "loss": 0.9611,
+      "step": 999500
+    },
+    {
+      "epoch": 42.63,
+      "learning_rate": 7.434343434343435e-09,
+      "loss": 0.9609,
+      "step": 1000000
+    },
+    {
+      "epoch": 42.63,
+      "eval_accuracy": 0.8068319957114065,
+      "eval_loss": 0.8024299144744873,
+      "eval_runtime": 56.1367,
+      "eval_samples_per_second": 921.749,
+      "eval_steps_per_second": 14.411,
+      "step": 1000000
+    },
+    {
+      "epoch": 42.63,
+      "step": 1000000,
+      "total_flos": 6.737707398739683e+19,
+      "train_loss": 0.07689679595947266,
+      "train_runtime": 32435.3069,
+      "train_samples_per_second": 7892.634,
+      "train_steps_per_second": 30.831
+    }
+  ],
+  "max_steps": 1000000,
+  "num_train_epochs": 43,
+  "total_flos": 6.737707398739683e+19,
+  "trial_name": null,
+  "trial_params": null
+}