diff --git "a/checkpoint-2900160/trainer_state.json" "b/checkpoint-2900160/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2900160/trainer_state.json"
@@ -0,0 +1,34313 @@
+{
+  "best_metric": 3.9728095531463623,
+  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-case/lstm/4/checkpoints/checkpoint-2900160",
+  "epoch": 0.025000606015738065,
+  "eval_steps": 10,
+  "global_step": 2900160,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998362119627e-05,
+      "loss": 10.8203,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999161405248948e-05,
+      "loss": 7.5554,
+      "step": 512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998322810497896e-05,
+      "loss": 7.0617,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997484215746844e-05,
+      "loss": 6.9928,
+      "step": 1536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996645620995792e-05,
+      "loss": 6.9463,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99580702624474e-05,
+      "loss": 6.8938,
+      "step": 2560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994968431493688e-05,
+      "loss": 6.7206,
+      "step": 3072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994131474623009e-05,
+      "loss": 6.6102,
+      "step": 3584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993292879871958e-05,
+      "loss": 6.5164,
+      "step": 4096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992454285120906e-05,
+      "loss": 6.447,
+      "step": 4608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.991615690369854e-05,
+      "loss": 6.3857,
+      "step": 5120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.990777095618801e-05,
+      "loss": 6.328,
+      "step": 5632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989938500867749e-05,
+      "loss": 6.2684,
+      "step": 6144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989099906116697e-05,
+      "loss": 6.2061,
+      "step": 6656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.988261311365645e-05,
+      "loss": 6.1519,
+      "step": 7168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.987422716614593e-05,
+      "loss": 6.0943,
+      "step": 7680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986584121863541e-05,
+      "loss": 6.0496,
+      "step": 8192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.985745527112489e-05,
+      "loss": 6.0038,
+      "step": 8704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984906932361437e-05,
+      "loss": 5.9633,
+      "step": 9216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984068337610385e-05,
+      "loss": 5.9267,
+      "step": 9728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.983229742859333e-05,
+      "loss": 5.8922,
+      "step": 10240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.982391148108281e-05,
+      "loss": 5.8559,
+      "step": 10752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.981552553357229e-05,
+      "loss": 5.8247,
+      "step": 11264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9807155964865506e-05,
+      "loss": 5.796,
+      "step": 11776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9798770017354986e-05,
+      "loss": 5.7777,
+      "step": 12288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9790384069844466e-05,
+      "loss": 5.7387,
+      "step": 12800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9781998122333946e-05,
+      "loss": 5.7171,
+      "step": 13312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9773612174823426e-05,
+      "loss": 5.6922,
+      "step": 13824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.97652262273129e-05,
+      "loss": 5.6726,
+      "step": 14336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.975684027980238e-05,
+      "loss": 5.6463,
+      "step": 14848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.974845433229186e-05,
+      "loss": 5.6222,
+      "step": 15360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.974006838478134e-05,
+      "loss": 5.6155,
+      "step": 15872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973168243727082e-05,
+      "loss": 5.5884,
+      "step": 16384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97232964897603e-05,
+      "loss": 5.5783,
+      "step": 16896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9714926921053515e-05,
+      "loss": 5.5648,
+      "step": 17408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9706557352346724e-05,
+      "loss": 5.5395,
+      "step": 17920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9698171404836204e-05,
+      "loss": 5.5286,
+      "step": 18432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9689785457325684e-05,
+      "loss": 5.4887,
+      "step": 18944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9681399509815164e-05,
+      "loss": 5.4899,
+      "step": 19456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.967302994110837e-05,
+      "loss": 5.4549,
+      "step": 19968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.966464399359785e-05,
+      "loss": 5.4551,
+      "step": 20480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965625804608733e-05,
+      "loss": 5.4362,
+      "step": 20992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964787209857681e-05,
+      "loss": 5.435,
+      "step": 21504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963948615106629e-05,
+      "loss": 5.4091,
+      "step": 22016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96311165823595e-05,
+      "loss": 5.4015,
+      "step": 22528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962273063484898e-05,
+      "loss": 5.3932,
+      "step": 23040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.961434468733847e-05,
+      "loss": 5.3755,
+      "step": 23552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960595873982795e-05,
+      "loss": 5.3797,
+      "step": 24064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.959757279231743e-05,
+      "loss": 5.3497,
+      "step": 24576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958920322361064e-05,
+      "loss": 5.3351,
+      "step": 25088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958081727610012e-05,
+      "loss": 5.3448,
+      "step": 25600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95724313285896e-05,
+      "loss": 5.3373,
+      "step": 26112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.956404538107908e-05,
+      "loss": 5.3124,
+      "step": 26624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.955565943356856e-05,
+      "loss": 5.301,
+      "step": 27136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.954728986486177e-05,
+      "loss": 5.2988,
+      "step": 27648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953890391735125e-05,
+      "loss": 5.283,
+      "step": 28160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9530517969840727e-05,
+      "loss": 5.2966,
+      "step": 28672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9522132022330207e-05,
+      "loss": 5.2574,
+      "step": 29184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951376245362342e-05,
+      "loss": 5.2572,
+      "step": 29696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95053765061129e-05,
+      "loss": 5.2491,
+      "step": 30208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.949700693740611e-05,
+      "loss": 5.2377,
+      "step": 30720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948862098989559e-05,
+      "loss": 5.2392,
+      "step": 31232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948023504238507e-05,
+      "loss": 5.2173,
+      "step": 31744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947184909487455e-05,
+      "loss": 5.2118,
+      "step": 32256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.946346314736403e-05,
+      "loss": 5.2034,
+      "step": 32768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.945507719985351e-05,
+      "loss": 5.2166,
+      "step": 33280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.944669125234299e-05,
+      "loss": 5.1889,
+      "step": 33792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.943830530483247e-05,
+      "loss": 5.1878,
+      "step": 34304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.942991935732195e-05,
+      "loss": 5.1617,
+      "step": 34816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.942154978861516e-05,
+      "loss": 5.1689,
+      "step": 35328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.941316384110464e-05,
+      "loss": 5.1778,
+      "step": 35840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9404794272397856e-05,
+      "loss": 5.1681,
+      "step": 36352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9396408324887336e-05,
+      "loss": 5.1465,
+      "step": 36864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9388022377376816e-05,
+      "loss": 5.1595,
+      "step": 37376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9379636429866296e-05,
+      "loss": 5.1548,
+      "step": 37888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9371250482355776e-05,
+      "loss": 5.1311,
+      "step": 38400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9362864534845256e-05,
+      "loss": 5.1289,
+      "step": 38912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.935447858733473e-05,
+      "loss": 5.1113,
+      "step": 39424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.934609263982421e-05,
+      "loss": 5.0985,
+      "step": 39936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.933770669231369e-05,
+      "loss": 5.0977,
+      "step": 40448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9329337123606905e-05,
+      "loss": 5.1003,
+      "step": 40960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.932095117609638e-05,
+      "loss": 5.0981,
+      "step": 41472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931256522858586e-05,
+      "loss": 5.1038,
+      "step": 41984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9304195659879074e-05,
+      "loss": 5.0892,
+      "step": 42496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9295809712368554e-05,
+      "loss": 5.0694,
+      "step": 43008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9287423764858034e-05,
+      "loss": 5.0711,
+      "step": 43520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9279037817347514e-05,
+      "loss": 5.0705,
+      "step": 44032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9270651869836994e-05,
+      "loss": 5.0586,
+      "step": 44544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9262265922326474e-05,
+      "loss": 5.0501,
+      "step": 45056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9253879974815954e-05,
+      "loss": 5.0399,
+      "step": 45568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9245494027305433e-05,
+      "loss": 5.0393,
+      "step": 46080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9237108079794913e-05,
+      "loss": 5.0403,
+      "step": 46592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922873851108812e-05,
+      "loss": 5.0274,
+      "step": 47104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922036894238133e-05,
+      "loss": 5.0294,
+      "step": 47616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921198299487081e-05,
+      "loss": 5.0181,
+      "step": 48128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.920359704736029e-05,
+      "loss": 5.0141,
+      "step": 48640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919521109984978e-05,
+      "loss": 4.9998,
+      "step": 49152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918684153114299e-05,
+      "loss": 4.9976,
+      "step": 49664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917845558363247e-05,
+      "loss": 4.9913,
+      "step": 50176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917006963612195e-05,
+      "loss": 4.9944,
+      "step": 50688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.916168368861143e-05,
+      "loss": 4.9953,
+      "step": 51200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915329774110091e-05,
+      "loss": 4.9728,
+      "step": 51712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914491179359039e-05,
+      "loss": 4.978,
+      "step": 52224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.913652584607987e-05,
+      "loss": 4.9667,
+      "step": 52736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9128156277373076e-05,
+      "loss": 4.9644,
+      "step": 53248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9119770329862556e-05,
+      "loss": 4.9573,
+      "step": 53760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9111384382352036e-05,
+      "loss": 4.9506,
+      "step": 54272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9102998434841516e-05,
+      "loss": 4.9497,
+      "step": 54784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9094612487330996e-05,
+      "loss": 4.9416,
+      "step": 55296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9086226539820476e-05,
+      "loss": 4.939,
+      "step": 55808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.907784059230996e-05,
+      "loss": 4.9309,
+      "step": 56320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906945464479944e-05,
+      "loss": 4.9213,
+      "step": 56832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906106869728892e-05,
+      "loss": 4.9316,
+      "step": 57344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9052682749778396e-05,
+      "loss": 4.9214,
+      "step": 57856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9044296802267876e-05,
+      "loss": 4.9222,
+      "step": 58368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.903592723356109e-05,
+      "loss": 4.9173,
+      "step": 58880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9027541286050565e-05,
+      "loss": 4.9176,
+      "step": 59392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.901917171734378e-05,
+      "loss": 4.9086,
+      "step": 59904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.901078576983326e-05,
+      "loss": 4.8933,
+      "step": 60416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.900239982232274e-05,
+      "loss": 4.901,
+      "step": 60928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8994013874812214e-05,
+      "loss": 4.9055,
+      "step": 61440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89856279273017e-05,
+      "loss": 4.8914,
+      "step": 61952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897724197979118e-05,
+      "loss": 4.8845,
+      "step": 62464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.896885603228066e-05,
+      "loss": 4.882,
+      "step": 62976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.896047008477014e-05,
+      "loss": 4.8806,
+      "step": 63488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.895210051606335e-05,
+      "loss": 4.8705,
+      "step": 64000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.894371456855283e-05,
+      "loss": 4.8736,
+      "step": 64512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.893532862104231e-05,
+      "loss": 4.8628,
+      "step": 65024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.892695905233552e-05,
+      "loss": 4.8634,
+      "step": 65536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8918573104825e-05,
+      "loss": 4.8545,
+      "step": 66048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891018715731448e-05,
+      "loss": 4.8728,
+      "step": 66560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.890180120980396e-05,
+      "loss": 4.855,
+      "step": 67072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.889343164109717e-05,
+      "loss": 4.846,
+      "step": 67584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.888504569358665e-05,
+      "loss": 4.8419,
+      "step": 68096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8876659746076134e-05,
+      "loss": 4.8434,
+      "step": 68608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8868273798565614e-05,
+      "loss": 4.8446,
+      "step": 69120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8859887851055094e-05,
+      "loss": 4.8441,
+      "step": 69632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8851501903544574e-05,
+      "loss": 4.8329,
+      "step": 70144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8843115956034054e-05,
+      "loss": 4.8338,
+      "step": 70656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8834730008523534e-05,
+      "loss": 4.8293,
+      "step": 71168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8826344061013014e-05,
+      "loss": 4.8271,
+      "step": 71680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.881797449230622e-05,
+      "loss": 4.8143,
+      "step": 72192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88095885447957e-05,
+      "loss": 4.8265,
+      "step": 72704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.880120259728518e-05,
+      "loss": 4.8126,
+      "step": 73216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879281664977466e-05,
+      "loss": 4.7996,
+      "step": 73728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.87844634598716e-05,
+      "loss": 4.8145,
+      "step": 74240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.877607751236109e-05,
+      "loss": 4.8026,
+      "step": 74752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.876769156485057e-05,
+      "loss": 4.8028,
+      "step": 75264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875930561734005e-05,
+      "loss": 4.8086,
+      "step": 75776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875093604863326e-05,
+      "loss": 4.7898,
+      "step": 76288
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.756908416748047,
+      "eval_runtime": 298.6182,
+      "eval_samples_per_second": 1277.856,
+      "eval_steps_per_second": 39.934,
+      "step": 76320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.874255010112274e-05,
+      "loss": 4.7827,
+      "step": 76800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.873416415361222e-05,
+      "loss": 4.7867,
+      "step": 77312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.87257782061017e-05,
+      "loss": 4.788,
+      "step": 77824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8717408637394906e-05,
+      "loss": 4.7828,
+      "step": 78336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8709022689884386e-05,
+      "loss": 4.7875,
+      "step": 78848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8700636742373866e-05,
+      "loss": 4.7709,
+      "step": 79360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8692250794863346e-05,
+      "loss": 4.7771,
+      "step": 79872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8683881226156555e-05,
+      "loss": 4.7643,
+      "step": 80384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.867549527864604e-05,
+      "loss": 4.7734,
+      "step": 80896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.866710933113552e-05,
+      "loss": 4.7597,
+      "step": 81408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8658723383625e-05,
+      "loss": 4.7591,
+      "step": 81920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.865033743611448e-05,
+      "loss": 4.7594,
+      "step": 82432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.864195148860396e-05,
+      "loss": 4.7523,
+      "step": 82944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.863356554109344e-05,
+      "loss": 4.7478,
+      "step": 83456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.862517959358292e-05,
+      "loss": 4.7445,
+      "step": 83968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.86167936460724e-05,
+      "loss": 4.7309,
+      "step": 84480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8608407698561874e-05,
+      "loss": 4.7425,
+      "step": 84992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8600021751051354e-05,
+      "loss": 4.7357,
+      "step": 85504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8591635803540834e-05,
+      "loss": 4.728,
+      "step": 86016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8583249856030314e-05,
+      "loss": 4.7474,
+      "step": 86528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8574863908519794e-05,
+      "loss": 4.7299,
+      "step": 87040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8566477961009274e-05,
+      "loss": 4.7319,
+      "step": 87552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.855810839230249e-05,
+      "loss": 4.7253,
+      "step": 88064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.854972244479197e-05,
+      "loss": 4.7435,
+      "step": 88576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.854133649728145e-05,
+      "loss": 4.7118,
+      "step": 89088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.853295054977093e-05,
+      "loss": 4.7135,
+      "step": 89600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.852456460226041e-05,
+      "loss": 4.7168,
+      "step": 90112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.851617865474989e-05,
+      "loss": 4.7186,
+      "step": 90624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.850779270723937e-05,
+      "loss": 4.703,
+      "step": 91136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.849942313853258e-05,
+      "loss": 4.7032,
+      "step": 91648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.849105356982579e-05,
+      "loss": 4.7127,
+      "step": 92160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.848266762231527e-05,
+      "loss": 4.7096,
+      "step": 92672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.847428167480475e-05,
+      "loss": 4.7108,
+      "step": 93184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8465912106097964e-05,
+      "loss": 4.7054,
+      "step": 93696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8457526158587444e-05,
+      "loss": 4.6966,
+      "step": 94208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8449140211076924e-05,
+      "loss": 4.7052,
+      "step": 94720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8440754263566404e-05,
+      "loss": 4.6758,
+      "step": 95232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8432368316055884e-05,
+      "loss": 4.6963,
+      "step": 95744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8423982368545363e-05,
+      "loss": 4.6715,
+      "step": 96256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8415596421034843e-05,
+      "loss": 4.6844,
+      "step": 96768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.840721047352432e-05,
+      "loss": 4.6757,
+      "step": 97280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.83988245260138e-05,
+      "loss": 4.6873,
+      "step": 97792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.839043857850328e-05,
+      "loss": 4.6754,
+      "step": 98304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.838205263099276e-05,
+      "loss": 4.6738,
+      "step": 98816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8373666683482236e-05,
+      "loss": 4.676,
+      "step": 99328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8365280735971716e-05,
+      "loss": 4.6671,
+      "step": 99840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.835691116726493e-05,
+      "loss": 4.682,
+      "step": 100352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834852521975441e-05,
+      "loss": 4.6564,
+      "step": 100864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834013927224389e-05,
+      "loss": 4.6505,
+      "step": 101376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.833175332473337e-05,
+      "loss": 4.67,
+      "step": 101888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832336737722285e-05,
+      "loss": 4.6691,
+      "step": 102400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831498142971233e-05,
+      "loss": 4.6536,
+      "step": 102912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.830659548220181e-05,
+      "loss": 4.6484,
+      "step": 103424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.829822591349502e-05,
+      "loss": 4.6545,
+      "step": 103936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.82898399659845e-05,
+      "loss": 4.6417,
+      "step": 104448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.828145401847398e-05,
+      "loss": 4.6647,
+      "step": 104960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.827306807096346e-05,
+      "loss": 4.6385,
+      "step": 105472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.826468212345294e-05,
+      "loss": 4.646,
+      "step": 105984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.825629617594242e-05,
+      "loss": 4.6395,
+      "step": 106496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.82479102284319e-05,
+      "loss": 4.6333,
+      "step": 107008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.823952428092138e-05,
+      "loss": 4.6439,
+      "step": 107520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.823113833341086e-05,
+      "loss": 4.6299,
+      "step": 108032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.822275238590034e-05,
+      "loss": 4.6259,
+      "step": 108544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8214382817193557e-05,
+      "loss": 4.6282,
+      "step": 109056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8205996869683037e-05,
+      "loss": 4.6424,
+      "step": 109568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8197627300976246e-05,
+      "loss": 4.6241,
+      "step": 110080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8189241353465726e-05,
+      "loss": 4.6216,
+      "step": 110592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8180855405955206e-05,
+      "loss": 4.6148,
+      "step": 111104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8172469458444685e-05,
+      "loss": 4.6161,
+      "step": 111616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8164083510934165e-05,
+      "loss": 4.6353,
+      "step": 112128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8155697563423645e-05,
+      "loss": 4.632,
+      "step": 112640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8147327994716854e-05,
+      "loss": 4.6147,
+      "step": 113152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8138958426010064e-05,
+      "loss": 4.627,
+      "step": 113664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.813057247849955e-05,
+      "loss": 4.6293,
+      "step": 114176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.812218653098903e-05,
+      "loss": 4.6164,
+      "step": 114688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.811380058347851e-05,
+      "loss": 4.617,
+      "step": 115200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.810541463596799e-05,
+      "loss": 4.6064,
+      "step": 115712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.809702868845747e-05,
+      "loss": 4.5952,
+      "step": 116224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.808864274094695e-05,
+      "loss": 4.5979,
+      "step": 116736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.808025679343642e-05,
+      "loss": 4.6045,
+      "step": 117248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.807188722472964e-05,
+      "loss": 4.6056,
+      "step": 117760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.806350127721912e-05,
+      "loss": 4.6232,
+      "step": 118272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.80551153297086e-05,
+      "loss": 4.6089,
+      "step": 118784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.804672938219807e-05,
+      "loss": 4.5892,
+      "step": 119296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.803834343468755e-05,
+      "loss": 4.5962,
+      "step": 119808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.802995748717703e-05,
+      "loss": 4.6017,
+      "step": 120320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.802157153966652e-05,
+      "loss": 4.5964,
+      "step": 120832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8013185592156e-05,
+      "loss": 4.5884,
+      "step": 121344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.800479964464548e-05,
+      "loss": 4.5873,
+      "step": 121856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.799643007593869e-05,
+      "loss": 4.5864,
+      "step": 122368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.798804412842817e-05,
+      "loss": 4.5927,
+      "step": 122880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.797965818091765e-05,
+      "loss": 4.5817,
+      "step": 123392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.797127223340713e-05,
+      "loss": 4.5869,
+      "step": 123904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.796288628589661e-05,
+      "loss": 4.5824,
+      "step": 124416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.795450033838609e-05,
+      "loss": 4.5824,
+      "step": 124928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.794611439087557e-05,
+      "loss": 4.5724,
+      "step": 125440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.793772844336505e-05,
+      "loss": 4.5773,
+      "step": 125952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.792934249585453e-05,
+      "loss": 4.5723,
+      "step": 126464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.792095654834401e-05,
+      "loss": 4.5749,
+      "step": 126976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.791257060083349e-05,
+      "loss": 4.5789,
+      "step": 127488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.790418465332297e-05,
+      "loss": 4.5645,
+      "step": 128000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.789581508461618e-05,
+      "loss": 4.5674,
+      "step": 128512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.788744551590939e-05,
+      "loss": 4.5644,
+      "step": 129024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.787905956839887e-05,
+      "loss": 4.5626,
+      "step": 129536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.787067362088835e-05,
+      "loss": 4.5564,
+      "step": 130048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.786228767337783e-05,
+      "loss": 4.5581,
+      "step": 130560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.785390172586731e-05,
+      "loss": 4.5614,
+      "step": 131072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.784553215716052e-05,
+      "loss": 4.5513,
+      "step": 131584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.783714620965e-05,
+      "loss": 4.5545,
+      "step": 132096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.782876026213948e-05,
+      "loss": 4.5494,
+      "step": 132608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.782037431462896e-05,
+      "loss": 4.5409,
+      "step": 133120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7811988367118434e-05,
+      "loss": 4.5633,
+      "step": 133632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.780361879841165e-05,
+      "loss": 4.5465,
+      "step": 134144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.779523285090114e-05,
+      "loss": 4.5533,
+      "step": 134656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7786863282194346e-05,
+      "loss": 4.5497,
+      "step": 135168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7778477334683826e-05,
+      "loss": 4.5589,
+      "step": 135680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7770091387173306e-05,
+      "loss": 4.5495,
+      "step": 136192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7761705439662786e-05,
+      "loss": 4.5353,
+      "step": 136704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.775331949215226e-05,
+      "loss": 4.5462,
+      "step": 137216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.774493354464174e-05,
+      "loss": 4.5517,
+      "step": 137728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.773654759713122e-05,
+      "loss": 4.5364,
+      "step": 138240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.77281616496207e-05,
+      "loss": 4.5401,
+      "step": 138752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.771979208091391e-05,
+      "loss": 4.5343,
+      "step": 139264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.771140613340339e-05,
+      "loss": 4.5426,
+      "step": 139776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7703020185892875e-05,
+      "loss": 4.5301,
+      "step": 140288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7694634238382355e-05,
+      "loss": 4.5373,
+      "step": 140800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7686264669675564e-05,
+      "loss": 4.5279,
+      "step": 141312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7677878722165044e-05,
+      "loss": 4.5341,
+      "step": 141824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7669492774654524e-05,
+      "loss": 4.5227,
+      "step": 142336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7661106827144004e-05,
+      "loss": 4.5453,
+      "step": 142848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7652720879633484e-05,
+      "loss": 4.5349,
+      "step": 143360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7644334932122964e-05,
+      "loss": 4.52,
+      "step": 143872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7635948984612444e-05,
+      "loss": 4.5239,
+      "step": 144384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.762757941590565e-05,
+      "loss": 4.5253,
+      "step": 144896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.761919346839513e-05,
+      "loss": 4.5294,
+      "step": 145408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.761080752088461e-05,
+      "loss": 4.5374,
+      "step": 145920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.760242157337409e-05,
+      "loss": 4.5186,
+      "step": 146432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.759403562586357e-05,
+      "loss": 4.5237,
+      "step": 146944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.758564967835306e-05,
+      "loss": 4.5207,
+      "step": 147456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.757728010964627e-05,
+      "loss": 4.5229,
+      "step": 147968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.756889416213575e-05,
+      "loss": 4.5139,
+      "step": 148480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.756050821462523e-05,
+      "loss": 4.5228,
+      "step": 148992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.755212226711471e-05,
+      "loss": 4.513,
+      "step": 149504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7543769077211646e-05,
+      "loss": 4.5074,
+      "step": 150016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7535383129701126e-05,
+      "loss": 4.5209,
+      "step": 150528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7526997182190606e-05,
+      "loss": 4.5128,
+      "step": 151040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7518611234680086e-05,
+      "loss": 4.5113,
+      "step": 151552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7510225287169566e-05,
+      "loss": 4.5144,
+      "step": 152064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7501839339659046e-05,
+      "loss": 4.5057,
+      "step": 152576
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.476215839385986,
+      "eval_runtime": 296.0227,
+      "eval_samples_per_second": 1289.06,
+      "eval_steps_per_second": 40.284,
+      "step": 152640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7493453392148526e-05,
+      "loss": 4.4964,
+      "step": 153088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.748506744463801e-05,
+      "loss": 4.4979,
+      "step": 153600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.747668149712749e-05,
+      "loss": 4.51,
+      "step": 154112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.746829554961697e-05,
+      "loss": 4.4999,
+      "step": 154624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7459909602106446e-05,
+      "loss": 4.51,
+      "step": 155136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7451523654595926e-05,
+      "loss": 4.4954,
+      "step": 155648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7443137707085406e-05,
+      "loss": 4.4992,
+      "step": 156160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7434751759574886e-05,
+      "loss": 4.4932,
+      "step": 156672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7426365812064366e-05,
+      "loss": 4.5048,
+      "step": 157184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7417996243357575e-05,
+      "loss": 4.4918,
+      "step": 157696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7409610295847055e-05,
+      "loss": 4.4915,
+      "step": 158208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7401224348336535e-05,
+      "loss": 4.4914,
+      "step": 158720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7392838400826015e-05,
+      "loss": 4.4894,
+      "step": 159232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7384452453315495e-05,
+      "loss": 4.4841,
+      "step": 159744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7376066505804975e-05,
+      "loss": 4.4854,
+      "step": 160256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.736768055829446e-05,
+      "loss": 4.4751,
+      "step": 160768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.735929461078394e-05,
+      "loss": 4.4805,
+      "step": 161280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.735090866327342e-05,
+      "loss": 4.4761,
+      "step": 161792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.73425227157629e-05,
+      "loss": 4.4729,
+      "step": 162304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.733413676825238e-05,
+      "loss": 4.4989,
+      "step": 162816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.732576719954559e-05,
+      "loss": 4.479,
+      "step": 163328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.731738125203507e-05,
+      "loss": 4.4748,
+      "step": 163840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.730899530452455e-05,
+      "loss": 4.4789,
+      "step": 164352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.730060935701403e-05,
+      "loss": 4.4933,
+      "step": 164864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.729223978830724e-05,
+      "loss": 4.4671,
+      "step": 165376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.728385384079672e-05,
+      "loss": 4.4709,
+      "step": 165888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.72754678932862e-05,
+      "loss": 4.4707,
+      "step": 166400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.726708194577568e-05,
+      "loss": 4.4775,
+      "step": 166912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.725869599826516e-05,
+      "loss": 4.4592,
+      "step": 167424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.725031005075464e-05,
+      "loss": 4.4656,
+      "step": 167936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.724192410324412e-05,
+      "loss": 4.4678,
+      "step": 168448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.72335381557336e-05,
+      "loss": 4.4745,
+      "step": 168960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.722515220822308e-05,
+      "loss": 4.4796,
+      "step": 169472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.721678263951629e-05,
+      "loss": 4.4692,
+      "step": 169984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.720839669200577e-05,
+      "loss": 4.4626,
+      "step": 170496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.720001074449525e-05,
+      "loss": 4.4718,
+      "step": 171008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.719162479698473e-05,
+      "loss": 4.4509,
+      "step": 171520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.718323884947421e-05,
+      "loss": 4.4637,
+      "step": 172032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.717485290196369e-05,
+      "loss": 4.4459,
+      "step": 172544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.71664833332569e-05,
+      "loss": 4.4526,
+      "step": 173056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7158097385746384e-05,
+      "loss": 4.4548,
+      "step": 173568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7149711438235864e-05,
+      "loss": 4.4638,
+      "step": 174080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7141325490725343e-05,
+      "loss": 4.4538,
+      "step": 174592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7132939543214823e-05,
+      "loss": 4.4517,
+      "step": 175104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7124553595704303e-05,
+      "loss": 4.4531,
+      "step": 175616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.711616764819378e-05,
+      "loss": 4.4486,
+      "step": 176128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.710778170068326e-05,
+      "loss": 4.4614,
+      "step": 176640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.70994285107802e-05,
+      "loss": 4.4412,
+      "step": 177152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.709104256326968e-05,
+      "loss": 4.431,
+      "step": 177664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.708265661575916e-05,
+      "loss": 4.4559,
+      "step": 178176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.707427066824864e-05,
+      "loss": 4.4506,
+      "step": 178688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.706588472073812e-05,
+      "loss": 4.4405,
+      "step": 179200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.70574987732276e-05,
+      "loss": 4.44,
+      "step": 179712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.704912920452082e-05,
+      "loss": 4.4428,
+      "step": 180224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.70407432570103e-05,
+      "loss": 4.4318,
+      "step": 180736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.703235730949978e-05,
+      "loss": 4.4518,
+      "step": 181248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.702397136198926e-05,
+      "loss": 4.432,
+      "step": 181760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.701558541447874e-05,
+      "loss": 4.4392,
+      "step": 182272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.700719946696822e-05,
+      "loss": 4.4324,
+      "step": 182784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.69988135194577e-05,
+      "loss": 4.4331,
+      "step": 183296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6990443950750906e-05,
+      "loss": 4.4335,
+      "step": 183808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6982058003240386e-05,
+      "loss": 4.4316,
+      "step": 184320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6973672055729866e-05,
+      "loss": 4.4238,
+      "step": 184832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6965302487023075e-05,
+      "loss": 4.4269,
+      "step": 185344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6956916539512555e-05,
+      "loss": 4.4403,
+      "step": 185856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6948530592002035e-05,
+      "loss": 4.4267,
+      "step": 186368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6940144644491515e-05,
+      "loss": 4.4232,
+      "step": 186880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6931758696980995e-05,
+      "loss": 4.4191,
+      "step": 187392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6923372749470475e-05,
+      "loss": 4.4201,
+      "step": 187904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6914986801959955e-05,
+      "loss": 4.4415,
+      "step": 188416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.690661723325317e-05,
+      "loss": 4.4354,
+      "step": 188928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6898231285742644e-05,
+      "loss": 4.4231,
+      "step": 189440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6889845338232124e-05,
+      "loss": 4.4328,
+      "step": 189952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6881459390721604e-05,
+      "loss": 4.4375,
+      "step": 190464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6873073443211084e-05,
+      "loss": 4.4212,
+      "step": 190976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6864687495700564e-05,
+      "loss": 4.4259,
+      "step": 191488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6856301548190044e-05,
+      "loss": 4.4203,
+      "step": 192000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.684793197948325e-05,
+      "loss": 4.4105,
+      "step": 192512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.683954603197274e-05,
+      "loss": 4.4167,
+      "step": 193024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.683116008446222e-05,
+      "loss": 4.4128,
+      "step": 193536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.68227741369517e-05,
+      "loss": 4.42,
+      "step": 194048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.681438818944118e-05,
+      "loss": 4.4327,
+      "step": 194560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.680600224193066e-05,
+      "loss": 4.429,
+      "step": 195072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.679761629442014e-05,
+      "loss": 4.4023,
+      "step": 195584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.678923034690962e-05,
+      "loss": 4.4198,
+      "step": 196096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.678086077820283e-05,
+      "loss": 4.416,
+      "step": 196608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.677247483069231e-05,
+      "loss": 4.4188,
+      "step": 197120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.676408888318179e-05,
+      "loss": 4.4118,
+      "step": 197632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6755719314475e-05,
+      "loss": 4.4044,
+      "step": 198144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.674733336696448e-05,
+      "loss": 4.4119,
+      "step": 198656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.673894741945396e-05,
+      "loss": 4.4094,
+      "step": 199168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.673056147194344e-05,
+      "loss": 4.4049,
+      "step": 199680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.672219190323665e-05,
+      "loss": 4.4104,
+      "step": 200192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.671380595572613e-05,
+      "loss": 4.4042,
+      "step": 200704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.670542000821561e-05,
+      "loss": 4.4095,
+      "step": 201216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.669703406070509e-05,
+      "loss": 4.3983,
+      "step": 201728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.668864811319457e-05,
+      "loss": 4.4009,
+      "step": 202240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.668026216568405e-05,
+      "loss": 4.3992,
+      "step": 202752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.667187621817353e-05,
+      "loss": 4.4075,
+      "step": 203264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.666350664946674e-05,
+      "loss": 4.4067,
+      "step": 203776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.665512070195622e-05,
+      "loss": 4.3931,
+      "step": 204288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.66467347544457e-05,
+      "loss": 4.3971,
+      "step": 204800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6638348806935175e-05,
+      "loss": 4.3958,
+      "step": 205312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.662996285942466e-05,
+      "loss": 4.3972,
+      "step": 205824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.662157691191414e-05,
+      "loss": 4.3863,
+      "step": 206336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.661320734320736e-05,
+      "loss": 4.3885,
+      "step": 206848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.660482139569683e-05,
+      "loss": 4.3991,
+      "step": 207360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.659643544818631e-05,
+      "loss": 4.391,
+      "step": 207872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.658804950067579e-05,
+      "loss": 4.3873,
+      "step": 208384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657966355316527e-05,
+      "loss": 4.3856,
+      "step": 208896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657127760565475e-05,
+      "loss": 4.3794,
+      "step": 209408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.656289165814423e-05,
+      "loss": 4.3972,
+      "step": 209920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.655452208943744e-05,
+      "loss": 4.3897,
+      "step": 210432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.654615252073065e-05,
+      "loss": 4.3851,
+      "step": 210944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.653776657322013e-05,
+      "loss": 4.3887,
+      "step": 211456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6529380625709615e-05,
+      "loss": 4.4018,
+      "step": 211968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6520994678199095e-05,
+      "loss": 4.3901,
+      "step": 212480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6512608730688575e-05,
+      "loss": 4.3772,
+      "step": 212992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6504222783178055e-05,
+      "loss": 4.3875,
+      "step": 213504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6495836835667535e-05,
+      "loss": 4.3912,
+      "step": 214016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6487450888157015e-05,
+      "loss": 4.3814,
+      "step": 214528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6479081319450224e-05,
+      "loss": 4.3844,
+      "step": 215040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6470695371939704e-05,
+      "loss": 4.3795,
+      "step": 215552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.646232580323291e-05,
+      "loss": 4.3883,
+      "step": 216064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.645393985572239e-05,
+      "loss": 4.3748,
+      "step": 216576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.644555390821187e-05,
+      "loss": 4.3865,
+      "step": 217088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.643716796070135e-05,
+      "loss": 4.3691,
+      "step": 217600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642878201319083e-05,
+      "loss": 4.3849,
+      "step": 218112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642039606568031e-05,
+      "loss": 4.368,
+      "step": 218624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.64120101181698e-05,
+      "loss": 4.3928,
+      "step": 219136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.640362417065928e-05,
+      "loss": 4.3853,
+      "step": 219648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.639523822314876e-05,
+      "loss": 4.3749,
+      "step": 220160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.638686865444197e-05,
+      "loss": 4.3695,
+      "step": 220672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.637848270693145e-05,
+      "loss": 4.3771,
+      "step": 221184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.637009675942093e-05,
+      "loss": 4.3769,
+      "step": 221696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.636171081191041e-05,
+      "loss": 4.3875,
+      "step": 222208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.635332486439989e-05,
+      "loss": 4.3679,
+      "step": 222720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.634493891688936e-05,
+      "loss": 4.3778,
+      "step": 223232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.633655296937884e-05,
+      "loss": 4.3738,
+      "step": 223744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.632816702186832e-05,
+      "loss": 4.3809,
+      "step": 224256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.63197810743578e-05,
+      "loss": 4.3641,
+      "step": 224768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.631141150565102e-05,
+      "loss": 4.377,
+      "step": 225280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.630304193694423e-05,
+      "loss": 4.3699,
+      "step": 225792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.629465598943371e-05,
+      "loss": 4.3633,
+      "step": 226304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.628627004192319e-05,
+      "loss": 4.3757,
+      "step": 226816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6277884094412666e-05,
+      "loss": 4.3724,
+      "step": 227328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6269498146902146e-05,
+      "loss": 4.3624,
+      "step": 227840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6261112199391626e-05,
+      "loss": 4.374,
+      "step": 228352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6252726251881106e-05,
+      "loss": 4.3637,
+      "step": 228864
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.342019557952881,
+      "eval_runtime": 293.1681,
+      "eval_samples_per_second": 1301.612,
+      "eval_steps_per_second": 40.676,
+      "step": 228960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6244356683174315e-05,
+      "loss": 4.3607,
+      "step": 229376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6235970735663795e-05,
+      "loss": 4.3545,
+      "step": 229888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6227584788153275e-05,
+      "loss": 4.3654,
+      "step": 230400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6219198840642755e-05,
+      "loss": 4.3622,
+      "step": 230912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6210812893132235e-05,
+      "loss": 4.3764,
+      "step": 231424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6202426945621715e-05,
+      "loss": 4.3528,
+      "step": 231936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.61940409981112e-05,
+      "loss": 4.3607,
+      "step": 232448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.618565505060068e-05,
+      "loss": 4.3528,
+      "step": 232960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.617728548189389e-05,
+      "loss": 4.3715,
+      "step": 233472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.616889953438337e-05,
+      "loss": 4.3513,
+      "step": 233984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.616051358687285e-05,
+      "loss": 4.3529,
+      "step": 234496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.615212763936233e-05,
+      "loss": 4.3591,
+      "step": 235008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.614375807065554e-05,
+      "loss": 4.3498,
+      "step": 235520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.613537212314502e-05,
+      "loss": 4.3495,
+      "step": 236032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.61269861756345e-05,
+      "loss": 4.3491,
+      "step": 236544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.611860022812398e-05,
+      "loss": 4.3439,
+      "step": 237056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.611021428061346e-05,
+      "loss": 4.3416,
+      "step": 237568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.610182833310294e-05,
+      "loss": 4.3449,
+      "step": 238080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.609344238559242e-05,
+      "loss": 4.3396,
+      "step": 238592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.60850564380819e-05,
+      "loss": 4.3674,
+      "step": 239104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6076686869375116e-05,
+      "loss": 4.3454,
+      "step": 239616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6068300921864595e-05,
+      "loss": 4.345,
+      "step": 240128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6059931353157805e-05,
+      "loss": 4.3488,
+      "step": 240640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6051545405647285e-05,
+      "loss": 4.3596,
+      "step": 241152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6043159458136764e-05,
+      "loss": 4.3388,
+      "step": 241664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6034773510626244e-05,
+      "loss": 4.3421,
+      "step": 242176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6026403941919454e-05,
+      "loss": 4.3403,
+      "step": 242688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6018017994408933e-05,
+      "loss": 4.3454,
+      "step": 243200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6009632046898413e-05,
+      "loss": 4.3334,
+      "step": 243712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6001246099387893e-05,
+      "loss": 4.3387,
+      "step": 244224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.599286015187737e-05,
+      "loss": 4.3426,
+      "step": 244736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.598447420436685e-05,
+      "loss": 4.3464,
+      "step": 245248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.597610463566007e-05,
+      "loss": 4.3523,
+      "step": 245760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.596771868814955e-05,
+      "loss": 4.3442,
+      "step": 246272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.595933274063903e-05,
+      "loss": 4.3373,
+      "step": 246784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.59509467931285e-05,
+      "loss": 4.3432,
+      "step": 247296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.594256084561798e-05,
+      "loss": 4.3287,
+      "step": 247808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.593417489810746e-05,
+      "loss": 4.3363,
+      "step": 248320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.592578895059694e-05,
+      "loss": 4.3226,
+      "step": 248832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.591740300308642e-05,
+      "loss": 4.327,
+      "step": 249344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.590903343437963e-05,
+      "loss": 4.3311,
+      "step": 249856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.590064748686911e-05,
+      "loss": 4.3443,
+      "step": 250368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.589226153935859e-05,
+      "loss": 4.3356,
+      "step": 250880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.588387559184808e-05,
+      "loss": 4.3273,
+      "step": 251392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.587550602314129e-05,
+      "loss": 4.3315,
+      "step": 251904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5867136454434496e-05,
+      "loss": 4.3272,
+      "step": 252416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5858750506923976e-05,
+      "loss": 4.339,
+      "step": 252928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5850364559413456e-05,
+      "loss": 4.3211,
+      "step": 253440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5841978611902936e-05,
+      "loss": 4.3117,
+      "step": 253952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5833592664392416e-05,
+      "loss": 4.335,
+      "step": 254464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5825206716881896e-05,
+      "loss": 4.3308,
+      "step": 254976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5816837148175105e-05,
+      "loss": 4.317,
+      "step": 255488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5808451200664585e-05,
+      "loss": 4.3255,
+      "step": 256000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5800065253154065e-05,
+      "loss": 4.3269,
+      "step": 256512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5791679305643545e-05,
+      "loss": 4.3115,
+      "step": 257024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5783293358133025e-05,
+      "loss": 4.3325,
+      "step": 257536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.577490741062251e-05,
+      "loss": 4.3115,
+      "step": 258048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.576653784191572e-05,
+      "loss": 4.3266,
+      "step": 258560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.57581518944052e-05,
+      "loss": 4.3122,
+      "step": 259072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.574976594689468e-05,
+      "loss": 4.3237,
+      "step": 259584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.574137999938416e-05,
+      "loss": 4.312,
+      "step": 260096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.573299405187364e-05,
+      "loss": 4.3162,
+      "step": 260608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.572460810436312e-05,
+      "loss": 4.3046,
+      "step": 261120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.57162221568526e-05,
+      "loss": 4.3144,
+      "step": 261632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.570783620934208e-05,
+      "loss": 4.3297,
+      "step": 262144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.569946664063529e-05,
+      "loss": 4.3113,
+      "step": 262656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.569108069312477e-05,
+      "loss": 4.3122,
+      "step": 263168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.568269474561425e-05,
+      "loss": 4.3029,
+      "step": 263680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5674325176907465e-05,
+      "loss": 4.3089,
+      "step": 264192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5665955608200674e-05,
+      "loss": 4.3284,
+      "step": 264704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5657569660690154e-05,
+      "loss": 4.3221,
+      "step": 265216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5649183713179634e-05,
+      "loss": 4.3152,
+      "step": 265728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5640797765669114e-05,
+      "loss": 4.3222,
+      "step": 266240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.563242819696232e-05,
+      "loss": 4.3225,
+      "step": 266752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.56240422494518e-05,
+      "loss": 4.3118,
+      "step": 267264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.561565630194128e-05,
+      "loss": 4.3145,
+      "step": 267776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.560727035443076e-05,
+      "loss": 4.3123,
+      "step": 268288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.559888440692024e-05,
+      "loss": 4.3034,
+      "step": 268800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.559049845940972e-05,
+      "loss": 4.3051,
+      "step": 269312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.55821125118992e-05,
+      "loss": 4.3031,
+      "step": 269824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.557372656438868e-05,
+      "loss": 4.31,
+      "step": 270336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.55653569956819e-05,
+      "loss": 4.3226,
+      "step": 270848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.555697104817138e-05,
+      "loss": 4.3194,
+      "step": 271360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.554858510066086e-05,
+      "loss": 4.2941,
+      "step": 271872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.554019915315033e-05,
+      "loss": 4.3174,
+      "step": 272384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.553181320563981e-05,
+      "loss": 4.3088,
+      "step": 272896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.552342725812929e-05,
+      "loss": 4.3092,
+      "step": 273408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.551504131061877e-05,
+      "loss": 4.3038,
+      "step": 273920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.550667174191198e-05,
+      "loss": 4.2965,
+      "step": 274432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.549828579440146e-05,
+      "loss": 4.3069,
+      "step": 274944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.548989984689094e-05,
+      "loss": 4.3016,
+      "step": 275456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.548151389938042e-05,
+      "loss": 4.3005,
+      "step": 275968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.54731279518699e-05,
+      "loss": 4.3034,
+      "step": 276480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.546474200435939e-05,
+      "loss": 4.2988,
+      "step": 276992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5456372435652596e-05,
+      "loss": 4.3089,
+      "step": 277504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5447986488142076e-05,
+      "loss": 4.2961,
+      "step": 278016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5439600540631556e-05,
+      "loss": 4.2949,
+      "step": 278528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5431214593121036e-05,
+      "loss": 4.2935,
+      "step": 279040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5422828645610516e-05,
+      "loss": 4.3071,
+      "step": 279552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5414442698099996e-05,
+      "loss": 4.3,
+      "step": 280064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5406056750589476e-05,
+      "loss": 4.2911,
+      "step": 280576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5397670803078956e-05,
+      "loss": 4.2907,
+      "step": 281088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5389301234372165e-05,
+      "loss": 4.2963,
+      "step": 281600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5380915286861645e-05,
+      "loss": 4.2986,
+      "step": 282112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5372529339351125e-05,
+      "loss": 4.2881,
+      "step": 282624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.536415977064434e-05,
+      "loss": 4.2844,
+      "step": 283136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.535577382313382e-05,
+      "loss": 4.2982,
+      "step": 283648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.53473878756233e-05,
+      "loss": 4.2866,
+      "step": 284160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.533900192811278e-05,
+      "loss": 4.2877,
+      "step": 284672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.533061598060226e-05,
+      "loss": 4.2837,
+      "step": 285184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.532223003309174e-05,
+      "loss": 4.2853,
+      "step": 285696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.531384408558122e-05,
+      "loss": 4.2975,
+      "step": 286208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5305458138070694e-05,
+      "loss": 4.2845,
+      "step": 286720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5297072190560174e-05,
+      "loss": 4.2862,
+      "step": 287232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.528871900065712e-05,
+      "loss": 4.2921,
+      "step": 287744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.52803330531466e-05,
+      "loss": 4.2997,
+      "step": 288256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.527194710563608e-05,
+      "loss": 4.2982,
+      "step": 288768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.526356115812556e-05,
+      "loss": 4.2771,
+      "step": 289280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.525517521061504e-05,
+      "loss": 4.2894,
+      "step": 289792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.524678926310452e-05,
+      "loss": 4.2892,
+      "step": 290304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5238403315594e-05,
+      "loss": 4.2877,
+      "step": 290816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5230033746887215e-05,
+      "loss": 4.2908,
+      "step": 291328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5221647799376694e-05,
+      "loss": 4.2796,
+      "step": 291840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.521326185186617e-05,
+      "loss": 4.2908,
+      "step": 292352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.520487590435565e-05,
+      "loss": 4.2777,
+      "step": 292864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.519648995684513e-05,
+      "loss": 4.2888,
+      "step": 293376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.518810400933461e-05,
+      "loss": 4.2738,
+      "step": 293888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5179734440627817e-05,
+      "loss": 4.2902,
+      "step": 294400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5171348493117297e-05,
+      "loss": 4.2718,
+      "step": 294912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5162962545606777e-05,
+      "loss": 4.3017,
+      "step": 295424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5154576598096256e-05,
+      "loss": 4.2832,
+      "step": 295936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.514619065058574e-05,
+      "loss": 4.2839,
+      "step": 296448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.513780470307522e-05,
+      "loss": 4.2733,
+      "step": 296960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.51294187555647e-05,
+      "loss": 4.2856,
+      "step": 297472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.512103280805418e-05,
+      "loss": 4.2824,
+      "step": 297984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.511266323934739e-05,
+      "loss": 4.293,
+      "step": 298496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.510427729183687e-05,
+      "loss": 4.2757,
+      "step": 299008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.509589134432635e-05,
+      "loss": 4.284,
+      "step": 299520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.508752177561956e-05,
+      "loss": 4.2806,
+      "step": 300032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.507913582810904e-05,
+      "loss": 4.2924,
+      "step": 300544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.507074988059852e-05,
+      "loss": 4.2686,
+      "step": 301056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5062363933088e-05,
+      "loss": 4.2843,
+      "step": 301568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.505399436438121e-05,
+      "loss": 4.2766,
+      "step": 302080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.50456084168707e-05,
+      "loss": 4.2751,
+      "step": 302592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.503722246936018e-05,
+      "loss": 4.2836,
+      "step": 303104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.502883652184966e-05,
+      "loss": 4.2806,
+      "step": 303616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.502045057433914e-05,
+      "loss": 4.2721,
+      "step": 304128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.501206462682862e-05,
+      "loss": 4.2812,
+      "step": 304640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.50036786793181e-05,
+      "loss": 4.2767,
+      "step": 305152
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.258913040161133,
+      "eval_runtime": 293.9356,
+      "eval_samples_per_second": 1298.213,
+      "eval_steps_per_second": 40.57,
+      "step": 305280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4995309110611306e-05,
+      "loss": 4.271,
+      "step": 305664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4986923163100786e-05,
+      "loss": 4.261,
+      "step": 306176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4978537215590266e-05,
+      "loss": 4.276,
+      "step": 306688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4970151268079746e-05,
+      "loss": 4.2723,
+      "step": 307200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4961765320569226e-05,
+      "loss": 4.2868,
+      "step": 307712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4953379373058706e-05,
+      "loss": 4.2678,
+      "step": 308224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.494499342554818e-05,
+      "loss": 4.2673,
+      "step": 308736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4936607478037665e-05,
+      "loss": 4.2653,
+      "step": 309248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4928221530527145e-05,
+      "loss": 4.2806,
+      "step": 309760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4919835583016625e-05,
+      "loss": 4.2678,
+      "step": 310272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4911449635506105e-05,
+      "loss": 4.264,
+      "step": 310784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4903063687995585e-05,
+      "loss": 4.2678,
+      "step": 311296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4894694119288794e-05,
+      "loss": 4.2617,
+      "step": 311808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4886308171778274e-05,
+      "loss": 4.2644,
+      "step": 312320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4877922224267754e-05,
+      "loss": 4.2674,
+      "step": 312832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4869536276757234e-05,
+      "loss": 4.2494,
+      "step": 313344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4861150329246714e-05,
+      "loss": 4.2585,
+      "step": 313856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4852764381736194e-05,
+      "loss": 4.2578,
+      "step": 314368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4844378434225674e-05,
+      "loss": 4.2528,
+      "step": 314880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4835992486715154e-05,
+      "loss": 4.2768,
+      "step": 315392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4827606539204634e-05,
+      "loss": 4.2656,
+      "step": 315904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.481923697049785e-05,
+      "loss": 4.2578,
+      "step": 316416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.481085102298733e-05,
+      "loss": 4.2633,
+      "step": 316928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.480246507547681e-05,
+      "loss": 4.2751,
+      "step": 317440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.479407912796629e-05,
+      "loss": 4.254,
+      "step": 317952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.478569318045577e-05,
+      "loss": 4.2588,
+      "step": 318464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.477730723294524e-05,
+      "loss": 4.2559,
+      "step": 318976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.476892128543472e-05,
+      "loss": 4.258,
+      "step": 319488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.47605353379242e-05,
+      "loss": 4.2497,
+      "step": 320000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.475218214802115e-05,
+      "loss": 4.2589,
+      "step": 320512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.474379620051063e-05,
+      "loss": 4.2591,
+      "step": 321024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.473541025300011e-05,
+      "loss": 4.2629,
+      "step": 321536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.472702430548959e-05,
+      "loss": 4.2691,
+      "step": 322048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.471863835797907e-05,
+      "loss": 4.2605,
+      "step": 322560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.471025241046855e-05,
+      "loss": 4.2527,
+      "step": 323072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.470186646295803e-05,
+      "loss": 4.2598,
+      "step": 323584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.469348051544751e-05,
+      "loss": 4.2512,
+      "step": 324096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4685110946740717e-05,
+      "loss": 4.2521,
+      "step": 324608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4676724999230197e-05,
+      "loss": 4.2426,
+      "step": 325120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4668339051719676e-05,
+      "loss": 4.2464,
+      "step": 325632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4659953104209156e-05,
+      "loss": 4.2502,
+      "step": 326144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4651583535502366e-05,
+      "loss": 4.2612,
+      "step": 326656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4643197587991846e-05,
+      "loss": 4.2516,
+      "step": 327168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4634811640481325e-05,
+      "loss": 4.2485,
+      "step": 327680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4626425692970805e-05,
+      "loss": 4.2508,
+      "step": 328192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4618039745460285e-05,
+      "loss": 4.248,
+      "step": 328704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.46096701767535e-05,
+      "loss": 4.2582,
+      "step": 329216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.460128422924298e-05,
+      "loss": 4.242,
+      "step": 329728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.459289828173246e-05,
+      "loss": 4.2303,
+      "step": 330240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.458451233422194e-05,
+      "loss": 4.2539,
+      "step": 330752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.457614276551515e-05,
+      "loss": 4.2521,
+      "step": 331264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.456775681800463e-05,
+      "loss": 4.237,
+      "step": 331776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455937087049411e-05,
+      "loss": 4.2447,
+      "step": 332288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455098492298359e-05,
+      "loss": 4.249,
+      "step": 332800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.454259897547307e-05,
+      "loss": 4.2321,
+      "step": 333312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.453422940676628e-05,
+      "loss": 4.2534,
+      "step": 333824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.452584345925576e-05,
+      "loss": 4.2351,
+      "step": 334336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.451745751174524e-05,
+      "loss": 4.2466,
+      "step": 334848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.450907156423472e-05,
+      "loss": 4.2353,
+      "step": 335360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4500685616724206e-05,
+      "loss": 4.2441,
+      "step": 335872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4492316048017415e-05,
+      "loss": 4.2367,
+      "step": 336384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4483930100506895e-05,
+      "loss": 4.242,
+      "step": 336896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4475544152996375e-05,
+      "loss": 4.2262,
+      "step": 337408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4467158205485855e-05,
+      "loss": 4.2404,
+      "step": 337920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4458772257975335e-05,
+      "loss": 4.2489,
+      "step": 338432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4450386310464815e-05,
+      "loss": 4.2369,
+      "step": 338944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4442000362954295e-05,
+      "loss": 4.2325,
+      "step": 339456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4433614415443775e-05,
+      "loss": 4.2262,
+      "step": 339968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4425228467933254e-05,
+      "loss": 4.2318,
+      "step": 340480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.441684252042273e-05,
+      "loss": 4.2483,
+      "step": 340992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.440845657291221e-05,
+      "loss": 4.2471,
+      "step": 341504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.440007062540169e-05,
+      "loss": 4.2405,
+      "step": 342016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4391684677891174e-05,
+      "loss": 4.2463,
+      "step": 342528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.438331510918438e-05,
+      "loss": 4.2485,
+      "step": 343040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.43749455404776e-05,
+      "loss": 4.2336,
+      "step": 343552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.436655959296708e-05,
+      "loss": 4.2356,
+      "step": 344064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.435819002426029e-05,
+      "loss": 4.2407,
+      "step": 344576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.434980407674977e-05,
+      "loss": 4.2285,
+      "step": 345088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.434141812923925e-05,
+      "loss": 4.2274,
+      "step": 345600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.433303218172873e-05,
+      "loss": 4.231,
+      "step": 346112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.43246462342182e-05,
+      "loss": 4.2327,
+      "step": 346624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.431626028670768e-05,
+      "loss": 4.2489,
+      "step": 347136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.430787433919716e-05,
+      "loss": 4.2474,
+      "step": 347648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.429948839168664e-05,
+      "loss": 4.2216,
+      "step": 348160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.429110244417613e-05,
+      "loss": 4.2394,
+      "step": 348672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.428271649666561e-05,
+      "loss": 4.2358,
+      "step": 349184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.427433054915509e-05,
+      "loss": 4.2357,
+      "step": 349696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.426594460164457e-05,
+      "loss": 4.2323,
+      "step": 350208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.425757503293778e-05,
+      "loss": 4.2224,
+      "step": 350720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.424918908542726e-05,
+      "loss": 4.2309,
+      "step": 351232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.424080313791674e-05,
+      "loss": 4.2294,
+      "step": 351744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.423241719040622e-05,
+      "loss": 4.2283,
+      "step": 352256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.42240312428957e-05,
+      "loss": 4.2266,
+      "step": 352768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4215661674188906e-05,
+      "loss": 4.231,
+      "step": 353280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4207275726678386e-05,
+      "loss": 4.233,
+      "step": 353792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4198889779167866e-05,
+      "loss": 4.2309,
+      "step": 354304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4190503831657346e-05,
+      "loss": 4.2195,
+      "step": 354816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4182117884146826e-05,
+      "loss": 4.2255,
+      "step": 355328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.417374831544004e-05,
+      "loss": 4.231,
+      "step": 355840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.416536236792952e-05,
+      "loss": 4.2326,
+      "step": 356352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4156976420419e-05,
+      "loss": 4.2192,
+      "step": 356864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.414859047290848e-05,
+      "loss": 4.2226,
+      "step": 357376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.414020452539796e-05,
+      "loss": 4.2267,
+      "step": 357888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.413183495669117e-05,
+      "loss": 4.2262,
+      "step": 358400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.412346538798438e-05,
+      "loss": 4.2196,
+      "step": 358912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.411507944047386e-05,
+      "loss": 4.2157,
+      "step": 359424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.410669349296334e-05,
+      "loss": 4.2235,
+      "step": 359936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.409830754545282e-05,
+      "loss": 4.2176,
+      "step": 360448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.40899215979423e-05,
+      "loss": 4.2161,
+      "step": 360960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.408153565043178e-05,
+      "loss": 4.2146,
+      "step": 361472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.407314970292126e-05,
+      "loss": 4.2151,
+      "step": 361984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.406476375541074e-05,
+      "loss": 4.2273,
+      "step": 362496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4056394186703955e-05,
+      "loss": 4.2179,
+      "step": 363008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4048008239193435e-05,
+      "loss": 4.2144,
+      "step": 363520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4039638670486644e-05,
+      "loss": 4.2235,
+      "step": 364032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4031252722976124e-05,
+      "loss": 4.2325,
+      "step": 364544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4022866775465604e-05,
+      "loss": 4.229,
+      "step": 365056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4014480827955084e-05,
+      "loss": 4.2089,
+      "step": 365568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.400609488044456e-05,
+      "loss": 4.2173,
+      "step": 366080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.399770893293404e-05,
+      "loss": 4.2214,
+      "step": 366592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.398932298542352e-05,
+      "loss": 4.2164,
+      "step": 367104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.398095341671673e-05,
+      "loss": 4.2243,
+      "step": 367616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.397256746920621e-05,
+      "loss": 4.211,
+      "step": 368128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.396418152169569e-05,
+      "loss": 4.2196,
+      "step": 368640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.395579557418517e-05,
+      "loss": 4.2168,
+      "step": 369152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.394740962667465e-05,
+      "loss": 4.2198,
+      "step": 369664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.393902367916413e-05,
+      "loss": 4.2063,
+      "step": 370176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.393065411045734e-05,
+      "loss": 4.2193,
+      "step": 370688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.392226816294682e-05,
+      "loss": 4.2123,
+      "step": 371200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.39138822154363e-05,
+      "loss": 4.2262,
+      "step": 371712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.390549626792578e-05,
+      "loss": 4.2143,
+      "step": 372224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.389711032041526e-05,
+      "loss": 4.221,
+      "step": 372736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.388872437290474e-05,
+      "loss": 4.206,
+      "step": 373248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.388035480419795e-05,
+      "loss": 4.2219,
+      "step": 373760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.387196885668744e-05,
+      "loss": 4.2172,
+      "step": 374272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.386358290917692e-05,
+      "loss": 4.2235,
+      "step": 374784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.38551969616664e-05,
+      "loss": 4.2068,
+      "step": 375296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.384681101415588e-05,
+      "loss": 4.2227,
+      "step": 375808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.383842506664536e-05,
+      "loss": 4.211,
+      "step": 376320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.383003911913484e-05,
+      "loss": 4.2268,
+      "step": 376832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3821669550428046e-05,
+      "loss": 4.2077,
+      "step": 377344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3813283602917526e-05,
+      "loss": 4.2185,
+      "step": 377856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3804897655407006e-05,
+      "loss": 4.2112,
+      "step": 378368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3796511707896486e-05,
+      "loss": 4.2068,
+      "step": 378880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3788125760385966e-05,
+      "loss": 4.2182,
+      "step": 379392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3779739812875446e-05,
+      "loss": 4.2147,
+      "step": 379904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.377135386536492e-05,
+      "loss": 4.2068,
+      "step": 380416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3762984296658135e-05,
+      "loss": 4.2136,
+      "step": 380928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.375459834914762e-05,
+      "loss": 4.215,
+      "step": 381440
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.2025299072265625,
+      "eval_runtime": 295.0207,
+      "eval_samples_per_second": 1293.438,
+      "eval_steps_per_second": 40.421,
+      "step": 381600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.37462124016371e-05,
+      "loss": 4.2057,
+      "step": 381952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3737826454126575e-05,
+      "loss": 4.199,
+      "step": 382464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3729440506616055e-05,
+      "loss": 4.2105,
+      "step": 382976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3721054559105535e-05,
+      "loss": 4.2082,
+      "step": 383488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3712668611595015e-05,
+      "loss": 4.2192,
+      "step": 384000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3704282664084495e-05,
+      "loss": 4.2062,
+      "step": 384512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3695896716573975e-05,
+      "loss": 4.203,
+      "step": 385024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3687510769063455e-05,
+      "loss": 4.2031,
+      "step": 385536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3679124821552935e-05,
+      "loss": 4.2163,
+      "step": 386048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3670738874042415e-05,
+      "loss": 4.2032,
+      "step": 386560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3662352926531895e-05,
+      "loss": 4.2038,
+      "step": 387072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3653966979021375e-05,
+      "loss": 4.2022,
+      "step": 387584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3645597410314584e-05,
+      "loss": 4.2015,
+      "step": 388096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.363721146280407e-05,
+      "loss": 4.2026,
+      "step": 388608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.362882551529355e-05,
+      "loss": 4.2039,
+      "step": 389120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.362043956778303e-05,
+      "loss": 4.1886,
+      "step": 389632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.361205362027251e-05,
+      "loss": 4.1932,
+      "step": 390144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.360366767276199e-05,
+      "loss": 4.198,
+      "step": 390656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3595281725251463e-05,
+      "loss": 4.1887,
+      "step": 391168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.358689577774094e-05,
+      "loss": 4.2128,
+      "step": 391680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.357850983023042e-05,
+      "loss": 4.2054,
+      "step": 392192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.35701238827199e-05,
+      "loss": 4.1999,
+      "step": 392704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.356173793520938e-05,
+      "loss": 4.1956,
+      "step": 393216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.355335198769886e-05,
+      "loss": 4.2112,
+      "step": 393728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.354498241899207e-05,
+      "loss": 4.1965,
+      "step": 394240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.353659647148155e-05,
+      "loss": 4.1988,
+      "step": 394752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.352821052397104e-05,
+      "loss": 4.1903,
+      "step": 395264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.351984095526425e-05,
+      "loss": 4.1968,
+      "step": 395776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.351145500775373e-05,
+      "loss": 4.1937,
+      "step": 396288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.350306906024321e-05,
+      "loss": 4.1933,
+      "step": 396800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.349468311273269e-05,
+      "loss": 4.1983,
+      "step": 397312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.348629716522217e-05,
+      "loss": 4.202,
+      "step": 397824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347792759651538e-05,
+      "loss": 4.205,
+      "step": 398336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.346954164900486e-05,
+      "loss": 4.2025,
+      "step": 398848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.346115570149434e-05,
+      "loss": 4.1906,
+      "step": 399360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.345276975398382e-05,
+      "loss": 4.2012,
+      "step": 399872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3444400185277026e-05,
+      "loss": 4.1934,
+      "step": 400384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3436014237766506e-05,
+      "loss": 4.1836,
+      "step": 400896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.342762829025599e-05,
+      "loss": 4.1882,
+      "step": 401408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.341924234274547e-05,
+      "loss": 4.1857,
+      "step": 401920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.341085639523495e-05,
+      "loss": 4.1894,
+      "step": 402432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.340247044772443e-05,
+      "loss": 4.2037,
+      "step": 402944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.339408450021391e-05,
+      "loss": 4.1889,
+      "step": 403456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.338569855270339e-05,
+      "loss": 4.1902,
+      "step": 403968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.33773289839966e-05,
+      "loss": 4.1968,
+      "step": 404480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.336894303648608e-05,
+      "loss": 4.1877,
+      "step": 404992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.336055708897556e-05,
+      "loss": 4.1941,
+      "step": 405504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.335217114146504e-05,
+      "loss": 4.1835,
+      "step": 406016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.334378519395452e-05,
+      "loss": 4.1707,
+      "step": 406528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.333541562524773e-05,
+      "loss": 4.1988,
+      "step": 407040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.332702967773721e-05,
+      "loss": 4.1894,
+      "step": 407552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.331864373022669e-05,
+      "loss": 4.1842,
+      "step": 408064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.331025778271618e-05,
+      "loss": 4.1843,
+      "step": 408576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3301888214009386e-05,
+      "loss": 4.1865,
+      "step": 409088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3293502266498866e-05,
+      "loss": 4.1725,
+      "step": 409600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3285132697792075e-05,
+      "loss": 4.1911,
+      "step": 410112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3276746750281555e-05,
+      "loss": 4.1819,
+      "step": 410624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3268360802771035e-05,
+      "loss": 4.1873,
+      "step": 411136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3259974855260515e-05,
+      "loss": 4.1806,
+      "step": 411648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3251588907749995e-05,
+      "loss": 4.1836,
+      "step": 412160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.324320296023947e-05,
+      "loss": 4.178,
+      "step": 412672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.323481701272895e-05,
+      "loss": 4.1869,
+      "step": 413184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.322643106521843e-05,
+      "loss": 4.1723,
+      "step": 413696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3218061496511644e-05,
+      "loss": 4.1811,
+      "step": 414208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3209675549001124e-05,
+      "loss": 4.1904,
+      "step": 414720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3201289601490604e-05,
+      "loss": 4.1814,
+      "step": 415232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3192903653980084e-05,
+      "loss": 4.1754,
+      "step": 415744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.31845340852733e-05,
+      "loss": 4.1732,
+      "step": 416256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.317614813776277e-05,
+      "loss": 4.1728,
+      "step": 416768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.316776219025225e-05,
+      "loss": 4.1883,
+      "step": 417280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.315937624274173e-05,
+      "loss": 4.192,
+      "step": 417792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.315100667403494e-05,
+      "loss": 4.1832,
+      "step": 418304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.314262072652442e-05,
+      "loss": 4.1898,
+      "step": 418816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.31342347790139e-05,
+      "loss": 4.1912,
+      "step": 419328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.312584883150338e-05,
+      "loss": 4.1848,
+      "step": 419840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.31174792627966e-05,
+      "loss": 4.1767,
+      "step": 420352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3109109694089814e-05,
+      "loss": 4.18,
+      "step": 420864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3100723746579294e-05,
+      "loss": 4.1777,
+      "step": 421376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3092337799068773e-05,
+      "loss": 4.1688,
+      "step": 421888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.308395185155825e-05,
+      "loss": 4.1797,
+      "step": 422400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.307556590404773e-05,
+      "loss": 4.1733,
+      "step": 422912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3067179956537207e-05,
+      "loss": 4.1901,
+      "step": 423424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3058794009026687e-05,
+      "loss": 4.1935,
+      "step": 423936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3050408061516166e-05,
+      "loss": 4.164,
+      "step": 424448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3042022114005646e-05,
+      "loss": 4.1863,
+      "step": 424960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3033636166495126e-05,
+      "loss": 4.1761,
+      "step": 425472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3025250218984606e-05,
+      "loss": 4.1803,
+      "step": 425984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3016880650277815e-05,
+      "loss": 4.1762,
+      "step": 426496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.30084947027673e-05,
+      "loss": 4.1686,
+      "step": 427008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.300010875525678e-05,
+      "loss": 4.177,
+      "step": 427520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.299172280774626e-05,
+      "loss": 4.1733,
+      "step": 428032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.298333686023574e-05,
+      "loss": 4.1743,
+      "step": 428544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.297495091272522e-05,
+      "loss": 4.1707,
+      "step": 429056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.29665649652147e-05,
+      "loss": 4.1784,
+      "step": 429568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.295817901770418e-05,
+      "loss": 4.1778,
+      "step": 430080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.294980944899739e-05,
+      "loss": 4.1754,
+      "step": 430592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.294142350148687e-05,
+      "loss": 4.1652,
+      "step": 431104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.293303755397635e-05,
+      "loss": 4.1721,
+      "step": 431616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.292465160646583e-05,
+      "loss": 4.1739,
+      "step": 432128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2916265658955304e-05,
+      "loss": 4.1802,
+      "step": 432640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.290789609024852e-05,
+      "loss": 4.1655,
+      "step": 433152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2899510142738e-05,
+      "loss": 4.1709,
+      "step": 433664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2891124195227487e-05,
+      "loss": 4.1672,
+      "step": 434176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.288273824771696e-05,
+      "loss": 4.175,
+      "step": 434688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.287435230020644e-05,
+      "loss": 4.1623,
+      "step": 435200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.286596635269592e-05,
+      "loss": 4.1667,
+      "step": 435712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.285759678398913e-05,
+      "loss": 4.1679,
+      "step": 436224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.284921083647861e-05,
+      "loss": 4.1665,
+      "step": 436736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.284082488896809e-05,
+      "loss": 4.1616,
+      "step": 437248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.283243894145757e-05,
+      "loss": 4.1667,
+      "step": 437760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.282406937275078e-05,
+      "loss": 4.1618,
+      "step": 438272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.281568342524026e-05,
+      "loss": 4.1686,
+      "step": 438784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.280729747772974e-05,
+      "loss": 4.1702,
+      "step": 439296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2798911530219224e-05,
+      "loss": 4.1582,
+      "step": 439808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2790541961512434e-05,
+      "loss": 4.1711,
+      "step": 440320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.278217239280565e-05,
+      "loss": 4.1806,
+      "step": 440832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.277378644529513e-05,
+      "loss": 4.1781,
+      "step": 441344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.276541687658834e-05,
+      "loss": 4.1569,
+      "step": 441856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.275703092907782e-05,
+      "loss": 4.165,
+      "step": 442368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.27486449815673e-05,
+      "loss": 4.1691,
+      "step": 442880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.274025903405678e-05,
+      "loss": 4.1625,
+      "step": 443392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.273188946534999e-05,
+      "loss": 4.1723,
+      "step": 443904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.272350351783947e-05,
+      "loss": 4.1584,
+      "step": 444416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.271511757032895e-05,
+      "loss": 4.1696,
+      "step": 444928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.270673162281843e-05,
+      "loss": 4.1629,
+      "step": 445440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.269834567530791e-05,
+      "loss": 4.1714,
+      "step": 445952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.268995972779739e-05,
+      "loss": 4.157,
+      "step": 446464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.268157378028687e-05,
+      "loss": 4.1625,
+      "step": 446976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.267318783277635e-05,
+      "loss": 4.166,
+      "step": 447488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.266480188526583e-05,
+      "loss": 4.1733,
+      "step": 448000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2656432316559036e-05,
+      "loss": 4.1664,
+      "step": 448512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2648046369048516e-05,
+      "loss": 4.1661,
+      "step": 449024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2639660421537996e-05,
+      "loss": 4.1561,
+      "step": 449536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2631274474027476e-05,
+      "loss": 4.1715,
+      "step": 450048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2622888526516956e-05,
+      "loss": 4.1653,
+      "step": 450560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2614502579006436e-05,
+      "loss": 4.1737,
+      "step": 451072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2606133010299645e-05,
+      "loss": 4.1601,
+      "step": 451584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.259774706278913e-05,
+      "loss": 4.1706,
+      "step": 452096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.258936111527861e-05,
+      "loss": 4.1599,
+      "step": 452608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.258097516776809e-05,
+      "loss": 4.1809,
+      "step": 453120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.257258922025757e-05,
+      "loss": 4.155,
+      "step": 453632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.256420327274705e-05,
+      "loss": 4.1712,
+      "step": 454144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.255581732523653e-05,
+      "loss": 4.1609,
+      "step": 454656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.254743137772601e-05,
+      "loss": 4.1597,
+      "step": 455168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.253906180901922e-05,
+      "loss": 4.1672,
+      "step": 455680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.25306758615087e-05,
+      "loss": 4.1612,
+      "step": 456192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.252228991399818e-05,
+      "loss": 4.1651,
+      "step": 456704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.251390396648766e-05,
+      "loss": 4.1585,
+      "step": 457216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.250553439778087e-05,
+      "loss": 4.1682,
+      "step": 457728
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.161040306091309,
+      "eval_runtime": 311.0389,
+      "eval_samples_per_second": 1226.827,
+      "eval_steps_per_second": 38.339,
+      "step": 457920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2497164829074085e-05,
+      "loss": 4.1598,
+      "step": 458240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2488778881563565e-05,
+      "loss": 4.149,
+      "step": 458752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2480392934053045e-05,
+      "loss": 4.1581,
+      "step": 459264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2472006986542525e-05,
+      "loss": 4.1583,
+      "step": 459776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2463621039032005e-05,
+      "loss": 4.1681,
+      "step": 460288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2455235091521485e-05,
+      "loss": 4.1609,
+      "step": 460800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2446849144010965e-05,
+      "loss": 4.1482,
+      "step": 461312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.243846319650044e-05,
+      "loss": 4.1589,
+      "step": 461824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2430093627793654e-05,
+      "loss": 4.1652,
+      "step": 462336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2421707680283134e-05,
+      "loss": 4.1563,
+      "step": 462848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2413321732772614e-05,
+      "loss": 4.1555,
+      "step": 463360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.240493578526209e-05,
+      "loss": 4.1531,
+      "step": 463872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.23965662165553e-05,
+      "loss": 4.1571,
+      "step": 464384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.238818026904478e-05,
+      "loss": 4.1519,
+      "step": 464896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.237979432153426e-05,
+      "loss": 4.158,
+      "step": 465408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.237140837402374e-05,
+      "loss": 4.141,
+      "step": 465920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.236302242651322e-05,
+      "loss": 4.1447,
+      "step": 466432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.23546364790027e-05,
+      "loss": 4.1485,
+      "step": 466944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.234626691029591e-05,
+      "loss": 4.1436,
+      "step": 467456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.233788096278539e-05,
+      "loss": 4.1638,
+      "step": 467968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.232949501527487e-05,
+      "loss": 4.1608,
+      "step": 468480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.232110906776435e-05,
+      "loss": 4.1528,
+      "step": 468992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.231272312025383e-05,
+      "loss": 4.1476,
+      "step": 469504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.230433717274331e-05,
+      "loss": 4.1603,
+      "step": 470016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.229596760403652e-05,
+      "loss": 4.1483,
+      "step": 470528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2287581656526e-05,
+      "loss": 4.153,
+      "step": 471040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.227919570901549e-05,
+      "loss": 4.1443,
+      "step": 471552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.227080976150497e-05,
+      "loss": 4.152,
+      "step": 472064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.226242381399445e-05,
+      "loss": 4.1491,
+      "step": 472576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.225403786648393e-05,
+      "loss": 4.1461,
+      "step": 473088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.224565191897341e-05,
+      "loss": 4.1508,
+      "step": 473600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.223726597146289e-05,
+      "loss": 4.1568,
+      "step": 474112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2228896402756096e-05,
+      "loss": 4.1548,
+      "step": 474624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2220510455245576e-05,
+      "loss": 4.1565,
+      "step": 475136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2212124507735056e-05,
+      "loss": 4.1439,
+      "step": 475648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2203738560224536e-05,
+      "loss": 4.1591,
+      "step": 476160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2195352612714016e-05,
+      "loss": 4.1477,
+      "step": 476672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2186966665203496e-05,
+      "loss": 4.1363,
+      "step": 477184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2178580717692976e-05,
+      "loss": 4.1425,
+      "step": 477696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2170211148986185e-05,
+      "loss": 4.1385,
+      "step": 478208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.216182520147567e-05,
+      "loss": 4.1454,
+      "step": 478720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.215343925396515e-05,
+      "loss": 4.1558,
+      "step": 479232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2145053306454625e-05,
+      "loss": 4.1422,
+      "step": 479744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2136667358944105e-05,
+      "loss": 4.1454,
+      "step": 480256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2128281411433585e-05,
+      "loss": 4.1511,
+      "step": 480768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.21199118427268e-05,
+      "loss": 4.1431,
+      "step": 481280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2111525895216274e-05,
+      "loss": 4.1505,
+      "step": 481792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2103139947705754e-05,
+      "loss": 4.1364,
+      "step": 482304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2094754000195234e-05,
+      "loss": 4.1269,
+      "step": 482816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2086368052684714e-05,
+      "loss": 4.1534,
+      "step": 483328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.207799848397792e-05,
+      "loss": 4.1481,
+      "step": 483840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.206961253646741e-05,
+      "loss": 4.1356,
+      "step": 484352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.206122658895689e-05,
+      "loss": 4.1461,
+      "step": 484864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.205284064144637e-05,
+      "loss": 4.1393,
+      "step": 485376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.204447107273958e-05,
+      "loss": 4.1298,
+      "step": 485888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.203608512522906e-05,
+      "loss": 4.1417,
+      "step": 486400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.202769917771854e-05,
+      "loss": 4.1399,
+      "step": 486912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.201931323020802e-05,
+      "loss": 4.1426,
+      "step": 487424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.20109272826975e-05,
+      "loss": 4.1356,
+      "step": 487936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.200254133518698e-05,
+      "loss": 4.1398,
+      "step": 488448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.199415538767646e-05,
+      "loss": 4.1344,
+      "step": 488960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.198578581896967e-05,
+      "loss": 4.1453,
+      "step": 489472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.197739987145915e-05,
+      "loss": 4.1279,
+      "step": 489984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.196901392394863e-05,
+      "loss": 4.1351,
+      "step": 490496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.196062797643811e-05,
+      "loss": 4.1451,
+      "step": 491008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.195224202892759e-05,
+      "loss": 4.1384,
+      "step": 491520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1943856081417074e-05,
+      "loss": 4.1335,
+      "step": 492032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1935470133906554e-05,
+      "loss": 4.13,
+      "step": 492544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1927084186396034e-05,
+      "loss": 4.1274,
+      "step": 493056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1918698238885514e-05,
+      "loss": 4.1402,
+      "step": 493568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.191032867017872e-05,
+      "loss": 4.1503,
+      "step": 494080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.19019427226682e-05,
+      "loss": 4.1402,
+      "step": 494592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.189357315396141e-05,
+      "loss": 4.1465,
+      "step": 495104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.188518720645089e-05,
+      "loss": 4.1453,
+      "step": 495616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.187680125894037e-05,
+      "loss": 4.1469,
+      "step": 496128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.186841531142985e-05,
+      "loss": 4.1373,
+      "step": 496640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.186002936391933e-05,
+      "loss": 4.1354,
+      "step": 497152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.185165979521254e-05,
+      "loss": 4.1359,
+      "step": 497664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.184327384770203e-05,
+      "loss": 4.122,
+      "step": 498176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.183488790019151e-05,
+      "loss": 4.1391,
+      "step": 498688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.182650195268099e-05,
+      "loss": 4.1328,
+      "step": 499200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.181811600517046e-05,
+      "loss": 4.1469,
+      "step": 499712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.180973005765994e-05,
+      "loss": 4.1477,
+      "step": 500224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.180136048895316e-05,
+      "loss": 4.1214,
+      "step": 500736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.179297454144264e-05,
+      "loss": 4.1483,
+      "step": 501248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.178458859393211e-05,
+      "loss": 4.1338,
+      "step": 501760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.177620264642159e-05,
+      "loss": 4.1379,
+      "step": 502272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.176781669891107e-05,
+      "loss": 4.134,
+      "step": 502784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.175943075140055e-05,
+      "loss": 4.1285,
+      "step": 503296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.175104480389003e-05,
+      "loss": 4.1358,
+      "step": 503808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.174265885637951e-05,
+      "loss": 4.1311,
+      "step": 504320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1734289287672726e-05,
+      "loss": 4.133,
+      "step": 504832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1725903340162206e-05,
+      "loss": 4.1311,
+      "step": 505344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1717533771455415e-05,
+      "loss": 4.1362,
+      "step": 505856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1709147823944895e-05,
+      "loss": 4.1406,
+      "step": 506368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1700761876434375e-05,
+      "loss": 4.1273,
+      "step": 506880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1692375928923855e-05,
+      "loss": 4.1278,
+      "step": 507392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1683989981413334e-05,
+      "loss": 4.13,
+      "step": 507904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1675604033902814e-05,
+      "loss": 4.1294,
+      "step": 508416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1667218086392294e-05,
+      "loss": 4.1419,
+      "step": 508928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1658832138881774e-05,
+      "loss": 4.1278,
+      "step": 509440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1650462570174983e-05,
+      "loss": 4.1278,
+      "step": 509952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.164207662266446e-05,
+      "loss": 4.1261,
+      "step": 510464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.163369067515395e-05,
+      "loss": 4.1379,
+      "step": 510976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.162530472764343e-05,
+      "loss": 4.1173,
+      "step": 511488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.161691878013291e-05,
+      "loss": 4.1253,
+      "step": 512000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.160854921142612e-05,
+      "loss": 4.1265,
+      "step": 512512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.16001632639156e-05,
+      "loss": 4.1267,
+      "step": 513024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.159177731640508e-05,
+      "loss": 4.12,
+      "step": 513536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.158339136889456e-05,
+      "loss": 4.1255,
+      "step": 514048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.157502180018777e-05,
+      "loss": 4.1195,
+      "step": 514560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.156663585267725e-05,
+      "loss": 4.1286,
+      "step": 515072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.155824990516673e-05,
+      "loss": 4.1293,
+      "step": 515584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.154986395765621e-05,
+      "loss": 4.1208,
+      "step": 516096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.154149438894942e-05,
+      "loss": 4.1266,
+      "step": 516608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1533108441438904e-05,
+      "loss": 4.1392,
+      "step": 517120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.152473887273211e-05,
+      "loss": 4.1414,
+      "step": 517632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.151635292522159e-05,
+      "loss": 4.1189,
+      "step": 518144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.150796697771107e-05,
+      "loss": 4.123,
+      "step": 518656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.149958103020055e-05,
+      "loss": 4.1304,
+      "step": 519168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.149119508269003e-05,
+      "loss": 4.1196,
+      "step": 519680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.148284189278697e-05,
+      "loss": 4.1358,
+      "step": 520192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.147445594527645e-05,
+      "loss": 4.1115,
+      "step": 520704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.146606999776593e-05,
+      "loss": 4.1339,
+      "step": 521216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.145768405025541e-05,
+      "loss": 4.1203,
+      "step": 521728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.144929810274489e-05,
+      "loss": 4.1327,
+      "step": 522240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.144091215523437e-05,
+      "loss": 4.1169,
+      "step": 522752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.143252620772386e-05,
+      "loss": 4.1237,
+      "step": 523264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.142414026021334e-05,
+      "loss": 4.1217,
+      "step": 523776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.141575431270282e-05,
+      "loss": 4.1342,
+      "step": 524288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.14073683651923e-05,
+      "loss": 4.1227,
+      "step": 524800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.139898241768177e-05,
+      "loss": 4.1295,
+      "step": 525312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.139059647017125e-05,
+      "loss": 4.1179,
+      "step": 525824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.138221052266073e-05,
+      "loss": 4.1347,
+      "step": 526336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1373840953953946e-05,
+      "loss": 4.1245,
+      "step": 526848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.136545500644342e-05,
+      "loss": 4.1303,
+      "step": 527360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1357085437736635e-05,
+      "loss": 4.1238,
+      "step": 527872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1348699490226115e-05,
+      "loss": 4.13,
+      "step": 528384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1340313542715595e-05,
+      "loss": 4.1222,
+      "step": 528896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1331927595205075e-05,
+      "loss": 4.1385,
+      "step": 529408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1323541647694555e-05,
+      "loss": 4.1172,
+      "step": 529920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1315155700184035e-05,
+      "loss": 4.1316,
+      "step": 530432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1306769752673515e-05,
+      "loss": 4.1223,
+      "step": 530944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1298383805162995e-05,
+      "loss": 4.1233,
+      "step": 531456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1289997857652475e-05,
+      "loss": 4.1218,
+      "step": 531968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1281628288945684e-05,
+      "loss": 4.1221,
+      "step": 532480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1273242341435164e-05,
+      "loss": 4.125,
+      "step": 532992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.126487277272837e-05,
+      "loss": 4.1218,
+      "step": 533504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.125648682521785e-05,
+      "loss": 4.1327,
+      "step": 534016
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.130150318145752,
+      "eval_runtime": 308.9439,
+      "eval_samples_per_second": 1235.147,
+      "eval_steps_per_second": 38.599,
+      "step": 534240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.124810087770733e-05,
+      "loss": 4.1213,
+      "step": 534528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.123971493019681e-05,
+      "loss": 4.1128,
+      "step": 535040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.123132898268629e-05,
+      "loss": 4.1199,
+      "step": 535552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.122294303517577e-05,
+      "loss": 4.1176,
+      "step": 536064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.121455708766526e-05,
+      "loss": 4.1285,
+      "step": 536576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.120617114015474e-05,
+      "loss": 4.125,
+      "step": 537088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.119778519264422e-05,
+      "loss": 4.118,
+      "step": 537600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.11893992451337e-05,
+      "loss": 4.1157,
+      "step": 538112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.118101329762318e-05,
+      "loss": 4.1276,
+      "step": 538624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.117262735011266e-05,
+      "loss": 4.1198,
+      "step": 539136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.116424140260213e-05,
+      "loss": 4.1164,
+      "step": 539648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.115585545509161e-05,
+      "loss": 4.1137,
+      "step": 540160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.114748588638483e-05,
+      "loss": 4.1199,
+      "step": 540672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.11390999388743e-05,
+      "loss": 4.1147,
+      "step": 541184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.113071399136378e-05,
+      "loss": 4.1185,
+      "step": 541696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.112232804385326e-05,
+      "loss": 4.1016,
+      "step": 542208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.111394209634274e-05,
+      "loss": 4.1092,
+      "step": 542720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.110555614883223e-05,
+      "loss": 4.1114,
+      "step": 543232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.109717020132171e-05,
+      "loss": 4.1065,
+      "step": 543744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.108878425381119e-05,
+      "loss": 4.1243,
+      "step": 544256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.108039830630067e-05,
+      "loss": 4.1241,
+      "step": 544768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.107202873759388e-05,
+      "loss": 4.1152,
+      "step": 545280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.106364279008336e-05,
+      "loss": 4.1094,
+      "step": 545792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.105525684257284e-05,
+      "loss": 4.1237,
+      "step": 546304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.104687089506232e-05,
+      "loss": 4.1122,
+      "step": 546816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.10384849475518e-05,
+      "loss": 4.115,
+      "step": 547328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.103009900004128e-05,
+      "loss": 4.1052,
+      "step": 547840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1021729431334486e-05,
+      "loss": 4.1164,
+      "step": 548352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1013343483823966e-05,
+      "loss": 4.1158,
+      "step": 548864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1004957536313446e-05,
+      "loss": 4.1041,
+      "step": 549376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0996571588802926e-05,
+      "loss": 4.112,
+      "step": 549888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0988185641292406e-05,
+      "loss": 4.1202,
+      "step": 550400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.097979969378189e-05,
+      "loss": 4.1142,
+      "step": 550912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.097141374627137e-05,
+      "loss": 4.1252,
+      "step": 551424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0963027798760846e-05,
+      "loss": 4.1086,
+      "step": 551936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0954641851250326e-05,
+      "loss": 4.1184,
+      "step": 552448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.094628866134727e-05,
+      "loss": 4.113,
+      "step": 552960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.093790271383675e-05,
+      "loss": 4.1014,
+      "step": 553472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.092951676632623e-05,
+      "loss": 4.1077,
+      "step": 553984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.092113081881571e-05,
+      "loss": 4.1017,
+      "step": 554496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.091274487130519e-05,
+      "loss": 4.1056,
+      "step": 555008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0904358923794664e-05,
+      "loss": 4.1193,
+      "step": 555520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.089597297628415e-05,
+      "loss": 4.1083,
+      "step": 556032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.088758702877363e-05,
+      "loss": 4.1076,
+      "step": 556544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.087920108126311e-05,
+      "loss": 4.1149,
+      "step": 557056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.087083151255632e-05,
+      "loss": 4.1077,
+      "step": 557568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.08624455650458e-05,
+      "loss": 4.109,
+      "step": 558080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0854075996339015e-05,
+      "loss": 4.1056,
+      "step": 558592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0845690048828495e-05,
+      "loss": 4.0837,
+      "step": 559104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.083730410131797e-05,
+      "loss": 4.1236,
+      "step": 559616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.082891815380745e-05,
+      "loss": 4.1059,
+      "step": 560128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.082053220629693e-05,
+      "loss": 4.1072,
+      "step": 560640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.081214625878641e-05,
+      "loss": 4.1053,
+      "step": 561152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.080376031127589e-05,
+      "loss": 4.1048,
+      "step": 561664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.079537436376537e-05,
+      "loss": 4.0954,
+      "step": 562176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0787004795058584e-05,
+      "loss": 4.1012,
+      "step": 562688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0778618847548064e-05,
+      "loss": 4.1054,
+      "step": 563200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0770232900037544e-05,
+      "loss": 4.1068,
+      "step": 563712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0761846952527024e-05,
+      "loss": 4.1084,
+      "step": 564224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0753461005016504e-05,
+      "loss": 4.0983,
+      "step": 564736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0745075057505984e-05,
+      "loss": 4.0992,
+      "step": 565248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0736689109995464e-05,
+      "loss": 4.1147,
+      "step": 565760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.072831954128867e-05,
+      "loss": 4.0881,
+      "step": 566272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.071993359377815e-05,
+      "loss": 4.099,
+      "step": 566784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.071154764626763e-05,
+      "loss": 4.1079,
+      "step": 567296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.070316169875711e-05,
+      "loss": 4.1042,
+      "step": 567808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.069477575124659e-05,
+      "loss": 4.0975,
+      "step": 568320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.068638980373607e-05,
+      "loss": 4.0996,
+      "step": 568832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.067800385622555e-05,
+      "loss": 4.0884,
+      "step": 569344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.066961790871503e-05,
+      "loss": 4.1038,
+      "step": 569856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.066123196120451e-05,
+      "loss": 4.1163,
+      "step": 570368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.065287877130146e-05,
+      "loss": 4.1061,
+      "step": 570880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.064449282379094e-05,
+      "loss": 4.1118,
+      "step": 571392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.063610687628042e-05,
+      "loss": 4.1116,
+      "step": 571904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.06277209287699e-05,
+      "loss": 4.1156,
+      "step": 572416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0619351360063106e-05,
+      "loss": 4.0988,
+      "step": 572928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0610965412552586e-05,
+      "loss": 4.1028,
+      "step": 573440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0602579465042066e-05,
+      "loss": 4.1025,
+      "step": 573952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0594193517531546e-05,
+      "loss": 4.0846,
+      "step": 574464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0585807570021026e-05,
+      "loss": 4.1028,
+      "step": 574976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0577421622510506e-05,
+      "loss": 4.099,
+      "step": 575488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0569035674999986e-05,
+      "loss": 4.1082,
+      "step": 576000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0560649727489466e-05,
+      "loss": 4.1172,
+      "step": 576512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.055228015878268e-05,
+      "loss": 4.0922,
+      "step": 577024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0543894211272155e-05,
+      "loss": 4.1076,
+      "step": 577536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0535508263761635e-05,
+      "loss": 4.1003,
+      "step": 578048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0527122316251115e-05,
+      "loss": 4.1049,
+      "step": 578560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0518736368740595e-05,
+      "loss": 4.0973,
+      "step": 579072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0510350421230075e-05,
+      "loss": 4.0975,
+      "step": 579584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0501964473719555e-05,
+      "loss": 4.0978,
+      "step": 580096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0493578526209035e-05,
+      "loss": 4.0974,
+      "step": 580608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0485208957502244e-05,
+      "loss": 4.0982,
+      "step": 581120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0476823009991724e-05,
+      "loss": 4.0952,
+      "step": 581632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0468437062481204e-05,
+      "loss": 4.1064,
+      "step": 582144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.046005111497069e-05,
+      "loss": 4.1046,
+      "step": 582656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.045166516746017e-05,
+      "loss": 4.0935,
+      "step": 583168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.044327921994965e-05,
+      "loss": 4.0916,
+      "step": 583680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.043489327243913e-05,
+      "loss": 4.0956,
+      "step": 584192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.042650732492861e-05,
+      "loss": 4.1011,
+      "step": 584704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.041812137741809e-05,
+      "loss": 4.1034,
+      "step": 585216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.04097518087113e-05,
+      "loss": 4.0976,
+      "step": 585728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.040136586120078e-05,
+      "loss": 4.0923,
+      "step": 586240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.039297991369026e-05,
+      "loss": 4.094,
+      "step": 586752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.038459396617974e-05,
+      "loss": 4.1091,
+      "step": 587264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.037620801866921e-05,
+      "loss": 4.0798,
+      "step": 587776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.036783844996243e-05,
+      "loss": 4.0944,
+      "step": 588288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.035945250245191e-05,
+      "loss": 4.0883,
+      "step": 588800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.035106655494139e-05,
+      "loss": 4.0964,
+      "step": 589312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.034268060743087e-05,
+      "loss": 4.0863,
+      "step": 589824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0334311038724084e-05,
+      "loss": 4.099,
+      "step": 590336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0325925091213564e-05,
+      "loss": 4.085,
+      "step": 590848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0317539143703044e-05,
+      "loss": 4.097,
+      "step": 591360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.030915319619252e-05,
+      "loss": 4.0978,
+      "step": 591872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0300767248682e-05,
+      "loss": 4.0865,
+      "step": 592384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.029239767997521e-05,
+      "loss": 4.096,
+      "step": 592896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0284011732464686e-05,
+      "loss": 4.1093,
+      "step": 593408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0275625784954166e-05,
+      "loss": 4.0983,
+      "step": 593920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0267239837443646e-05,
+      "loss": 4.0958,
+      "step": 594432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.025887026873686e-05,
+      "loss": 4.0822,
+      "step": 594944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.025048432122634e-05,
+      "loss": 4.0986,
+      "step": 595456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.024209837371582e-05,
+      "loss": 4.096,
+      "step": 595968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.023372880500904e-05,
+      "loss": 4.0984,
+      "step": 596480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.022534285749852e-05,
+      "loss": 4.0796,
+      "step": 596992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.021695690998799e-05,
+      "loss": 4.1014,
+      "step": 597504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.020857096247747e-05,
+      "loss": 4.09,
+      "step": 598016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.020018501496695e-05,
+      "loss": 4.0975,
+      "step": 598528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.019179906745643e-05,
+      "loss": 4.0861,
+      "step": 599040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.018341311994591e-05,
+      "loss": 4.0939,
+      "step": 599552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.017502717243539e-05,
+      "loss": 4.088,
+      "step": 600064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.01666576037286e-05,
+      "loss": 4.1037,
+      "step": 600576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.015827165621808e-05,
+      "loss": 4.0935,
+      "step": 601088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0149902087511296e-05,
+      "loss": 4.096,
+      "step": 601600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0141516140000776e-05,
+      "loss": 4.0834,
+      "step": 602112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0133130192490256e-05,
+      "loss": 4.1038,
+      "step": 602624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0124744244979736e-05,
+      "loss": 4.093,
+      "step": 603136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0116358297469216e-05,
+      "loss": 4.0984,
+      "step": 603648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0107972349958696e-05,
+      "loss": 4.0942,
+      "step": 604160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0099586402448175e-05,
+      "loss": 4.0984,
+      "step": 604672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0091216833741385e-05,
+      "loss": 4.0889,
+      "step": 605184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0082830886230865e-05,
+      "loss": 4.1062,
+      "step": 605696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0074444938720344e-05,
+      "loss": 4.0867,
+      "step": 606208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0066058991209824e-05,
+      "loss": 4.1008,
+      "step": 606720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0057673043699304e-05,
+      "loss": 4.0962,
+      "step": 607232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0049287096188784e-05,
+      "loss": 4.0896,
+      "step": 607744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0040901148678264e-05,
+      "loss": 4.0891,
+      "step": 608256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.003253157997148e-05,
+      "loss": 4.0915,
+      "step": 608768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.002414563246096e-05,
+      "loss": 4.0954,
+      "step": 609280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.001575968495044e-05,
+      "loss": 4.091,
+      "step": 609792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000737373743992e-05,
+      "loss": 4.0996,
+      "step": 610304
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.106700420379639,
+      "eval_runtime": 292.0792,
+      "eval_samples_per_second": 1306.464,
+      "eval_steps_per_second": 40.828,
+      "step": 610560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.99989877899294e-05,
+      "loss": 4.0795,
+      "step": 610816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.999060184241887e-05,
+      "loss": 4.0835,
+      "step": 611328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.998221589490835e-05,
+      "loss": 4.0858,
+      "step": 611840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.997382994739783e-05,
+      "loss": 4.0888,
+      "step": 612352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.996546037869105e-05,
+      "loss": 4.1004,
+      "step": 612864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.995707443118052e-05,
+      "loss": 4.0893,
+      "step": 613376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.994868848367e-05,
+      "loss": 4.0917,
+      "step": 613888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.994030253615948e-05,
+      "loss": 4.0835,
+      "step": 614400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.993191658864897e-05,
+      "loss": 4.0945,
+      "step": 614912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.992353064113845e-05,
+      "loss": 4.0925,
+      "step": 615424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.991514469362793e-05,
+      "loss": 4.0853,
+      "step": 615936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.990675874611741e-05,
+      "loss": 4.0806,
+      "step": 616448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.989838917741062e-05,
+      "loss": 4.0932,
+      "step": 616960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.98900032299001e-05,
+      "loss": 4.0829,
+      "step": 617472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.988161728238958e-05,
+      "loss": 4.0839,
+      "step": 617984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.987323133487906e-05,
+      "loss": 4.0764,
+      "step": 618496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.986486176617227e-05,
+      "loss": 4.0758,
+      "step": 619008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.985647581866175e-05,
+      "loss": 4.0829,
+      "step": 619520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.984808987115123e-05,
+      "loss": 4.0746,
+      "step": 620032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9839703923640707e-05,
+      "loss": 4.0936,
+      "step": 620544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.983133435493392e-05,
+      "loss": 4.0928,
+      "step": 621056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.98229484074234e-05,
+      "loss": 4.0899,
+      "step": 621568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.981456245991288e-05,
+      "loss": 4.0796,
+      "step": 622080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.980617651240236e-05,
+      "loss": 4.0868,
+      "step": 622592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.979780694369557e-05,
+      "loss": 4.0846,
+      "step": 623104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.978942099618505e-05,
+      "loss": 4.0832,
+      "step": 623616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.978103504867453e-05,
+      "loss": 4.0793,
+      "step": 624128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.977266547996774e-05,
+      "loss": 4.0866,
+      "step": 624640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.976427953245722e-05,
+      "loss": 4.0821,
+      "step": 625152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.97558935849467e-05,
+      "loss": 4.0766,
+      "step": 625664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.974750763743618e-05,
+      "loss": 4.0786,
+      "step": 626176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.973912168992566e-05,
+      "loss": 4.0912,
+      "step": 626688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.973075212121887e-05,
+      "loss": 4.0886,
+      "step": 627200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9722366173708356e-05,
+      "loss": 4.0925,
+      "step": 627712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9713980226197836e-05,
+      "loss": 4.078,
+      "step": 628224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9705594278687316e-05,
+      "loss": 4.0824,
+      "step": 628736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9697224709980525e-05,
+      "loss": 4.089,
+      "step": 629248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9688838762470005e-05,
+      "loss": 4.0731,
+      "step": 629760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9680452814959485e-05,
+      "loss": 4.0742,
+      "step": 630272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9672066867448965e-05,
+      "loss": 4.0723,
+      "step": 630784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9663680919938445e-05,
+      "loss": 4.0786,
+      "step": 631296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9655294972427925e-05,
+      "loss": 4.0809,
+      "step": 631808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9646909024917405e-05,
+      "loss": 4.0849,
+      "step": 632320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9638523077406885e-05,
+      "loss": 4.083,
+      "step": 632832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.963013712989636e-05,
+      "loss": 4.0774,
+      "step": 633344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9621767561189574e-05,
+      "loss": 4.0815,
+      "step": 633856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9613381613679054e-05,
+      "loss": 4.0757,
+      "step": 634368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.960499566616854e-05,
+      "loss": 4.0781,
+      "step": 634880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9596609718658014e-05,
+      "loss": 4.0514,
+      "step": 635392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.958824014995123e-05,
+      "loss": 4.0946,
+      "step": 635904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.957985420244071e-05,
+      "loss": 4.075,
+      "step": 636416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.957146825493018e-05,
+      "loss": 4.0847,
+      "step": 636928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.956308230741966e-05,
+      "loss": 4.0751,
+      "step": 637440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.955469635990914e-05,
+      "loss": 4.0753,
+      "step": 637952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.954632679120236e-05,
+      "loss": 4.0659,
+      "step": 638464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.953794084369183e-05,
+      "loss": 4.0703,
+      "step": 638976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.952955489618131e-05,
+      "loss": 4.0776,
+      "step": 639488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.952116894867079e-05,
+      "loss": 4.075,
+      "step": 640000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.951279937996401e-05,
+      "loss": 4.0809,
+      "step": 640512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.950441343245349e-05,
+      "loss": 4.0731,
+      "step": 641024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.949602748494297e-05,
+      "loss": 4.0664,
+      "step": 641536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.948765791623618e-05,
+      "loss": 4.0866,
+      "step": 642048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9479271968725656e-05,
+      "loss": 4.0575,
+      "step": 642560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9470886021215136e-05,
+      "loss": 4.0661,
+      "step": 643072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9462500073704616e-05,
+      "loss": 4.0819,
+      "step": 643584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9454114126194096e-05,
+      "loss": 4.0799,
+      "step": 644096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9445728178683576e-05,
+      "loss": 4.0613,
+      "step": 644608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9437342231173056e-05,
+      "loss": 4.0759,
+      "step": 645120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9428956283662536e-05,
+      "loss": 4.0553,
+      "step": 645632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9420586714955745e-05,
+      "loss": 4.078,
+      "step": 646144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.941221714624896e-05,
+      "loss": 4.0827,
+      "step": 646656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.940383119873844e-05,
+      "loss": 4.08,
+      "step": 647168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.939544525122792e-05,
+      "loss": 4.0772,
+      "step": 647680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.93870593037174e-05,
+      "loss": 4.0819,
+      "step": 648192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.937867335620688e-05,
+      "loss": 4.0905,
+      "step": 648704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.937030378750009e-05,
+      "loss": 4.0693,
+      "step": 649216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.936191783998957e-05,
+      "loss": 4.0762,
+      "step": 649728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.935354827128278e-05,
+      "loss": 4.0769,
+      "step": 650240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.934516232377226e-05,
+      "loss": 4.0509,
+      "step": 650752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.933677637626174e-05,
+      "loss": 4.0753,
+      "step": 651264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.932839042875122e-05,
+      "loss": 4.0732,
+      "step": 651776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.93200044812407e-05,
+      "loss": 4.0775,
+      "step": 652288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9311618533730186e-05,
+      "loss": 4.0936,
+      "step": 652800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9303232586219666e-05,
+      "loss": 4.0622,
+      "step": 653312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9294846638709146e-05,
+      "loss": 4.0738,
+      "step": 653824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9286460691198626e-05,
+      "loss": 4.0716,
+      "step": 654336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9278074743688105e-05,
+      "loss": 4.0768,
+      "step": 654848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9269688796177585e-05,
+      "loss": 4.0673,
+      "step": 655360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9261302848667065e-05,
+      "loss": 4.0733,
+      "step": 655872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9252933279960274e-05,
+      "loss": 4.069,
+      "step": 656384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9244547332449754e-05,
+      "loss": 4.0704,
+      "step": 656896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9236177763742964e-05,
+      "loss": 4.0667,
+      "step": 657408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9227791816232443e-05,
+      "loss": 4.0703,
+      "step": 657920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9219405868721923e-05,
+      "loss": 4.0782,
+      "step": 658432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9211019921211403e-05,
+      "loss": 4.0691,
+      "step": 658944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.920263397370088e-05,
+      "loss": 4.074,
+      "step": 659456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.91942644049941e-05,
+      "loss": 4.0601,
+      "step": 659968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.918587845748358e-05,
+      "loss": 4.0664,
+      "step": 660480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.917749250997306e-05,
+      "loss": 4.0714,
+      "step": 660992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.916910656246254e-05,
+      "loss": 4.078,
+      "step": 661504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.916072061495202e-05,
+      "loss": 4.0691,
+      "step": 662016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.915233466744149e-05,
+      "loss": 4.0613,
+      "step": 662528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.914394871993097e-05,
+      "loss": 4.0697,
+      "step": 663040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.913557915122419e-05,
+      "loss": 4.0736,
+      "step": 663552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.912719320371367e-05,
+      "loss": 4.0558,
+      "step": 664064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.911880725620314e-05,
+      "loss": 4.0681,
+      "step": 664576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.911042130869262e-05,
+      "loss": 4.0537,
+      "step": 665088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.91020353611821e-05,
+      "loss": 4.074,
+      "step": 665600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.909364941367159e-05,
+      "loss": 4.0592,
+      "step": 666112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.908526346616107e-05,
+      "loss": 4.0671,
+      "step": 666624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.907689389745428e-05,
+      "loss": 4.0583,
+      "step": 667136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.906850794994376e-05,
+      "loss": 4.067,
+      "step": 667648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.906012200243324e-05,
+      "loss": 4.0727,
+      "step": 668160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.905173605492272e-05,
+      "loss": 4.0598,
+      "step": 668672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.90433501074122e-05,
+      "loss": 4.0677,
+      "step": 669184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9034980538705406e-05,
+      "loss": 4.0794,
+      "step": 669696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9026594591194886e-05,
+      "loss": 4.0751,
+      "step": 670208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9018208643684366e-05,
+      "loss": 4.0642,
+      "step": 670720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9009822696173846e-05,
+      "loss": 4.0551,
+      "step": 671232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9001436748663326e-05,
+      "loss": 4.0658,
+      "step": 671744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8993050801152806e-05,
+      "loss": 4.0717,
+      "step": 672256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.898468123244602e-05,
+      "loss": 4.066,
+      "step": 672768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.89762952849355e-05,
+      "loss": 4.0562,
+      "step": 673280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.896790933742498e-05,
+      "loss": 4.0732,
+      "step": 673792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.895952338991446e-05,
+      "loss": 4.0651,
+      "step": 674304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.895113744240394e-05,
+      "loss": 4.0697,
+      "step": 674816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.894275149489342e-05,
+      "loss": 4.0571,
+      "step": 675328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.89343655473829e-05,
+      "loss": 4.0675,
+      "step": 675840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.892597959987238e-05,
+      "loss": 4.0593,
+      "step": 676352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8917593652361854e-05,
+      "loss": 4.0761,
+      "step": 676864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.890922408365507e-05,
+      "loss": 4.0661,
+      "step": 677376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.890083813614455e-05,
+      "loss": 4.067,
+      "step": 677888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.889245218863403e-05,
+      "loss": 4.0568,
+      "step": 678400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.888406624112351e-05,
+      "loss": 4.0804,
+      "step": 678912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8875713051220455e-05,
+      "loss": 4.0628,
+      "step": 679424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8867327103709935e-05,
+      "loss": 4.0729,
+      "step": 679936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8858941156199415e-05,
+      "loss": 4.0663,
+      "step": 680448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8850555208688895e-05,
+      "loss": 4.065,
+      "step": 680960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8842169261178375e-05,
+      "loss": 4.0646,
+      "step": 681472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8833783313667855e-05,
+      "loss": 4.0782,
+      "step": 681984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.882539736615733e-05,
+      "loss": 4.0601,
+      "step": 682496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.881701141864681e-05,
+      "loss": 4.0744,
+      "step": 683008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8808641849940024e-05,
+      "loss": 4.0662,
+      "step": 683520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.88002559024295e-05,
+      "loss": 4.061,
+      "step": 684032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.879186995491898e-05,
+      "loss": 4.0619,
+      "step": 684544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8783484007408464e-05,
+      "loss": 4.063,
+      "step": 685056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8775098059897944e-05,
+      "loss": 4.0662,
+      "step": 685568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.876672849119115e-05,
+      "loss": 4.0658,
+      "step": 686080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.875835892248437e-05,
+      "loss": 4.0729,
+      "step": 686592
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.0872039794921875,
+      "eval_runtime": 292.1205,
+      "eval_samples_per_second": 1306.279,
+      "eval_steps_per_second": 40.822,
+      "step": 686880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.874997297497385e-05,
+      "loss": 4.0613,
+      "step": 687104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.874158702746333e-05,
+      "loss": 4.0569,
+      "step": 687616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.87332010799528e-05,
+      "loss": 4.0586,
+      "step": 688128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.872481513244228e-05,
+      "loss": 4.0634,
+      "step": 688640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.871642918493176e-05,
+      "loss": 4.0735,
+      "step": 689152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.870804323742124e-05,
+      "loss": 4.0627,
+      "step": 689664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.869965728991072e-05,
+      "loss": 4.0628,
+      "step": 690176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.86912713424002e-05,
+      "loss": 4.0592,
+      "step": 690688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.868288539488968e-05,
+      "loss": 4.0632,
+      "step": 691200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.867449944737916e-05,
+      "loss": 4.07,
+      "step": 691712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.866611349986864e-05,
+      "loss": 4.0569,
+      "step": 692224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.865772755235813e-05,
+      "loss": 4.0529,
+      "step": 692736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.864935798365134e-05,
+      "loss": 4.0691,
+      "step": 693248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.864097203614082e-05,
+      "loss": 4.0557,
+      "step": 693760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.86325860886303e-05,
+      "loss": 4.0574,
+      "step": 694272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.862420014111978e-05,
+      "loss": 4.0486,
+      "step": 694784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.861581419360926e-05,
+      "loss": 4.0508,
+      "step": 695296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.860742824609874e-05,
+      "loss": 4.0534,
+      "step": 695808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.859904229858822e-05,
+      "loss": 4.0495,
+      "step": 696320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.859065635107769e-05,
+      "loss": 4.0627,
+      "step": 696832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.858227040356717e-05,
+      "loss": 4.0672,
+      "step": 697344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.857388445605665e-05,
+      "loss": 4.0656,
+      "step": 697856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.856549850854613e-05,
+      "loss": 4.0563,
+      "step": 698368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.855711256103561e-05,
+      "loss": 4.0568,
+      "step": 698880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.854875937113256e-05,
+      "loss": 4.0649,
+      "step": 699392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.854037342362204e-05,
+      "loss": 4.0541,
+      "step": 699904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8531987476111515e-05,
+      "loss": 4.0513,
+      "step": 700416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8523601528600995e-05,
+      "loss": 4.0662,
+      "step": 700928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8515215581090475e-05,
+      "loss": 4.054,
+      "step": 701440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8506829633579955e-05,
+      "loss": 4.0498,
+      "step": 701952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8498443686069435e-05,
+      "loss": 4.0525,
+      "step": 702464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8490057738558915e-05,
+      "loss": 4.0622,
+      "step": 702976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8481671791048395e-05,
+      "loss": 4.0644,
+      "step": 703488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8473285843537875e-05,
+      "loss": 4.0715,
+      "step": 704000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8464916274831084e-05,
+      "loss": 4.0497,
+      "step": 704512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8456530327320564e-05,
+      "loss": 4.0594,
+      "step": 705024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.844814437981005e-05,
+      "loss": 4.0666,
+      "step": 705536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.843977481110326e-05,
+      "loss": 4.0411,
+      "step": 706048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.843138886359274e-05,
+      "loss": 4.0515,
+      "step": 706560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.842300291608222e-05,
+      "loss": 4.0474,
+      "step": 707072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.84146169685717e-05,
+      "loss": 4.0522,
+      "step": 707584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.840624739986491e-05,
+      "loss": 4.0583,
+      "step": 708096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.839786145235439e-05,
+      "loss": 4.0539,
+      "step": 708608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.838947550484387e-05,
+      "loss": 4.0591,
+      "step": 709120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.838108955733335e-05,
+      "loss": 4.0532,
+      "step": 709632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.837270360982283e-05,
+      "loss": 4.0541,
+      "step": 710144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.836433404111604e-05,
+      "loss": 4.0527,
+      "step": 710656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.835594809360552e-05,
+      "loss": 4.0533,
+      "step": 711168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8347562146095004e-05,
+      "loss": 4.0241,
+      "step": 711680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.833919257738821e-05,
+      "loss": 4.068,
+      "step": 712192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.833080662987769e-05,
+      "loss": 4.0474,
+      "step": 712704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.832242068236717e-05,
+      "loss": 4.0584,
+      "step": 713216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.831403473485665e-05,
+      "loss": 4.0492,
+      "step": 713728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.830564878734613e-05,
+      "loss": 4.0527,
+      "step": 714240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.829726283983561e-05,
+      "loss": 4.0397,
+      "step": 714752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.828887689232509e-05,
+      "loss": 4.0428,
+      "step": 715264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.828049094481457e-05,
+      "loss": 4.0573,
+      "step": 715776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.827212137610778e-05,
+      "loss": 4.0431,
+      "step": 716288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.826373542859726e-05,
+      "loss": 4.0562,
+      "step": 716800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.825534948108674e-05,
+      "loss": 4.0513,
+      "step": 717312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.824696353357622e-05,
+      "loss": 4.0359,
+      "step": 717824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.82385775860657e-05,
+      "loss": 4.0645,
+      "step": 718336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.823019163855518e-05,
+      "loss": 4.035,
+      "step": 718848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.82218220698484e-05,
+      "loss": 4.0442,
+      "step": 719360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.821343612233788e-05,
+      "loss": 4.0575,
+      "step": 719872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.820505017482735e-05,
+      "loss": 4.052,
+      "step": 720384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.819666422731683e-05,
+      "loss": 4.0376,
+      "step": 720896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.818827827980631e-05,
+      "loss": 4.0516,
+      "step": 721408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.817989233229579e-05,
+      "loss": 4.031,
+      "step": 721920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.817150638478527e-05,
+      "loss": 4.0563,
+      "step": 722432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.816312043727475e-05,
+      "loss": 4.0583,
+      "step": 722944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.815475086856796e-05,
+      "loss": 4.0499,
+      "step": 723456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.814636492105744e-05,
+      "loss": 4.054,
+      "step": 723968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.813797897354692e-05,
+      "loss": 4.0587,
+      "step": 724480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8129593026036406e-05,
+      "loss": 4.0671,
+      "step": 724992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8121207078525886e-05,
+      "loss": 4.0452,
+      "step": 725504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8112853888622824e-05,
+      "loss": 4.0479,
+      "step": 726016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8104467941112304e-05,
+      "loss": 4.0569,
+      "step": 726528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8096081993601784e-05,
+      "loss": 4.0254,
+      "step": 727040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8087696046091264e-05,
+      "loss": 4.0489,
+      "step": 727552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8079310098580744e-05,
+      "loss": 4.0518,
+      "step": 728064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8070924151070224e-05,
+      "loss": 4.0508,
+      "step": 728576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8062538203559704e-05,
+      "loss": 4.0681,
+      "step": 729088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8054152256049184e-05,
+      "loss": 4.0384,
+      "step": 729600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.804578268734239e-05,
+      "loss": 4.0495,
+      "step": 730112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.803739673983187e-05,
+      "loss": 4.0497,
+      "step": 730624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.802901079232136e-05,
+      "loss": 4.0509,
+      "step": 731136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.802062484481084e-05,
+      "loss": 4.0472,
+      "step": 731648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.801223889730032e-05,
+      "loss": 4.0476,
+      "step": 732160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.80038529497898e-05,
+      "loss": 4.0405,
+      "step": 732672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.799548338108301e-05,
+      "loss": 4.0492,
+      "step": 733184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.798709743357249e-05,
+      "loss": 4.0428,
+      "step": 733696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.797871148606197e-05,
+      "loss": 4.0455,
+      "step": 734208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.797032553855145e-05,
+      "loss": 4.0543,
+      "step": 734720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.796193959104093e-05,
+      "loss": 4.0452,
+      "step": 735232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.795355364353041e-05,
+      "loss": 4.0506,
+      "step": 735744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.794518407482362e-05,
+      "loss": 4.0386,
+      "step": 736256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.79367981273131e-05,
+      "loss": 4.0418,
+      "step": 736768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.792841217980258e-05,
+      "loss": 4.0471,
+      "step": 737280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.792002623229206e-05,
+      "loss": 4.0528,
+      "step": 737792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.791164028478154e-05,
+      "loss": 4.0452,
+      "step": 738304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.790325433727102e-05,
+      "loss": 4.0417,
+      "step": 738816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.78948683897605e-05,
+      "loss": 4.0476,
+      "step": 739328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.788649882105371e-05,
+      "loss": 4.0493,
+      "step": 739840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7878112873543186e-05,
+      "loss": 4.033,
+      "step": 740352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7869726926032666e-05,
+      "loss": 4.0454,
+      "step": 740864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7861340978522146e-05,
+      "loss": 4.0303,
+      "step": 741376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7852955031011626e-05,
+      "loss": 4.0509,
+      "step": 741888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7844569083501106e-05,
+      "loss": 4.0382,
+      "step": 742400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7836199514794315e-05,
+      "loss": 4.0414,
+      "step": 742912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7827813567283795e-05,
+      "loss": 4.0321,
+      "step": 743424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.781942761977328e-05,
+      "loss": 4.0503,
+      "step": 743936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.781104167226276e-05,
+      "loss": 4.044,
+      "step": 744448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.780265572475224e-05,
+      "loss": 4.0362,
+      "step": 744960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.779428615604545e-05,
+      "loss": 4.0473,
+      "step": 745472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.778590020853493e-05,
+      "loss": 4.0552,
+      "step": 745984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.777751426102441e-05,
+      "loss": 4.0528,
+      "step": 746496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.776912831351389e-05,
+      "loss": 4.0406,
+      "step": 747008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.776074236600337e-05,
+      "loss": 4.0336,
+      "step": 747520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.775235641849285e-05,
+      "loss": 4.0462,
+      "step": 748032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.774397047098233e-05,
+      "loss": 4.0505,
+      "step": 748544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.773560090227554e-05,
+      "loss": 4.0385,
+      "step": 749056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.772721495476502e-05,
+      "loss": 4.0358,
+      "step": 749568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.77188290072545e-05,
+      "loss": 4.0505,
+      "step": 750080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.771044305974398e-05,
+      "loss": 4.046,
+      "step": 750592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7702057112233467e-05,
+      "loss": 4.046,
+      "step": 751104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7693671164722946e-05,
+      "loss": 4.0342,
+      "step": 751616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7685285217212426e-05,
+      "loss": 4.0422,
+      "step": 752128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.76768992697019e-05,
+      "loss": 4.0366,
+      "step": 752640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.766851332219138e-05,
+      "loss": 4.0536,
+      "step": 753152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7660143753484595e-05,
+      "loss": 4.0406,
+      "step": 753664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.765175780597407e-05,
+      "loss": 4.0482,
+      "step": 754176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7643388237267284e-05,
+      "loss": 4.0298,
+      "step": 754688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7635002289756764e-05,
+      "loss": 4.0588,
+      "step": 755200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7626616342246244e-05,
+      "loss": 4.0364,
+      "step": 755712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.761823039473572e-05,
+      "loss": 4.0547,
+      "step": 756224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7609860826028933e-05,
+      "loss": 4.0432,
+      "step": 756736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7601474878518413e-05,
+      "loss": 4.0447,
+      "step": 757248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.75930889310079e-05,
+      "loss": 4.0375,
+      "step": 757760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.758470298349737e-05,
+      "loss": 4.0587,
+      "step": 758272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.757631703598685e-05,
+      "loss": 4.0378,
+      "step": 758784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.756793108847633e-05,
+      "loss": 4.0502,
+      "step": 759296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.755954514096581e-05,
+      "loss": 4.0463,
+      "step": 759808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.755115919345529e-05,
+      "loss": 4.0369,
+      "step": 760320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.754277324594477e-05,
+      "loss": 4.0424,
+      "step": 760832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.753440367723798e-05,
+      "loss": 4.0385,
+      "step": 761344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.752601772972746e-05,
+      "loss": 4.0424,
+      "step": 761856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.751764816102067e-05,
+      "loss": 4.0446,
+      "step": 762368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.750926221351015e-05,
+      "loss": 4.0478,
+      "step": 762880
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.071236610412598,
+      "eval_runtime": 296.4774,
+      "eval_samples_per_second": 1287.083,
+      "eval_steps_per_second": 40.222,
+      "step": 763200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.750087626599964e-05,
+      "loss": 4.0372,
+      "step": 763392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.749249031848912e-05,
+      "loss": 4.0339,
+      "step": 763904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.74841043709786e-05,
+      "loss": 4.0371,
+      "step": 764416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.747571842346808e-05,
+      "loss": 4.0403,
+      "step": 764928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.746733247595756e-05,
+      "loss": 4.0506,
+      "step": 765440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.745894652844704e-05,
+      "loss": 4.042,
+      "step": 765952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.745056058093652e-05,
+      "loss": 4.041,
+      "step": 766464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7442174633426e-05,
+      "loss": 4.0371,
+      "step": 766976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.743378868591548e-05,
+      "loss": 4.0414,
+      "step": 767488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.742540273840496e-05,
+      "loss": 4.0452,
+      "step": 768000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.741701679089443e-05,
+      "loss": 4.0368,
+      "step": 768512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.740863084338391e-05,
+      "loss": 4.0357,
+      "step": 769024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7400261274677127e-05,
+      "loss": 4.0461,
+      "step": 769536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7391875327166607e-05,
+      "loss": 4.0314,
+      "step": 770048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7383489379656086e-05,
+      "loss": 4.0348,
+      "step": 770560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7375103432145566e-05,
+      "loss": 4.0235,
+      "step": 771072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7366717484635046e-05,
+      "loss": 4.0326,
+      "step": 771584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7358331537124526e-05,
+      "loss": 4.032,
+      "step": 772096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7349945589614006e-05,
+      "loss": 4.0257,
+      "step": 772608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7341559642103486e-05,
+      "loss": 4.0417,
+      "step": 773120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7333173694592966e-05,
+      "loss": 4.0436,
+      "step": 773632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7324804125886175e-05,
+      "loss": 4.0412,
+      "step": 774144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7316418178375655e-05,
+      "loss": 4.0351,
+      "step": 774656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7308048609668864e-05,
+      "loss": 4.0369,
+      "step": 775168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7299662662158344e-05,
+      "loss": 4.0427,
+      "step": 775680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7291276714647824e-05,
+      "loss": 4.0297,
+      "step": 776192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7282890767137304e-05,
+      "loss": 4.0254,
+      "step": 776704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.727450481962679e-05,
+      "loss": 4.0439,
+      "step": 777216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.726611887211627e-05,
+      "loss": 4.0331,
+      "step": 777728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.725773292460575e-05,
+      "loss": 4.0299,
+      "step": 778240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.724934697709523e-05,
+      "loss": 4.0282,
+      "step": 778752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.724096102958471e-05,
+      "loss": 4.0399,
+      "step": 779264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.723257508207419e-05,
+      "loss": 4.0353,
+      "step": 779776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.722418913456367e-05,
+      "loss": 4.0524,
+      "step": 780288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.721580318705315e-05,
+      "loss": 4.0299,
+      "step": 780800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.720743361834636e-05,
+      "loss": 4.0333,
+      "step": 781312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.719906404963957e-05,
+      "loss": 4.0478,
+      "step": 781824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.719067810212905e-05,
+      "loss": 4.0154,
+      "step": 782336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.718229215461853e-05,
+      "loss": 4.031,
+      "step": 782848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.717390620710801e-05,
+      "loss": 4.0287,
+      "step": 783360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.716552025959749e-05,
+      "loss": 4.0249,
+      "step": 783872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.715713431208697e-05,
+      "loss": 4.0384,
+      "step": 784384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.714874836457645e-05,
+      "loss": 4.0332,
+      "step": 784896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.714036241706593e-05,
+      "loss": 4.0394,
+      "step": 785408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7131992848359144e-05,
+      "loss": 4.0296,
+      "step": 785920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.712360690084862e-05,
+      "loss": 4.0344,
+      "step": 786432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.71152209533381e-05,
+      "loss": 4.0261,
+      "step": 786944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.710683500582758e-05,
+      "loss": 4.0369,
+      "step": 787456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.709846543712079e-05,
+      "loss": 4.0053,
+      "step": 787968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7090095868414e-05,
+      "loss": 4.0429,
+      "step": 788480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.708170992090348e-05,
+      "loss": 4.0284,
+      "step": 788992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.707332397339296e-05,
+      "loss": 4.0377,
+      "step": 789504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.706493802588244e-05,
+      "loss": 4.0288,
+      "step": 790016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.705655207837192e-05,
+      "loss": 4.0287,
+      "step": 790528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.70481661308614e-05,
+      "loss": 4.0174,
+      "step": 791040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.703979656215462e-05,
+      "loss": 4.0234,
+      "step": 791552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.703141061464409e-05,
+      "loss": 4.0403,
+      "step": 792064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.702302466713357e-05,
+      "loss": 4.0174,
+      "step": 792576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.701463871962305e-05,
+      "loss": 4.0356,
+      "step": 793088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.700625277211253e-05,
+      "loss": 4.0344,
+      "step": 793600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.699786682460201e-05,
+      "loss": 4.0129,
+      "step": 794112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.698948087709149e-05,
+      "loss": 4.0412,
+      "step": 794624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.698109492958097e-05,
+      "loss": 4.0173,
+      "step": 795136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.697270898207045e-05,
+      "loss": 4.0212,
+      "step": 795648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.696433941336366e-05,
+      "loss": 4.0335,
+      "step": 796160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.695595346585315e-05,
+      "loss": 4.0282,
+      "step": 796672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.694756751834263e-05,
+      "loss": 4.0186,
+      "step": 797184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.693918157083211e-05,
+      "loss": 4.0315,
+      "step": 797696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6930812002125316e-05,
+      "loss": 4.0135,
+      "step": 798208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6922442433418525e-05,
+      "loss": 4.0328,
+      "step": 798720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6914056485908005e-05,
+      "loss": 4.0357,
+      "step": 799232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6905670538397485e-05,
+      "loss": 4.0311,
+      "step": 799744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6897284590886965e-05,
+      "loss": 4.0324,
+      "step": 800256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6888915022180174e-05,
+      "loss": 4.036,
+      "step": 800768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6880529074669654e-05,
+      "loss": 4.0427,
+      "step": 801280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.687215950596287e-05,
+      "loss": 4.0229,
+      "step": 801792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.686377355845235e-05,
+      "loss": 4.0299,
+      "step": 802304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.685538761094183e-05,
+      "loss": 4.0337,
+      "step": 802816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.684700166343131e-05,
+      "loss": 4.011,
+      "step": 803328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.683861571592079e-05,
+      "loss": 4.0251,
+      "step": 803840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.683022976841027e-05,
+      "loss": 4.0332,
+      "step": 804352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.682184382089975e-05,
+      "loss": 4.0273,
+      "step": 804864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.681345787338923e-05,
+      "loss": 4.0488,
+      "step": 805376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.680507192587871e-05,
+      "loss": 4.0151,
+      "step": 805888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.679668597836819e-05,
+      "loss": 4.0284,
+      "step": 806400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.678830003085767e-05,
+      "loss": 4.0301,
+      "step": 806912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.677991408334715e-05,
+      "loss": 4.0283,
+      "step": 807424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.677156089344409e-05,
+      "loss": 4.0301,
+      "step": 807936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.676317494593357e-05,
+      "loss": 4.0266,
+      "step": 808448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6754788998423054e-05,
+      "loss": 4.0168,
+      "step": 808960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6746403050912534e-05,
+      "loss": 4.0276,
+      "step": 809472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6738017103402014e-05,
+      "loss": 4.028,
+      "step": 809984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.672964753469522e-05,
+      "loss": 4.0215,
+      "step": 810496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.67212615871847e-05,
+      "loss": 4.0326,
+      "step": 811008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.671287563967418e-05,
+      "loss": 4.0311,
+      "step": 811520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.670450607096739e-05,
+      "loss": 4.0233,
+      "step": 812032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.669612012345687e-05,
+      "loss": 4.0198,
+      "step": 812544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.668773417594635e-05,
+      "loss": 4.0205,
+      "step": 813056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.667934822843583e-05,
+      "loss": 4.0232,
+      "step": 813568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.667096228092531e-05,
+      "loss": 4.0334,
+      "step": 814080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.666257633341479e-05,
+      "loss": 4.0285,
+      "step": 814592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.665419038590427e-05,
+      "loss": 4.0194,
+      "step": 815104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.664580443839375e-05,
+      "loss": 4.0262,
+      "step": 815616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.663741849088323e-05,
+      "loss": 4.0227,
+      "step": 816128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.662903254337271e-05,
+      "loss": 4.0188,
+      "step": 816640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.662064659586219e-05,
+      "loss": 4.0237,
+      "step": 817152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.661226064835167e-05,
+      "loss": 4.0093,
+      "step": 817664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.660389107964488e-05,
+      "loss": 4.0317,
+      "step": 818176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.659550513213436e-05,
+      "loss": 4.0218,
+      "step": 818688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.658711918462384e-05,
+      "loss": 4.0152,
+      "step": 819200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.657873323711332e-05,
+      "loss": 4.0109,
+      "step": 819712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.65703472896028e-05,
+      "loss": 4.0268,
+      "step": 820224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6561994099699746e-05,
+      "loss": 4.026,
+      "step": 820736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6553608152189226e-05,
+      "loss": 4.0153,
+      "step": 821248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6545222204678705e-05,
+      "loss": 4.0266,
+      "step": 821760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6536836257168185e-05,
+      "loss": 4.0373,
+      "step": 822272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6528450309657665e-05,
+      "loss": 4.0296,
+      "step": 822784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6520080740950875e-05,
+      "loss": 4.0221,
+      "step": 823296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6511694793440354e-05,
+      "loss": 4.0142,
+      "step": 823808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6503308845929834e-05,
+      "loss": 4.0212,
+      "step": 824320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6494922898419314e-05,
+      "loss": 4.0328,
+      "step": 824832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6486553329712523e-05,
+      "loss": 4.0189,
+      "step": 825344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6478167382202003e-05,
+      "loss": 4.0137,
+      "step": 825856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.646978143469148e-05,
+      "loss": 4.0297,
+      "step": 826368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.646139548718096e-05,
+      "loss": 4.0236,
+      "step": 826880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.645300953967044e-05,
+      "loss": 4.0225,
+      "step": 827392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.644462359215992e-05,
+      "loss": 4.0174,
+      "step": 827904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.643623764464941e-05,
+      "loss": 4.0213,
+      "step": 828416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.642785169713889e-05,
+      "loss": 4.0182,
+      "step": 828928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.64194821284321e-05,
+      "loss": 4.0331,
+      "step": 829440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.641109618092158e-05,
+      "loss": 4.0193,
+      "step": 829952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.640271023341106e-05,
+      "loss": 4.0271,
+      "step": 830464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.639432428590054e-05,
+      "loss": 4.0123,
+      "step": 830976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.638595471719375e-05,
+      "loss": 4.0325,
+      "step": 831488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.637756876968323e-05,
+      "loss": 4.0232,
+      "step": 832000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.636919920097644e-05,
+      "loss": 4.0292,
+      "step": 832512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.636081325346592e-05,
+      "loss": 4.0257,
+      "step": 833024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.63524273059554e-05,
+      "loss": 4.0212,
+      "step": 833536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.634404135844488e-05,
+      "loss": 4.024,
+      "step": 834048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6335655410934364e-05,
+      "loss": 4.0347,
+      "step": 834560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6327269463423844e-05,
+      "loss": 4.0204,
+      "step": 835072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6318883515913324e-05,
+      "loss": 4.0244,
+      "step": 835584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6310497568402804e-05,
+      "loss": 4.0291,
+      "step": 836096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6302111620892283e-05,
+      "loss": 4.0181,
+      "step": 836608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6293725673381763e-05,
+      "loss": 4.0203,
+      "step": 837120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6285339725871237e-05,
+      "loss": 4.0221,
+      "step": 837632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.627697015716445e-05,
+      "loss": 4.0226,
+      "step": 838144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.626858420965393e-05,
+      "loss": 4.0236,
+      "step": 838656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.626019826214341e-05,
+      "loss": 4.0291,
+      "step": 839168
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.057768821716309,
+      "eval_runtime": 288.1865,
+      "eval_samples_per_second": 1324.111,
+      "eval_steps_per_second": 41.379,
+      "step": 839520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6251812314632886e-05,
+      "loss": 4.0109,
+      "step": 839680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6243426367122366e-05,
+      "loss": 4.0172,
+      "step": 840192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6235040419611845e-05,
+      "loss": 4.0178,
+      "step": 840704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.622665447210133e-05,
+      "loss": 4.0215,
+      "step": 841216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.621826852459081e-05,
+      "loss": 4.03,
+      "step": 841728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.620989895588402e-05,
+      "loss": 4.0225,
+      "step": 842240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.62015130083735e-05,
+      "loss": 4.0211,
+      "step": 842752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.619312706086298e-05,
+      "loss": 4.0184,
+      "step": 843264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.618474111335246e-05,
+      "loss": 4.0215,
+      "step": 843776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.617635516584194e-05,
+      "loss": 4.0231,
+      "step": 844288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.616798559713515e-05,
+      "loss": 4.0207,
+      "step": 844800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.615959964962463e-05,
+      "loss": 4.0132,
+      "step": 845312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.615123008091784e-05,
+      "loss": 4.0258,
+      "step": 845824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.614284413340732e-05,
+      "loss": 4.0105,
+      "step": 846336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.61344581858968e-05,
+      "loss": 4.0173,
+      "step": 846848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6126072238386286e-05,
+      "loss": 4.0037,
+      "step": 847360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6117686290875766e-05,
+      "loss": 4.0174,
+      "step": 847872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6109300343365246e-05,
+      "loss": 4.0133,
+      "step": 848384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6100914395854726e-05,
+      "loss": 4.0046,
+      "step": 848896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6092528448344206e-05,
+      "loss": 4.0219,
+      "step": 849408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6084158879637415e-05,
+      "loss": 4.0222,
+      "step": 849920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6075772932126895e-05,
+      "loss": 4.0203,
+      "step": 850432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6067386984616375e-05,
+      "loss": 4.0175,
+      "step": 850944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6059001037105855e-05,
+      "loss": 4.0169,
+      "step": 851456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6050615089595335e-05,
+      "loss": 4.0243,
+      "step": 851968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.604226189969227e-05,
+      "loss": 4.0172,
+      "step": 852480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.603387595218175e-05,
+      "loss": 4.0013,
+      "step": 852992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.602549000467124e-05,
+      "loss": 4.0237,
+      "step": 853504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.601710405716072e-05,
+      "loss": 4.0133,
+      "step": 854016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.60087181096502e-05,
+      "loss": 4.0104,
+      "step": 854528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.600033216213968e-05,
+      "loss": 4.0139,
+      "step": 855040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.599194621462916e-05,
+      "loss": 4.0173,
+      "step": 855552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.598357664592237e-05,
+      "loss": 4.0179,
+      "step": 856064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.597519069841185e-05,
+      "loss": 4.0345,
+      "step": 856576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.596680475090133e-05,
+      "loss": 4.0114,
+      "step": 857088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.595841880339081e-05,
+      "loss": 4.0138,
+      "step": 857600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.595004923468402e-05,
+      "loss": 4.0283,
+      "step": 858112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.59416632871735e-05,
+      "loss": 3.9944,
+      "step": 858624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.593327733966298e-05,
+      "loss": 4.0118,
+      "step": 859136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.592489139215246e-05,
+      "loss": 4.0093,
+      "step": 859648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.591650544464194e-05,
+      "loss": 4.0063,
+      "step": 860160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.590811949713142e-05,
+      "loss": 4.0192,
+      "step": 860672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.58997335496209e-05,
+      "loss": 4.0137,
+      "step": 861184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.589134760211038e-05,
+      "loss": 4.0176,
+      "step": 861696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.588296165459986e-05,
+      "loss": 4.0124,
+      "step": 862208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.587459208589307e-05,
+      "loss": 4.0174,
+      "step": 862720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.586622251718628e-05,
+      "loss": 4.005,
+      "step": 863232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.585783656967576e-05,
+      "loss": 4.021,
+      "step": 863744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.584945062216524e-05,
+      "loss": 3.9869,
+      "step": 864256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5841064674654715e-05,
+      "loss": 4.0175,
+      "step": 864768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5832678727144195e-05,
+      "loss": 4.0104,
+      "step": 865280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5824292779633675e-05,
+      "loss": 4.022,
+      "step": 865792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5815906832123155e-05,
+      "loss": 4.0061,
+      "step": 866304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.580752088461264e-05,
+      "loss": 4.013,
+      "step": 866816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.579915131590585e-05,
+      "loss": 4.0029,
+      "step": 867328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.579078174719907e-05,
+      "loss": 4.0025,
+      "step": 867840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.578239579968855e-05,
+      "loss": 4.0247,
+      "step": 868352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.577400985217802e-05,
+      "loss": 3.9962,
+      "step": 868864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.57656239046675e-05,
+      "loss": 4.0155,
+      "step": 869376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.575723795715698e-05,
+      "loss": 4.0168,
+      "step": 869888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.574885200964646e-05,
+      "loss": 3.9924,
+      "step": 870400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.574046606213594e-05,
+      "loss": 4.025,
+      "step": 870912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.573208011462542e-05,
+      "loss": 3.9984,
+      "step": 871424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.57236941671149e-05,
+      "loss": 4.0017,
+      "step": 871936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.571532459840811e-05,
+      "loss": 4.0167,
+      "step": 872448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5706938650897595e-05,
+      "loss": 4.0131,
+      "step": 872960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5698552703387075e-05,
+      "loss": 4.0003,
+      "step": 873472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5690183134680284e-05,
+      "loss": 4.0119,
+      "step": 873984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5681797187169764e-05,
+      "loss": 3.9926,
+      "step": 874496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5673411239659244e-05,
+      "loss": 4.0145,
+      "step": 875008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5665025292148724e-05,
+      "loss": 4.0198,
+      "step": 875520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5656639344638204e-05,
+      "loss": 4.014,
+      "step": 876032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5648253397127684e-05,
+      "loss": 4.0105,
+      "step": 876544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5639867449617164e-05,
+      "loss": 4.0151,
+      "step": 877056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5631481502106644e-05,
+      "loss": 4.0269,
+      "step": 877568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.562311193339985e-05,
+      "loss": 4.0064,
+      "step": 878080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.561472598588933e-05,
+      "loss": 4.0146,
+      "step": 878592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.560634003837881e-05,
+      "loss": 4.0058,
+      "step": 879104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.559795409086829e-05,
+      "loss": 3.9979,
+      "step": 879616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.558956814335778e-05,
+      "loss": 4.0047,
+      "step": 880128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.558118219584726e-05,
+      "loss": 4.0126,
+      "step": 880640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.557279624833673e-05,
+      "loss": 4.0099,
+      "step": 881152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.556441030082621e-05,
+      "loss": 4.0319,
+      "step": 881664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.555604073211943e-05,
+      "loss": 3.9945,
+      "step": 882176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.554767116341264e-05,
+      "loss": 4.0118,
+      "step": 882688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.553928521590212e-05,
+      "loss": 4.0126,
+      "step": 883200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.55308992683916e-05,
+      "loss": 4.0085,
+      "step": 883712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.552251332088108e-05,
+      "loss": 4.0096,
+      "step": 884224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.551412737337055e-05,
+      "loss": 4.0069,
+      "step": 884736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.550574142586003e-05,
+      "loss": 4.0027,
+      "step": 885248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.549735547834952e-05,
+      "loss": 4.0094,
+      "step": 885760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5488969530839e-05,
+      "loss": 4.0103,
+      "step": 886272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.548058358332848e-05,
+      "loss": 4.0028,
+      "step": 886784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.547221401462169e-05,
+      "loss": 4.0137,
+      "step": 887296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5463828067111167e-05,
+      "loss": 4.0143,
+      "step": 887808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5455442119600647e-05,
+      "loss": 4.0048,
+      "step": 888320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5447056172090126e-05,
+      "loss": 4.0033,
+      "step": 888832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.543870298218707e-05,
+      "loss": 4.0042,
+      "step": 889344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.543031703467655e-05,
+      "loss": 4.005,
+      "step": 889856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5421931087166025e-05,
+      "loss": 4.0141,
+      "step": 890368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5413545139655505e-05,
+      "loss": 4.008,
+      "step": 890880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5405159192144985e-05,
+      "loss": 3.9993,
+      "step": 891392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.539677324463447e-05,
+      "loss": 4.0083,
+      "step": 891904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.538838729712395e-05,
+      "loss": 4.0064,
+      "step": 892416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.538000134961343e-05,
+      "loss": 4.0027,
+      "step": 892928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.537163178090664e-05,
+      "loss": 4.0025,
+      "step": 893440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.536324583339612e-05,
+      "loss": 3.994,
+      "step": 893952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.53548598858856e-05,
+      "loss": 4.0112,
+      "step": 894464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.534647393837508e-05,
+      "loss": 4.0065,
+      "step": 894976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.533810436966829e-05,
+      "loss": 3.9972,
+      "step": 895488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.532971842215777e-05,
+      "loss": 3.9927,
+      "step": 896000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.532133247464725e-05,
+      "loss": 4.0077,
+      "step": 896512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.531294652713673e-05,
+      "loss": 4.0084,
+      "step": 897024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.530456057962621e-05,
+      "loss": 3.996,
+      "step": 897536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5296191010919425e-05,
+      "loss": 4.0107,
+      "step": 898048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5287805063408905e-05,
+      "loss": 4.0191,
+      "step": 898560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5279419115898385e-05,
+      "loss": 4.0092,
+      "step": 899072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5271033168387865e-05,
+      "loss": 4.0057,
+      "step": 899584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5262647220877345e-05,
+      "loss": 3.999,
+      "step": 900096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5254277652170554e-05,
+      "loss": 4.0068,
+      "step": 900608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5245891704660034e-05,
+      "loss": 4.0147,
+      "step": 901120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5237505757149514e-05,
+      "loss": 4.0018,
+      "step": 901632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5229119809638994e-05,
+      "loss": 3.9964,
+      "step": 902144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5220733862128474e-05,
+      "loss": 4.0082,
+      "step": 902656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5212347914617954e-05,
+      "loss": 4.0051,
+      "step": 903168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5203961967107434e-05,
+      "loss": 4.0071,
+      "step": 903680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5195576019596914e-05,
+      "loss": 3.9991,
+      "step": 904192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.518720645089012e-05,
+      "loss": 3.9985,
+      "step": 904704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.51788205033796e-05,
+      "loss": 3.9999,
+      "step": 905216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.517043455586909e-05,
+      "loss": 4.0143,
+      "step": 905728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.516204860835857e-05,
+      "loss": 4.0037,
+      "step": 906240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.515367903965178e-05,
+      "loss": 4.0116,
+      "step": 906752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.514529309214126e-05,
+      "loss": 3.997,
+      "step": 907264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.513692352343447e-05,
+      "loss": 4.0105,
+      "step": 907776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.512853757592395e-05,
+      "loss": 4.0061,
+      "step": 908288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.512015162841343e-05,
+      "loss": 4.0084,
+      "step": 908800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.511176568090291e-05,
+      "loss": 4.0117,
+      "step": 909312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.510337973339239e-05,
+      "loss": 4.0033,
+      "step": 909824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.509499378588186e-05,
+      "loss": 4.0059,
+      "step": 910336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.508660783837134e-05,
+      "loss": 4.018,
+      "step": 910848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.507822189086083e-05,
+      "loss": 4.0037,
+      "step": 911360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.506985232215404e-05,
+      "loss": 4.005,
+      "step": 911872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5061466374643516e-05,
+      "loss": 4.0126,
+      "step": 912384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5053080427132996e-05,
+      "loss": 3.9992,
+      "step": 912896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5044694479622476e-05,
+      "loss": 4.0056,
+      "step": 913408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5036308532111956e-05,
+      "loss": 4.0105,
+      "step": 913920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5027938963405165e-05,
+      "loss": 4.001,
+      "step": 914432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5019553015894645e-05,
+      "loss": 4.0067,
+      "step": 914944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5011167068384125e-05,
+      "loss": 4.0116,
+      "step": 915456
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.047426223754883,
+      "eval_runtime": 288.5281,
+      "eval_samples_per_second": 1322.544,
+      "eval_steps_per_second": 41.33,
+      "step": 915840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5002781120873605e-05,
+      "loss": 4.0073,
+      "step": 915968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4994395173363085e-05,
+      "loss": 3.9961,
+      "step": 916480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4986009225852565e-05,
+      "loss": 3.9987,
+      "step": 916992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4977623278342045e-05,
+      "loss": 4.0055,
+      "step": 917504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4969237330831525e-05,
+      "loss": 4.0056,
+      "step": 918016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.496085138332101e-05,
+      "loss": 4.0076,
+      "step": 918528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.495246543581049e-05,
+      "loss": 4.0076,
+      "step": 919040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.494407948829997e-05,
+      "loss": 4.0002,
+      "step": 919552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.493569354078945e-05,
+      "loss": 4.0008,
+      "step": 920064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.492730759327893e-05,
+      "loss": 4.0037,
+      "step": 920576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4918921645768405e-05,
+      "loss": 4.0054,
+      "step": 921088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4910535698257885e-05,
+      "loss": 3.9927,
+      "step": 921600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.49021661295511e-05,
+      "loss": 4.013,
+      "step": 922112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4893780182040574e-05,
+      "loss": 3.9962,
+      "step": 922624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4885394234530054e-05,
+      "loss": 3.9961,
+      "step": 923136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4877008287019533e-05,
+      "loss": 3.9898,
+      "step": 923648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4868622339509013e-05,
+      "loss": 3.9995,
+      "step": 924160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4860236391998493e-05,
+      "loss": 3.9931,
+      "step": 924672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.485185044448798e-05,
+      "loss": 3.9912,
+      "step": 925184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.484346449697746e-05,
+      "loss": 3.9987,
+      "step": 925696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.483507854946694e-05,
+      "loss": 4.0104,
+      "step": 926208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.482670898076015e-05,
+      "loss": 4.0012,
+      "step": 926720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.481832303324963e-05,
+      "loss": 3.997,
+      "step": 927232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.480993708573911e-05,
+      "loss": 4.0034,
+      "step": 927744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.480155113822859e-05,
+      "loss": 4.0066,
+      "step": 928256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.479316519071807e-05,
+      "loss": 3.9941,
+      "step": 928768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.478477924320755e-05,
+      "loss": 3.987,
+      "step": 929280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.477639329569703e-05,
+      "loss": 4.006,
+      "step": 929792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.476800734818651e-05,
+      "loss": 3.9981,
+      "step": 930304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.475962140067599e-05,
+      "loss": 3.9897,
+      "step": 930816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.475123545316546e-05,
+      "loss": 3.9967,
+      "step": 931328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.474286588445868e-05,
+      "loss": 3.9982,
+      "step": 931840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.473447993694816e-05,
+      "loss": 4.0049,
+      "step": 932352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4726093989437645e-05,
+      "loss": 4.0179,
+      "step": 932864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.471770804192712e-05,
+      "loss": 3.9915,
+      "step": 933376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4709338473220334e-05,
+      "loss": 4.0049,
+      "step": 933888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.470096890451354e-05,
+      "loss": 4.0091,
+      "step": 934400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.469258295700302e-05,
+      "loss": 3.9758,
+      "step": 934912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.46841970094925e-05,
+      "loss": 3.997,
+      "step": 935424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.467581106198198e-05,
+      "loss": 3.9956,
+      "step": 935936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.466742511447146e-05,
+      "loss": 3.985,
+      "step": 936448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4659039166960936e-05,
+      "loss": 4.0078,
+      "step": 936960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4650653219450416e-05,
+      "loss": 3.9939,
+      "step": 937472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4642267271939896e-05,
+      "loss": 4.0022,
+      "step": 937984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.463389770323311e-05,
+      "loss": 3.9954,
+      "step": 938496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.462551175572259e-05,
+      "loss": 4.0008,
+      "step": 939008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.461714218701581e-05,
+      "loss": 3.987,
+      "step": 939520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.460875623950529e-05,
+      "loss": 4.0044,
+      "step": 940032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.460037029199476e-05,
+      "loss": 3.9743,
+      "step": 940544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.459198434448424e-05,
+      "loss": 4.0016,
+      "step": 941056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.458359839697372e-05,
+      "loss": 3.9945,
+      "step": 941568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.45752124494632e-05,
+      "loss": 4.0061,
+      "step": 942080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.456682650195268e-05,
+      "loss": 3.9877,
+      "step": 942592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.455844055444216e-05,
+      "loss": 3.9917,
+      "step": 943104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.455007098573537e-05,
+      "loss": 3.9897,
+      "step": 943616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4541701417028585e-05,
+      "loss": 3.9869,
+      "step": 944128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4533315469518065e-05,
+      "loss": 4.0082,
+      "step": 944640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4524929522007545e-05,
+      "loss": 3.9761,
+      "step": 945152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4516543574497025e-05,
+      "loss": 4.0032,
+      "step": 945664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4508157626986505e-05,
+      "loss": 4.0017,
+      "step": 946176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4499788058279714e-05,
+      "loss": 3.9746,
+      "step": 946688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4491402110769194e-05,
+      "loss": 4.0064,
+      "step": 947200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4483016163258674e-05,
+      "loss": 3.9808,
+      "step": 947712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4474630215748154e-05,
+      "loss": 3.9851,
+      "step": 948224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.446626064704136e-05,
+      "loss": 3.9983,
+      "step": 948736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.445787469953084e-05,
+      "loss": 3.9993,
+      "step": 949248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.444948875202032e-05,
+      "loss": 3.9835,
+      "step": 949760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.44411028045098e-05,
+      "loss": 3.9952,
+      "step": 950272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.443271685699929e-05,
+      "loss": 3.9765,
+      "step": 950784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.442433090948877e-05,
+      "loss": 3.9963,
+      "step": 951296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.441594496197825e-05,
+      "loss": 4.0052,
+      "step": 951808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.440755901446773e-05,
+      "loss": 3.9962,
+      "step": 952320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.439917306695721e-05,
+      "loss": 3.9982,
+      "step": 952832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.439078711944669e-05,
+      "loss": 3.9968,
+      "step": 953344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.438240117193617e-05,
+      "loss": 4.0087,
+      "step": 953856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.437403160322938e-05,
+      "loss": 3.9917,
+      "step": 954368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.436564565571886e-05,
+      "loss": 4.0008,
+      "step": 954880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.435725970820834e-05,
+      "loss": 3.991,
+      "step": 955392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.434887376069782e-05,
+      "loss": 3.9797,
+      "step": 955904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.434050419199103e-05,
+      "loss": 3.9914,
+      "step": 956416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.433211824448051e-05,
+      "loss": 3.9963,
+      "step": 956928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.432373229696999e-05,
+      "loss": 3.9957,
+      "step": 957440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.431534634945947e-05,
+      "loss": 4.0132,
+      "step": 957952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.430697678075268e-05,
+      "loss": 3.9847,
+      "step": 958464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.429859083324216e-05,
+      "loss": 3.9916,
+      "step": 958976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.429020488573164e-05,
+      "loss": 3.993,
+      "step": 959488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.428181893822112e-05,
+      "loss": 3.9985,
+      "step": 960000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.427346574831806e-05,
+      "loss": 3.9908,
+      "step": 960512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.426507980080754e-05,
+      "loss": 3.9902,
+      "step": 961024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.425669385329702e-05,
+      "loss": 3.9889,
+      "step": 961536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.42483079057865e-05,
+      "loss": 3.991,
+      "step": 962048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.423992195827598e-05,
+      "loss": 3.9926,
+      "step": 962560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.42315523895692e-05,
+      "loss": 3.9852,
+      "step": 963072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.422316644205868e-05,
+      "loss": 4.0008,
+      "step": 963584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.421478049454816e-05,
+      "loss": 3.999,
+      "step": 964096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.420639454703764e-05,
+      "loss": 3.9895,
+      "step": 964608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.419800859952712e-05,
+      "loss": 3.9866,
+      "step": 965120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4189639030820326e-05,
+      "loss": 3.9862,
+      "step": 965632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4181253083309806e-05,
+      "loss": 3.9878,
+      "step": 966144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4172867135799286e-05,
+      "loss": 4.002,
+      "step": 966656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4164481188288766e-05,
+      "loss": 3.9891,
+      "step": 967168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4156095240778246e-05,
+      "loss": 3.9837,
+      "step": 967680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4147725672071455e-05,
+      "loss": 3.9947,
+      "step": 968192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4139339724560935e-05,
+      "loss": 3.9912,
+      "step": 968704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4130953777050415e-05,
+      "loss": 3.9915,
+      "step": 969216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4122567829539895e-05,
+      "loss": 3.989,
+      "step": 969728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4114181882029375e-05,
+      "loss": 3.9751,
+      "step": 970240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4105795934518855e-05,
+      "loss": 3.9943,
+      "step": 970752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4097409987008335e-05,
+      "loss": 3.9878,
+      "step": 971264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4089024039497815e-05,
+      "loss": 3.9843,
+      "step": 971776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4080638091987294e-05,
+      "loss": 3.9753,
+      "step": 972288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4072268523280504e-05,
+      "loss": 3.9917,
+      "step": 972800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4063882575769984e-05,
+      "loss": 3.9948,
+      "step": 973312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.405551300706319e-05,
+      "loss": 3.9779,
+      "step": 973824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.404712705955267e-05,
+      "loss": 3.9946,
+      "step": 974336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.403874111204215e-05,
+      "loss": 3.9991,
+      "step": 974848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.403035516453163e-05,
+      "loss": 3.9996,
+      "step": 975360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.402196921702111e-05,
+      "loss": 3.9891,
+      "step": 975872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.40135832695106e-05,
+      "loss": 3.983,
+      "step": 976384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.400519732200008e-05,
+      "loss": 3.9893,
+      "step": 976896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.399681137448956e-05,
+      "loss": 4.0007,
+      "step": 977408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.398844180578277e-05,
+      "loss": 3.9837,
+      "step": 977920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.398007223707598e-05,
+      "loss": 3.9846,
+      "step": 978432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.397168628956546e-05,
+      "loss": 3.9843,
+      "step": 978944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.396330034205494e-05,
+      "loss": 3.9932,
+      "step": 979456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.395491439454442e-05,
+      "loss": 3.9903,
+      "step": 979968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.39465284470339e-05,
+      "loss": 3.9872,
+      "step": 980480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3938158878327106e-05,
+      "loss": 3.9815,
+      "step": 980992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3929772930816586e-05,
+      "loss": 3.9842,
+      "step": 981504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3921386983306066e-05,
+      "loss": 4.003,
+      "step": 982016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.391300103579555e-05,
+      "loss": 3.9879,
+      "step": 982528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.390461508828503e-05,
+      "loss": 3.9959,
+      "step": 983040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.389624551957824e-05,
+      "loss": 3.9774,
+      "step": 983552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.388785957206772e-05,
+      "loss": 3.9981,
+      "step": 984064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.38794736245572e-05,
+      "loss": 3.9913,
+      "step": 984576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.387108767704668e-05,
+      "loss": 3.9911,
+      "step": 985088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.386270172953616e-05,
+      "loss": 3.9948,
+      "step": 985600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.385431578202564e-05,
+      "loss": 3.9842,
+      "step": 986112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.384592983451512e-05,
+      "loss": 3.9961,
+      "step": 986624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.38375438870046e-05,
+      "loss": 3.9997,
+      "step": 987136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.382917431829781e-05,
+      "loss": 3.9905,
+      "step": 987648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.382078837078729e-05,
+      "loss": 3.9885,
+      "step": 988160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.381240242327677e-05,
+      "loss": 3.9955,
+      "step": 988672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.380401647576625e-05,
+      "loss": 3.982,
+      "step": 989184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3795646907059466e-05,
+      "loss": 3.9876,
+      "step": 989696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3787260959548946e-05,
+      "loss": 3.9955,
+      "step": 990208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3778875012038426e-05,
+      "loss": 3.9852,
+      "step": 990720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3770489064527906e-05,
+      "loss": 3.991,
+      "step": 991232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3762119495821115e-05,
+      "loss": 3.9934,
+      "step": 991744
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.037771224975586,
+      "eval_runtime": 302.0435,
+      "eval_samples_per_second": 1263.364,
+      "eval_steps_per_second": 39.481,
+      "step": 992160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3753733548310595e-05,
+      "loss": 3.9828,
+      "step": 992256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3745347600800075e-05,
+      "loss": 3.9795,
+      "step": 992768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3736961653289555e-05,
+      "loss": 3.9847,
+      "step": 993280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.372857570577903e-05,
+      "loss": 3.9895,
+      "step": 993792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.372018975826851e-05,
+      "loss": 3.9928,
+      "step": 994304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.371180381075799e-05,
+      "loss": 3.9927,
+      "step": 994816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3703417863247475e-05,
+      "loss": 3.9856,
+      "step": 995328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3695031915736955e-05,
+      "loss": 3.9891,
+      "step": 995840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3686662347030164e-05,
+      "loss": 3.9822,
+      "step": 996352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3678276399519644e-05,
+      "loss": 3.9953,
+      "step": 996864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3669890452009124e-05,
+      "loss": 3.9836,
+      "step": 997376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3661504504498604e-05,
+      "loss": 3.981,
+      "step": 997888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.365313493579181e-05,
+      "loss": 3.9975,
+      "step": 998400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.364474898828129e-05,
+      "loss": 3.9764,
+      "step": 998912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.363636304077077e-05,
+      "loss": 3.9848,
+      "step": 999424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.362797709326025e-05,
+      "loss": 3.9745,
+      "step": 999936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.361960752455346e-05,
+      "loss": 3.9837,
+      "step": 1000448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.361122157704294e-05,
+      "loss": 3.9786,
+      "step": 1000960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.360283562953243e-05,
+      "loss": 3.9764,
+      "step": 1001472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.359444968202191e-05,
+      "loss": 3.9774,
+      "step": 1001984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.358606373451139e-05,
+      "loss": 3.998,
+      "step": 1002496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.357767778700087e-05,
+      "loss": 3.9858,
+      "step": 1003008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.356929183949035e-05,
+      "loss": 3.9827,
+      "step": 1003520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.356090589197983e-05,
+      "loss": 3.9902,
+      "step": 1004032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.355251994446931e-05,
+      "loss": 3.9954,
+      "step": 1004544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.354415037576252e-05,
+      "loss": 3.9768,
+      "step": 1005056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3535764428252e-05,
+      "loss": 3.9746,
+      "step": 1005568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.352737848074148e-05,
+      "loss": 3.9909,
+      "step": 1006080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.351899253323096e-05,
+      "loss": 3.9786,
+      "step": 1006592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.351060658572044e-05,
+      "loss": 3.977,
+      "step": 1007104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.350222063820991e-05,
+      "loss": 3.9828,
+      "step": 1007616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.349383469069939e-05,
+      "loss": 3.9779,
+      "step": 1008128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3485465121992606e-05,
+      "loss": 3.9933,
+      "step": 1008640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.347707917448209e-05,
+      "loss": 3.9976,
+      "step": 1009152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.34687096057753e-05,
+      "loss": 3.9792,
+      "step": 1009664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.346032365826478e-05,
+      "loss": 3.9853,
+      "step": 1010176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.345195408955799e-05,
+      "loss": 3.9943,
+      "step": 1010688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.344356814204747e-05,
+      "loss": 3.9641,
+      "step": 1011200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.343518219453695e-05,
+      "loss": 3.9876,
+      "step": 1011712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.342679624702643e-05,
+      "loss": 3.9739,
+      "step": 1012224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.341841029951591e-05,
+      "loss": 3.973,
+      "step": 1012736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3410024352005384e-05,
+      "loss": 3.9885,
+      "step": 1013248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3401638404494864e-05,
+      "loss": 3.9866,
+      "step": 1013760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3393252456984344e-05,
+      "loss": 3.9832,
+      "step": 1014272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.338486650947383e-05,
+      "loss": 3.9798,
+      "step": 1014784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.337649694076704e-05,
+      "loss": 3.9842,
+      "step": 1015296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3368127372060256e-05,
+      "loss": 3.9756,
+      "step": 1015808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3359741424549736e-05,
+      "loss": 3.9877,
+      "step": 1016320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3351355477039216e-05,
+      "loss": 3.9607,
+      "step": 1016832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.334296952952869e-05,
+      "loss": 3.9833,
+      "step": 1017344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.333458358201817e-05,
+      "loss": 3.9824,
+      "step": 1017856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.332619763450765e-05,
+      "loss": 3.9906,
+      "step": 1018368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.331781168699713e-05,
+      "loss": 3.9732,
+      "step": 1018880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.330942573948661e-05,
+      "loss": 3.9784,
+      "step": 1019392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.330103979197609e-05,
+      "loss": 3.9711,
+      "step": 1019904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.329265384446557e-05,
+      "loss": 3.9728,
+      "step": 1020416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.328426789695505e-05,
+      "loss": 3.9937,
+      "step": 1020928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.327588194944453e-05,
+      "loss": 3.956,
+      "step": 1021440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3267512380737745e-05,
+      "loss": 3.9902,
+      "step": 1021952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3259126433227224e-05,
+      "loss": 3.9855,
+      "step": 1022464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3250740485716704e-05,
+      "loss": 3.9641,
+      "step": 1022976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3242354538206184e-05,
+      "loss": 3.9922,
+      "step": 1023488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3233968590695664e-05,
+      "loss": 3.9636,
+      "step": 1024000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3225599021988873e-05,
+      "loss": 3.9677,
+      "step": 1024512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.321722945328208e-05,
+      "loss": 3.9834,
+      "step": 1025024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320884350577156e-05,
+      "loss": 3.9852,
+      "step": 1025536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320045755826104e-05,
+      "loss": 3.9661,
+      "step": 1026048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.319207161075052e-05,
+      "loss": 3.9898,
+      "step": 1026560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.318368566324e-05,
+      "loss": 3.9596,
+      "step": 1027072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.317529971572948e-05,
+      "loss": 3.9756,
+      "step": 1027584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.31669301470227e-05,
+      "loss": 3.9924,
+      "step": 1028096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.315854419951218e-05,
+      "loss": 3.9829,
+      "step": 1028608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.315015825200166e-05,
+      "loss": 3.9837,
+      "step": 1029120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.314177230449114e-05,
+      "loss": 3.9856,
+      "step": 1029632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.313338635698062e-05,
+      "loss": 3.9952,
+      "step": 1030144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.31250004094701e-05,
+      "loss": 3.9759,
+      "step": 1030656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.311661446195958e-05,
+      "loss": 3.985,
+      "step": 1031168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.310824489325279e-05,
+      "loss": 3.9703,
+      "step": 1031680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.309985894574227e-05,
+      "loss": 3.9675,
+      "step": 1032192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.309147299823175e-05,
+      "loss": 3.9777,
+      "step": 1032704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.308308705072122e-05,
+      "loss": 3.9779,
+      "step": 1033216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.307470110321071e-05,
+      "loss": 3.9835,
+      "step": 1033728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.306631515570019e-05,
+      "loss": 3.9996,
+      "step": 1034240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.305792920818967e-05,
+      "loss": 3.9765,
+      "step": 1034752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.304954326067915e-05,
+      "loss": 3.9721,
+      "step": 1035264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3041173691972356e-05,
+      "loss": 3.9763,
+      "step": 1035776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3032787744461836e-05,
+      "loss": 3.9857,
+      "step": 1036288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3024401796951316e-05,
+      "loss": 3.9755,
+      "step": 1036800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3016015849440796e-05,
+      "loss": 3.9747,
+      "step": 1037312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3007646280734005e-05,
+      "loss": 3.9755,
+      "step": 1037824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2999260333223485e-05,
+      "loss": 3.9741,
+      "step": 1038336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2990874385712965e-05,
+      "loss": 3.979,
+      "step": 1038848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2982488438202445e-05,
+      "loss": 3.9713,
+      "step": 1039360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2974102490691925e-05,
+      "loss": 3.9845,
+      "step": 1039872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.296573292198514e-05,
+      "loss": 3.985,
+      "step": 1040384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.295734697447462e-05,
+      "loss": 3.972,
+      "step": 1040896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.29489610269641e-05,
+      "loss": 3.9729,
+      "step": 1041408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.294059145825731e-05,
+      "loss": 3.9718,
+      "step": 1041920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.293220551074679e-05,
+      "loss": 3.9768,
+      "step": 1042432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.292381956323627e-05,
+      "loss": 3.9857,
+      "step": 1042944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.291543361572575e-05,
+      "loss": 3.9786,
+      "step": 1043456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.290704766821523e-05,
+      "loss": 3.9671,
+      "step": 1043968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.289866172070471e-05,
+      "loss": 3.9827,
+      "step": 1044480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.289027577319419e-05,
+      "loss": 3.9756,
+      "step": 1044992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.288188982568367e-05,
+      "loss": 3.9713,
+      "step": 1045504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.287350387817315e-05,
+      "loss": 3.9741,
+      "step": 1046016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.286513430946636e-05,
+      "loss": 3.965,
+      "step": 1046528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.285674836195584e-05,
+      "loss": 3.9794,
+      "step": 1047040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2848362414445325e-05,
+      "loss": 3.9746,
+      "step": 1047552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2839976466934805e-05,
+      "loss": 3.9686,
+      "step": 1048064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.283162327703174e-05,
+      "loss": 3.9617,
+      "step": 1048576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.282323732952122e-05,
+      "loss": 3.9746,
+      "step": 1049088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.28148513820107e-05,
+      "loss": 3.9811,
+      "step": 1049600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.280646543450018e-05,
+      "loss": 3.9689,
+      "step": 1050112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.279809586579339e-05,
+      "loss": 3.9782,
+      "step": 1050624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.278970991828287e-05,
+      "loss": 3.9823,
+      "step": 1051136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.278132397077235e-05,
+      "loss": 3.9855,
+      "step": 1051648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.277293802326183e-05,
+      "loss": 3.9789,
+      "step": 1052160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.276455207575131e-05,
+      "loss": 3.965,
+      "step": 1052672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.275616612824079e-05,
+      "loss": 3.9736,
+      "step": 1053184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.274778018073028e-05,
+      "loss": 3.9869,
+      "step": 1053696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.273941061202349e-05,
+      "loss": 3.9697,
+      "step": 1054208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.273102466451297e-05,
+      "loss": 3.97,
+      "step": 1054720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.272263871700245e-05,
+      "loss": 3.9728,
+      "step": 1055232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.271425276949193e-05,
+      "loss": 3.9774,
+      "step": 1055744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.270586682198141e-05,
+      "loss": 3.9737,
+      "step": 1056256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.269748087447088e-05,
+      "loss": 3.9792,
+      "step": 1056768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.268909492696036e-05,
+      "loss": 3.9645,
+      "step": 1057280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.268070897944984e-05,
+      "loss": 3.9722,
+      "step": 1057792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.267232303193932e-05,
+      "loss": 3.9871,
+      "step": 1058304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.26639370844288e-05,
+      "loss": 3.9775,
+      "step": 1058816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.265555113691828e-05,
+      "loss": 3.9792,
+      "step": 1059328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.264716518940776e-05,
+      "loss": 3.9655,
+      "step": 1059840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2638795620700976e-05,
+      "loss": 3.9837,
+      "step": 1060352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2630409673190456e-05,
+      "loss": 3.9785,
+      "step": 1060864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2622023725679936e-05,
+      "loss": 3.9775,
+      "step": 1061376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2613637778169416e-05,
+      "loss": 3.9773,
+      "step": 1061888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2605251830658896e-05,
+      "loss": 3.9742,
+      "step": 1062400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2596865883148376e-05,
+      "loss": 3.9857,
+      "step": 1062912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2588479935637856e-05,
+      "loss": 3.9813,
+      "step": 1063424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2580093988127336e-05,
+      "loss": 3.9801,
+      "step": 1063936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2571724419420545e-05,
+      "loss": 3.9711,
+      "step": 1064448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2563338471910025e-05,
+      "loss": 3.9828,
+      "step": 1064960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2554952524399505e-05,
+      "loss": 3.9713,
+      "step": 1065472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2546566576888985e-05,
+      "loss": 3.97,
+      "step": 1065984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2538180629378465e-05,
+      "loss": 3.9841,
+      "step": 1066496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.252982743947541e-05,
+      "loss": 3.9698,
+      "step": 1067008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.252144149196489e-05,
+      "loss": 3.9832,
+      "step": 1067520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.251305554445437e-05,
+      "loss": 3.976,
+      "step": 1068032
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.029815673828125,
+      "eval_runtime": 301.4279,
+      "eval_samples_per_second": 1265.945,
+      "eval_steps_per_second": 39.562,
+      "step": 1068480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.250466959694385e-05,
+      "loss": 3.9599,
+      "step": 1068544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.249628364943333e-05,
+      "loss": 3.9704,
+      "step": 1069056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.248789770192281e-05,
+      "loss": 3.9729,
+      "step": 1069568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.247951175441229e-05,
+      "loss": 3.9755,
+      "step": 1070080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.247112580690177e-05,
+      "loss": 3.9813,
+      "step": 1070592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.246273985939124e-05,
+      "loss": 3.978,
+      "step": 1071104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.245435391188072e-05,
+      "loss": 3.9718,
+      "step": 1071616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.24459679643702e-05,
+      "loss": 3.9721,
+      "step": 1072128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.243758201685968e-05,
+      "loss": 3.9715,
+      "step": 1072640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.242919606934916e-05,
+      "loss": 3.9823,
+      "step": 1073152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.242081012183865e-05,
+      "loss": 3.9703,
+      "step": 1073664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.241242417432813e-05,
+      "loss": 3.9712,
+      "step": 1074176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.240405460562134e-05,
+      "loss": 3.9823,
+      "step": 1074688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.239566865811082e-05,
+      "loss": 3.9616,
+      "step": 1075200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.23872827106003e-05,
+      "loss": 3.9675,
+      "step": 1075712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.237889676308978e-05,
+      "loss": 3.9648,
+      "step": 1076224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.237051081557926e-05,
+      "loss": 3.9722,
+      "step": 1076736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.236212486806874e-05,
+      "loss": 3.9623,
+      "step": 1077248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.235373892055822e-05,
+      "loss": 3.9629,
+      "step": 1077760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.23453529730477e-05,
+      "loss": 3.9684,
+      "step": 1078272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.233696702553718e-05,
+      "loss": 3.9794,
+      "step": 1078784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.232858107802666e-05,
+      "loss": 3.9747,
+      "step": 1079296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.232019513051613e-05,
+      "loss": 3.9668,
+      "step": 1079808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.231180918300562e-05,
+      "loss": 3.9777,
+      "step": 1080320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.23034232354951e-05,
+      "loss": 3.9809,
+      "step": 1080832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2295053666788314e-05,
+      "loss": 3.9654,
+      "step": 1081344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.228668409808152e-05,
+      "loss": 3.9585,
+      "step": 1081856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2278298150571e-05,
+      "loss": 3.9772,
+      "step": 1082368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.226991220306048e-05,
+      "loss": 3.962,
+      "step": 1082880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2261526255549956e-05,
+      "loss": 3.9657,
+      "step": 1083392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2253140308039436e-05,
+      "loss": 3.9669,
+      "step": 1083904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.224477073933265e-05,
+      "loss": 3.9671,
+      "step": 1084416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.223638479182213e-05,
+      "loss": 3.9756,
+      "step": 1084928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2227998844311605e-05,
+      "loss": 3.9845,
+      "step": 1085440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2219612896801085e-05,
+      "loss": 3.9691,
+      "step": 1085952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.221122694929057e-05,
+      "loss": 3.9709,
+      "step": 1086464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.220284100178005e-05,
+      "loss": 3.9795,
+      "step": 1086976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.219447143307326e-05,
+      "loss": 3.9519,
+      "step": 1087488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.218608548556274e-05,
+      "loss": 3.9749,
+      "step": 1088000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.217769953805222e-05,
+      "loss": 3.9589,
+      "step": 1088512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.21693135905417e-05,
+      "loss": 3.9617,
+      "step": 1089024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.216092764303118e-05,
+      "loss": 3.9712,
+      "step": 1089536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.215254169552066e-05,
+      "loss": 3.9766,
+      "step": 1090048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.214415574801014e-05,
+      "loss": 3.9648,
+      "step": 1090560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.213578617930335e-05,
+      "loss": 3.9682,
+      "step": 1091072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.212740023179283e-05,
+      "loss": 3.9698,
+      "step": 1091584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.211901428428231e-05,
+      "loss": 3.9596,
+      "step": 1092096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2110644715575525e-05,
+      "loss": 3.9758,
+      "step": 1092608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2102258768065005e-05,
+      "loss": 3.9502,
+      "step": 1093120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2093872820554485e-05,
+      "loss": 3.9694,
+      "step": 1093632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2085486873043965e-05,
+      "loss": 3.9702,
+      "step": 1094144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2077100925533445e-05,
+      "loss": 3.9781,
+      "step": 1094656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2068714978022925e-05,
+      "loss": 3.9589,
+      "step": 1095168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2060329030512405e-05,
+      "loss": 3.9603,
+      "step": 1095680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2051943083001885e-05,
+      "loss": 3.959,
+      "step": 1096192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2043573514295094e-05,
+      "loss": 3.9564,
+      "step": 1096704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.20352039455883e-05,
+      "loss": 3.9828,
+      "step": 1097216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.202681799807778e-05,
+      "loss": 3.9487,
+      "step": 1097728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.201843205056726e-05,
+      "loss": 3.9728,
+      "step": 1098240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.201004610305674e-05,
+      "loss": 3.971,
+      "step": 1098752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.200166015554622e-05,
+      "loss": 3.9555,
+      "step": 1099264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.19932742080357e-05,
+      "loss": 3.9756,
+      "step": 1099776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.198488826052519e-05,
+      "loss": 3.9524,
+      "step": 1100288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.197650231301467e-05,
+      "loss": 3.9558,
+      "step": 1100800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.196813274430788e-05,
+      "loss": 3.9683,
+      "step": 1101312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.195974679679736e-05,
+      "loss": 3.9735,
+      "step": 1101824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.195136084928684e-05,
+      "loss": 3.9542,
+      "step": 1102336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.194297490177632e-05,
+      "loss": 3.9736,
+      "step": 1102848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.193458895426579e-05,
+      "loss": 3.9453,
+      "step": 1103360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.192621938555901e-05,
+      "loss": 3.9621,
+      "step": 1103872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.191783343804849e-05,
+      "loss": 3.9797,
+      "step": 1104384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.190944749053797e-05,
+      "loss": 3.9718,
+      "step": 1104896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.190106154302744e-05,
+      "loss": 3.9665,
+      "step": 1105408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1892691974320657e-05,
+      "loss": 3.9716,
+      "step": 1105920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.188430602681014e-05,
+      "loss": 3.9805,
+      "step": 1106432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.187593645810335e-05,
+      "loss": 3.9643,
+      "step": 1106944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.186755051059283e-05,
+      "loss": 3.9772,
+      "step": 1107456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.185916456308231e-05,
+      "loss": 3.9563,
+      "step": 1107968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.185077861557179e-05,
+      "loss": 3.9539,
+      "step": 1108480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1842392668061265e-05,
+      "loss": 3.9624,
+      "step": 1108992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.183402309935448e-05,
+      "loss": 3.966,
+      "step": 1109504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.182563715184396e-05,
+      "loss": 3.9664,
+      "step": 1110016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.181725120433344e-05,
+      "loss": 3.987,
+      "step": 1110528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1808865256822914e-05,
+      "loss": 3.9673,
+      "step": 1111040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1800479309312394e-05,
+      "loss": 3.9566,
+      "step": 1111552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.179210974060561e-05,
+      "loss": 3.965,
+      "step": 1112064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.178372379309509e-05,
+      "loss": 3.9706,
+      "step": 1112576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.177533784558457e-05,
+      "loss": 3.9615,
+      "step": 1113088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.176695189807405e-05,
+      "loss": 3.9609,
+      "step": 1113600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.175856595056353e-05,
+      "loss": 3.965,
+      "step": 1114112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.175018000305301e-05,
+      "loss": 3.9603,
+      "step": 1114624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.174181043434622e-05,
+      "loss": 3.9705,
+      "step": 1115136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.17334244868357e-05,
+      "loss": 3.9552,
+      "step": 1115648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.172503853932518e-05,
+      "loss": 3.9698,
+      "step": 1116160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.171666897061839e-05,
+      "loss": 3.9729,
+      "step": 1116672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.170828302310787e-05,
+      "loss": 3.9636,
+      "step": 1117184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.169989707559735e-05,
+      "loss": 3.9563,
+      "step": 1117696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1691511128086835e-05,
+      "loss": 3.9619,
+      "step": 1118208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1683125180576315e-05,
+      "loss": 3.9602,
+      "step": 1118720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1674739233065795e-05,
+      "loss": 3.9708,
+      "step": 1119232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1666353285555275e-05,
+      "loss": 3.9664,
+      "step": 1119744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1657967338044755e-05,
+      "loss": 3.9547,
+      "step": 1120256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1649581390534234e-05,
+      "loss": 3.9658,
+      "step": 1120768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1641195443023714e-05,
+      "loss": 3.9626,
+      "step": 1121280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1632809495513194e-05,
+      "loss": 3.9623,
+      "step": 1121792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1624423548002674e-05,
+      "loss": 3.9605,
+      "step": 1122304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1616053979295883e-05,
+      "loss": 3.9548,
+      "step": 1122816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1607668031785363e-05,
+      "loss": 3.9641,
+      "step": 1123328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.159928208427484e-05,
+      "loss": 3.9608,
+      "step": 1123840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.159091251556805e-05,
+      "loss": 3.9539,
+      "step": 1124352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.158252656805753e-05,
+      "loss": 3.9505,
+      "step": 1124864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.157414062054702e-05,
+      "loss": 3.9569,
+      "step": 1125376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.15657546730365e-05,
+      "loss": 3.9732,
+      "step": 1125888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.155736872552598e-05,
+      "loss": 3.9562,
+      "step": 1126400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.154898277801545e-05,
+      "loss": 3.9657,
+      "step": 1126912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.154059683050493e-05,
+      "loss": 3.962,
+      "step": 1127424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.153221088299441e-05,
+      "loss": 3.9811,
+      "step": 1127936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.152382493548389e-05,
+      "loss": 3.9639,
+      "step": 1128448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.15154553667771e-05,
+      "loss": 3.9518,
+      "step": 1128960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.150706941926658e-05,
+      "loss": 3.9637,
+      "step": 1129472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.149868347175606e-05,
+      "loss": 3.9696,
+      "step": 1129984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.149031390304928e-05,
+      "loss": 3.9563,
+      "step": 1130496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1481944334342486e-05,
+      "loss": 3.9586,
+      "step": 1131008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.147355838683197e-05,
+      "loss": 3.9602,
+      "step": 1131520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.146517243932145e-05,
+      "loss": 3.9653,
+      "step": 1132032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1456786491810926e-05,
+      "loss": 3.9561,
+      "step": 1132544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1448400544300406e-05,
+      "loss": 3.9651,
+      "step": 1133056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1440014596789886e-05,
+      "loss": 3.9584,
+      "step": 1133568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1431628649279366e-05,
+      "loss": 3.9614,
+      "step": 1134080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1423242701768846e-05,
+      "loss": 3.9715,
+      "step": 1134592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1414856754258326e-05,
+      "loss": 3.9659,
+      "step": 1135104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1406470806747806e-05,
+      "loss": 3.9636,
+      "step": 1135616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1398084859237286e-05,
+      "loss": 3.9558,
+      "step": 1136128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1389698911726766e-05,
+      "loss": 3.9693,
+      "step": 1136640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1381312964216246e-05,
+      "loss": 3.9683,
+      "step": 1137152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1372943395509455e-05,
+      "loss": 3.963,
+      "step": 1137664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1364557447998935e-05,
+      "loss": 3.9684,
+      "step": 1138176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.135618787929215e-05,
+      "loss": 3.9582,
+      "step": 1138688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.134780193178163e-05,
+      "loss": 3.973,
+      "step": 1139200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.133941598427111e-05,
+      "loss": 3.9682,
+      "step": 1139712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.133103003676059e-05,
+      "loss": 3.9672,
+      "step": 1140224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.132264408925007e-05,
+      "loss": 3.9635,
+      "step": 1140736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.131425814173955e-05,
+      "loss": 3.9682,
+      "step": 1141248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.130587219422903e-05,
+      "loss": 3.9569,
+      "step": 1141760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.129748624671851e-05,
+      "loss": 3.9585,
+      "step": 1142272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.128910029920799e-05,
+      "loss": 3.974,
+      "step": 1142784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.128071435169746e-05,
+      "loss": 3.9586,
+      "step": 1143296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.127234478299068e-05,
+      "loss": 3.9682,
+      "step": 1143808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.126397521428389e-05,
+      "loss": 3.9673,
+      "step": 1144320
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.0232462882995605,
+      "eval_runtime": 289.8657,
+      "eval_samples_per_second": 1316.441,
+      "eval_steps_per_second": 41.14,
+      "step": 1144800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1255589266773375e-05,
+      "loss": 3.9543,
+      "step": 1144832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1247203319262855e-05,
+      "loss": 3.9585,
+      "step": 1145344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1238817371752335e-05,
+      "loss": 3.9584,
+      "step": 1145856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1230431424241815e-05,
+      "loss": 3.9634,
+      "step": 1146368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1222061855535024e-05,
+      "loss": 3.9653,
+      "step": 1146880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1213675908024504e-05,
+      "loss": 3.969,
+      "step": 1147392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1205289960513984e-05,
+      "loss": 3.9603,
+      "step": 1147904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1196904013003464e-05,
+      "loss": 3.9592,
+      "step": 1148416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.118851806549294e-05,
+      "loss": 3.9603,
+      "step": 1148928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.118013211798242e-05,
+      "loss": 3.9713,
+      "step": 1149440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.11717461704719e-05,
+      "loss": 3.9537,
+      "step": 1149952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.116336022296138e-05,
+      "loss": 3.9626,
+      "step": 1150464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.115500703305833e-05,
+      "loss": 3.9666,
+      "step": 1150976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.114662108554781e-05,
+      "loss": 3.9548,
+      "step": 1151488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.113823513803729e-05,
+      "loss": 3.9555,
+      "step": 1152000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.112984919052676e-05,
+      "loss": 3.9549,
+      "step": 1152512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.112146324301624e-05,
+      "loss": 3.952,
+      "step": 1153024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.111307729550572e-05,
+      "loss": 3.9545,
+      "step": 1153536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.11046913479952e-05,
+      "loss": 3.9511,
+      "step": 1154048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.109630540048468e-05,
+      "loss": 3.9536,
+      "step": 1154560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.108791945297416e-05,
+      "loss": 3.969,
+      "step": 1155072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.107953350546364e-05,
+      "loss": 3.9603,
+      "step": 1155584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.107114755795312e-05,
+      "loss": 3.9565,
+      "step": 1156096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.10627616104426e-05,
+      "loss": 3.9651,
+      "step": 1156608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.105437566293208e-05,
+      "loss": 3.9712,
+      "step": 1157120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.10460060942253e-05,
+      "loss": 3.9498,
+      "step": 1157632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1037636525518506e-05,
+      "loss": 3.9498,
+      "step": 1158144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1029250578007986e-05,
+      "loss": 3.9636,
+      "step": 1158656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1020864630497466e-05,
+      "loss": 3.9531,
+      "step": 1159168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1012478682986946e-05,
+      "loss": 3.9533,
+      "step": 1159680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1004092735476426e-05,
+      "loss": 3.952,
+      "step": 1160192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0995706787965906e-05,
+      "loss": 3.9589,
+      "step": 1160704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0987320840455386e-05,
+      "loss": 3.9659,
+      "step": 1161216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0978934892944866e-05,
+      "loss": 3.9682,
+      "step": 1161728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0970565324238075e-05,
+      "loss": 3.9609,
+      "step": 1162240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0962179376727555e-05,
+      "loss": 3.9567,
+      "step": 1162752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0953793429217035e-05,
+      "loss": 3.9658,
+      "step": 1163264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.094542386051025e-05,
+      "loss": 3.9432,
+      "step": 1163776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.093703791299973e-05,
+      "loss": 3.9664,
+      "step": 1164288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.092865196548921e-05,
+      "loss": 3.9449,
+      "step": 1164800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.092026601797869e-05,
+      "loss": 3.9511,
+      "step": 1165312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.09118964492719e-05,
+      "loss": 3.9584,
+      "step": 1165824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.090351050176138e-05,
+      "loss": 3.9628,
+      "step": 1166336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.089512455425086e-05,
+      "loss": 3.9552,
+      "step": 1166848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.088673860674034e-05,
+      "loss": 3.9569,
+      "step": 1167360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.087835265922982e-05,
+      "loss": 3.9566,
+      "step": 1167872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.08699667117193e-05,
+      "loss": 3.9511,
+      "step": 1168384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.086158076420877e-05,
+      "loss": 3.9649,
+      "step": 1168896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.085319481669825e-05,
+      "loss": 3.9336,
+      "step": 1169408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.084480886918773e-05,
+      "loss": 3.9505,
+      "step": 1169920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.083643930048095e-05,
+      "loss": 3.9676,
+      "step": 1170432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.082805335297043e-05,
+      "loss": 3.961,
+      "step": 1170944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.081966740545991e-05,
+      "loss": 3.9465,
+      "step": 1171456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.081128145794939e-05,
+      "loss": 3.9533,
+      "step": 1171968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.08029118892426e-05,
+      "loss": 3.9489,
+      "step": 1172480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.079452594173208e-05,
+      "loss": 3.9429,
+      "step": 1172992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0786156373025293e-05,
+      "loss": 3.9717,
+      "step": 1173504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.077777042551477e-05,
+      "loss": 3.9365,
+      "step": 1174016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0769384478004247e-05,
+      "loss": 3.9608,
+      "step": 1174528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0760998530493726e-05,
+      "loss": 3.9555,
+      "step": 1175040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0752612582983206e-05,
+      "loss": 3.9489,
+      "step": 1175552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0744226635472686e-05,
+      "loss": 3.9603,
+      "step": 1176064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0735840687962166e-05,
+      "loss": 3.9435,
+      "step": 1176576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.072745474045165e-05,
+      "loss": 3.9469,
+      "step": 1177088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.071908517174486e-05,
+      "loss": 3.9519,
+      "step": 1177600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.071069922423434e-05,
+      "loss": 3.9661,
+      "step": 1178112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.070231327672382e-05,
+      "loss": 3.941,
+      "step": 1178624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.06939273292133e-05,
+      "loss": 3.9615,
+      "step": 1179136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.068554138170278e-05,
+      "loss": 3.9379,
+      "step": 1179648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.067715543419226e-05,
+      "loss": 3.9478,
+      "step": 1180160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.066878586548547e-05,
+      "loss": 3.9679,
+      "step": 1180672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.066039991797495e-05,
+      "loss": 3.9646,
+      "step": 1181184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.065201397046443e-05,
+      "loss": 3.9548,
+      "step": 1181696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.064362802295391e-05,
+      "loss": 3.961,
+      "step": 1182208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.063524207544339e-05,
+      "loss": 3.9678,
+      "step": 1182720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.062687250673661e-05,
+      "loss": 3.9511,
+      "step": 1183232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.061848655922609e-05,
+      "loss": 3.9626,
+      "step": 1183744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.061010061171557e-05,
+      "loss": 3.9477,
+      "step": 1184256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.060171466420505e-05,
+      "loss": 3.945,
+      "step": 1184768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0593328716694527e-05,
+      "loss": 3.9473,
+      "step": 1185280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0584975526791465e-05,
+      "loss": 3.9523,
+      "step": 1185792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0576589579280945e-05,
+      "loss": 3.9589,
+      "step": 1186304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0568203631770425e-05,
+      "loss": 3.9707,
+      "step": 1186816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0559817684259905e-05,
+      "loss": 3.9638,
+      "step": 1187328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0551431736749385e-05,
+      "loss": 3.9443,
+      "step": 1187840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0543045789238865e-05,
+      "loss": 3.9502,
+      "step": 1188352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0534659841728345e-05,
+      "loss": 3.9604,
+      "step": 1188864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0526273894217824e-05,
+      "loss": 3.9526,
+      "step": 1189376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.051790432551104e-05,
+      "loss": 3.9498,
+      "step": 1189888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0509518378000517e-05,
+      "loss": 3.9542,
+      "step": 1190400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0501132430489997e-05,
+      "loss": 3.9465,
+      "step": 1190912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0492746482979477e-05,
+      "loss": 3.9552,
+      "step": 1191424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.048437691427269e-05,
+      "loss": 3.9459,
+      "step": 1191936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.047599096676217e-05,
+      "loss": 3.9583,
+      "step": 1192448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.046760501925165e-05,
+      "loss": 3.958,
+      "step": 1192960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.045921907174113e-05,
+      "loss": 3.9549,
+      "step": 1193472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0450833124230606e-05,
+      "loss": 3.9465,
+      "step": 1193984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0442447176720086e-05,
+      "loss": 3.9518,
+      "step": 1194496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0434061229209566e-05,
+      "loss": 3.9503,
+      "step": 1195008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0425675281699046e-05,
+      "loss": 3.9594,
+      "step": 1195520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0417289334188526e-05,
+      "loss": 3.9553,
+      "step": 1196032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0408919765481735e-05,
+      "loss": 3.9422,
+      "step": 1196544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0400533817971215e-05,
+      "loss": 3.9526,
+      "step": 1197056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0392147870460698e-05,
+      "loss": 3.9512,
+      "step": 1197568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0383761922950178e-05,
+      "loss": 3.9526,
+      "step": 1198080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0375408733047123e-05,
+      "loss": 3.9487,
+      "step": 1198592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0367022785536603e-05,
+      "loss": 3.9428,
+      "step": 1199104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0358636838026076e-05,
+      "loss": 3.9554,
+      "step": 1199616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0350267269319292e-05,
+      "loss": 3.9521,
+      "step": 1200128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0341881321808775e-05,
+      "loss": 3.9405,
+      "step": 1200640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0333495374298255e-05,
+      "loss": 3.9415,
+      "step": 1201152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.032510942678773e-05,
+      "loss": 3.943,
+      "step": 1201664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.031672347927721e-05,
+      "loss": 3.9593,
+      "step": 1202176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.030833753176669e-05,
+      "loss": 3.9471,
+      "step": 1202688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.029995158425617e-05,
+      "loss": 3.9517,
+      "step": 1203200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.029156563674565e-05,
+      "loss": 3.9557,
+      "step": 1203712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.028317968923513e-05,
+      "loss": 3.968,
+      "step": 1204224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.027479374172461e-05,
+      "loss": 3.952,
+      "step": 1204736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.026640779421409e-05,
+      "loss": 3.9395,
+      "step": 1205248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.02580382255073e-05,
+      "loss": 3.954,
+      "step": 1205760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.024965227799678e-05,
+      "loss": 3.9549,
+      "step": 1206272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.024126633048626e-05,
+      "loss": 3.9468,
+      "step": 1206784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0232896761779473e-05,
+      "loss": 3.9469,
+      "step": 1207296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0224510814268953e-05,
+      "loss": 3.952,
+      "step": 1207808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0216124866758433e-05,
+      "loss": 3.9545,
+      "step": 1208320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0207738919247913e-05,
+      "loss": 3.9471,
+      "step": 1208832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0199352971737393e-05,
+      "loss": 3.9544,
+      "step": 1209344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0190967024226873e-05,
+      "loss": 3.9447,
+      "step": 1209856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0182581076716353e-05,
+      "loss": 3.9516,
+      "step": 1210368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0174195129205836e-05,
+      "loss": 3.9503,
+      "step": 1210880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0165809181695316e-05,
+      "loss": 3.9603,
+      "step": 1211392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0157423234184796e-05,
+      "loss": 3.9545,
+      "step": 1211904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0149053665478005e-05,
+      "loss": 3.9425,
+      "step": 1212416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0140684096771214e-05,
+      "loss": 3.9566,
+      "step": 1212928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0132298149260698e-05,
+      "loss": 3.9581,
+      "step": 1213440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0123912201750177e-05,
+      "loss": 3.9496,
+      "step": 1213952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0115526254239657e-05,
+      "loss": 3.9593,
+      "step": 1214464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0107140306729137e-05,
+      "loss": 3.949,
+      "step": 1214976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0098754359218617e-05,
+      "loss": 3.9628,
+      "step": 1215488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.009036841170809e-05,
+      "loss": 3.9549,
+      "step": 1216000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0081982464197574e-05,
+      "loss": 3.958,
+      "step": 1216512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0073596516687054e-05,
+      "loss": 3.948,
+      "step": 1217024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0065210569176534e-05,
+      "loss": 3.96,
+      "step": 1217536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0056824621666014e-05,
+      "loss": 3.9474,
+      "step": 1218048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0048455052959223e-05,
+      "loss": 3.9446,
+      "step": 1218560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0040069105448703e-05,
+      "loss": 3.9636,
+      "step": 1219072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0031683157938183e-05,
+      "loss": 3.9505,
+      "step": 1219584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0023297210427666e-05,
+      "loss": 3.9542,
+      "step": 1220096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0014927641720875e-05,
+      "loss": 3.9574,
+      "step": 1220608
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.0006541694210355e-05,
+      "loss": 3.9437,
+      "step": 1221120
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.017386436462402,
+      "eval_runtime": 290.0631,
+      "eval_samples_per_second": 1315.545,
+      "eval_steps_per_second": 41.112,
+      "step": 1221120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9998155746699835e-05,
+      "loss": 3.9466,
+      "step": 1221632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9989769799189315e-05,
+      "loss": 3.9457,
+      "step": 1222144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9981383851678795e-05,
+      "loss": 3.9532,
+      "step": 1222656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9972997904168275e-05,
+      "loss": 3.9544,
+      "step": 1223168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9964611956657755e-05,
+      "loss": 3.9595,
+      "step": 1223680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9956226009147238e-05,
+      "loss": 3.9499,
+      "step": 1224192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9947840061636718e-05,
+      "loss": 3.9506,
+      "step": 1224704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9939454114126198e-05,
+      "loss": 3.9454,
+      "step": 1225216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9931068166615678e-05,
+      "loss": 3.9631,
+      "step": 1225728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.992268221910515e-05,
+      "loss": 3.9422,
+      "step": 1226240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.991429627159463e-05,
+      "loss": 3.9514,
+      "step": 1226752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9905926702887847e-05,
+      "loss": 3.956,
+      "step": 1227264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.989754075537733e-05,
+      "loss": 3.9473,
+      "step": 1227776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9889154807866804e-05,
+      "loss": 3.9413,
+      "step": 1228288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9880768860356284e-05,
+      "loss": 3.9411,
+      "step": 1228800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9872382912845764e-05,
+      "loss": 3.9397,
+      "step": 1229312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9863996965335244e-05,
+      "loss": 3.9428,
+      "step": 1229824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9855611017824723e-05,
+      "loss": 3.9421,
+      "step": 1230336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9847225070314207e-05,
+      "loss": 3.9392,
+      "step": 1230848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9838839122803687e-05,
+      "loss": 3.9619,
+      "step": 1231360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9830469554096896e-05,
+      "loss": 3.9489,
+      "step": 1231872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9822083606586376e-05,
+      "loss": 3.9456,
+      "step": 1232384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9813714037879585e-05,
+      "loss": 3.9526,
+      "step": 1232896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9805328090369068e-05,
+      "loss": 3.9621,
+      "step": 1233408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9796942142858548e-05,
+      "loss": 3.9391,
+      "step": 1233920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9788556195348028e-05,
+      "loss": 3.9428,
+      "step": 1234432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9780170247837508e-05,
+      "loss": 3.946,
+      "step": 1234944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9771784300326988e-05,
+      "loss": 3.9488,
+      "step": 1235456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9763398352816468e-05,
+      "loss": 3.9418,
+      "step": 1235968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9755012405305948e-05,
+      "loss": 3.9386,
+      "step": 1236480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.974665921540289e-05,
+      "loss": 3.9495,
+      "step": 1236992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.973827326789237e-05,
+      "loss": 3.9547,
+      "step": 1237504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.972988732038185e-05,
+      "loss": 3.9571,
+      "step": 1238016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.972150137287133e-05,
+      "loss": 3.9511,
+      "step": 1238528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.971311542536081e-05,
+      "loss": 3.9474,
+      "step": 1239040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.970472947785029e-05,
+      "loss": 3.9501,
+      "step": 1239552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9696359909143502e-05,
+      "loss": 3.9337,
+      "step": 1240064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9687973961632982e-05,
+      "loss": 3.9535,
+      "step": 1240576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9679588014122462e-05,
+      "loss": 3.9371,
+      "step": 1241088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9671202066611942e-05,
+      "loss": 3.9402,
+      "step": 1241600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9662816119101422e-05,
+      "loss": 3.9434,
+      "step": 1242112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.96544301715909e-05,
+      "loss": 3.9556,
+      "step": 1242624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.964604422408038e-05,
+      "loss": 3.9425,
+      "step": 1243136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.963765827656986e-05,
+      "loss": 3.9443,
+      "step": 1243648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9629272329059345e-05,
+      "loss": 3.9501,
+      "step": 1244160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9620902760352554e-05,
+      "loss": 3.9378,
+      "step": 1244672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9612533191645763e-05,
+      "loss": 3.9543,
+      "step": 1245184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9604147244135243e-05,
+      "loss": 3.9267,
+      "step": 1245696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9595761296624723e-05,
+      "loss": 3.9373,
+      "step": 1246208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9587375349114206e-05,
+      "loss": 3.9534,
+      "step": 1246720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9578989401603686e-05,
+      "loss": 3.9522,
+      "step": 1247232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9570603454093166e-05,
+      "loss": 3.9387,
+      "step": 1247744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.956221750658264e-05,
+      "loss": 3.9407,
+      "step": 1248256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.955383155907212e-05,
+      "loss": 3.9433,
+      "step": 1248768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.95454456115616e-05,
+      "loss": 3.9266,
+      "step": 1249280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9537076042854815e-05,
+      "loss": 3.9618,
+      "step": 1249792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9528690095344292e-05,
+      "loss": 3.9317,
+      "step": 1250304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9520304147833772e-05,
+      "loss": 3.9514,
+      "step": 1250816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9511918200323252e-05,
+      "loss": 3.9454,
+      "step": 1251328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9503532252812732e-05,
+      "loss": 3.9353,
+      "step": 1251840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.949514630530221e-05,
+      "loss": 3.946,
+      "step": 1252352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.948676035779169e-05,
+      "loss": 3.936,
+      "step": 1252864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.947837441028117e-05,
+      "loss": 3.9372,
+      "step": 1253376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9470004841574384e-05,
+      "loss": 3.9359,
+      "step": 1253888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9461618894063864e-05,
+      "loss": 3.9561,
+      "step": 1254400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9453232946553344e-05,
+      "loss": 3.937,
+      "step": 1254912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9444846999042824e-05,
+      "loss": 3.9425,
+      "step": 1255424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9436461051532304e-05,
+      "loss": 3.9271,
+      "step": 1255936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9428091482825516e-05,
+      "loss": 3.9417,
+      "step": 1256448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9419705535314996e-05,
+      "loss": 3.9553,
+      "step": 1256960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9411319587804476e-05,
+      "loss": 3.9508,
+      "step": 1257472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9402950019097685e-05,
+      "loss": 3.9419,
+      "step": 1257984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9394564071587165e-05,
+      "loss": 3.9537,
+      "step": 1258496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9386178124076645e-05,
+      "loss": 3.9555,
+      "step": 1259008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9377792176566125e-05,
+      "loss": 3.9397,
+      "step": 1259520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.936940622905561e-05,
+      "loss": 3.9509,
+      "step": 1260032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.936102028154509e-05,
+      "loss": 3.9402,
+      "step": 1260544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.935263433403457e-05,
+      "loss": 3.9348,
+      "step": 1261056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.934424838652405e-05,
+      "loss": 3.9395,
+      "step": 1261568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9335878817817258e-05,
+      "loss": 3.9396,
+      "step": 1262080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9327492870306737e-05,
+      "loss": 3.9486,
+      "step": 1262592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9319106922796217e-05,
+      "loss": 3.9634,
+      "step": 1263104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.931073735408943e-05,
+      "loss": 3.9529,
+      "step": 1263616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.930235140657891e-05,
+      "loss": 3.9308,
+      "step": 1264128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.929396545906839e-05,
+      "loss": 3.9442,
+      "step": 1264640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.928557951155787e-05,
+      "loss": 3.9467,
+      "step": 1265152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.927719356404735e-05,
+      "loss": 3.9417,
+      "step": 1265664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9268807616536826e-05,
+      "loss": 3.9391,
+      "step": 1266176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9260421669026306e-05,
+      "loss": 3.9392,
+      "step": 1266688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9252052100319522e-05,
+      "loss": 3.9394,
+      "step": 1267200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9243666152809002e-05,
+      "loss": 3.9447,
+      "step": 1267712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9235280205298475e-05,
+      "loss": 3.9376,
+      "step": 1268224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9226894257787955e-05,
+      "loss": 3.9514,
+      "step": 1268736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.921850831027744e-05,
+      "loss": 3.9443,
+      "step": 1269248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9210138741570648e-05,
+      "loss": 3.949,
+      "step": 1269760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9201752794060128e-05,
+      "loss": 3.9321,
+      "step": 1270272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9193366846549608e-05,
+      "loss": 3.9395,
+      "step": 1270784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9184980899039088e-05,
+      "loss": 3.9422,
+      "step": 1271296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9176594951528568e-05,
+      "loss": 3.9469,
+      "step": 1271808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9168209004018047e-05,
+      "loss": 3.9505,
+      "step": 1272320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.915982305650753e-05,
+      "loss": 3.932,
+      "step": 1272832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.915143710899701e-05,
+      "loss": 3.9416,
+      "step": 1273344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.914306754029022e-05,
+      "loss": 3.9363,
+      "step": 1273856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.913469797158343e-05,
+      "loss": 3.9451,
+      "step": 1274368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.912631202407291e-05,
+      "loss": 3.9368,
+      "step": 1274880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9117926076562392e-05,
+      "loss": 3.9351,
+      "step": 1275392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9109540129051872e-05,
+      "loss": 3.9466,
+      "step": 1275904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9101154181541352e-05,
+      "loss": 3.9375,
+      "step": 1276416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9092768234030832e-05,
+      "loss": 3.9343,
+      "step": 1276928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9084382286520312e-05,
+      "loss": 3.9316,
+      "step": 1277440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9075996339009792e-05,
+      "loss": 3.9313,
+      "step": 1277952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9067626770303e-05,
+      "loss": 3.9482,
+      "step": 1278464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9059240822792484e-05,
+      "loss": 3.9365,
+      "step": 1278976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9050854875281964e-05,
+      "loss": 3.9404,
+      "step": 1279488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9042468927771444e-05,
+      "loss": 3.946,
+      "step": 1280000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9034099359064653e-05,
+      "loss": 3.9565,
+      "step": 1280512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9025713411554133e-05,
+      "loss": 3.9402,
+      "step": 1281024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9017327464043613e-05,
+      "loss": 3.9328,
+      "step": 1281536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9008941516533093e-05,
+      "loss": 3.9413,
+      "step": 1282048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9000571947826306e-05,
+      "loss": 3.9468,
+      "step": 1282560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8992186000315786e-05,
+      "loss": 3.9366,
+      "step": 1283072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8983816431608995e-05,
+      "loss": 3.9383,
+      "step": 1283584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8975430484098475e-05,
+      "loss": 3.9393,
+      "step": 1284096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8967044536587955e-05,
+      "loss": 3.9438,
+      "step": 1284608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8958658589077438e-05,
+      "loss": 3.9391,
+      "step": 1285120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8950272641566918e-05,
+      "loss": 3.942,
+      "step": 1285632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8941886694056398e-05,
+      "loss": 3.9361,
+      "step": 1286144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8933517125349607e-05,
+      "loss": 3.9409,
+      "step": 1286656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8925131177839087e-05,
+      "loss": 3.9364,
+      "step": 1287168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8916745230328567e-05,
+      "loss": 3.9561,
+      "step": 1287680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8908359282818047e-05,
+      "loss": 3.9429,
+      "step": 1288192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8899973335307527e-05,
+      "loss": 3.9384,
+      "step": 1288704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.889160376660074e-05,
+      "loss": 3.9383,
+      "step": 1289216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.888321781909022e-05,
+      "loss": 3.9457,
+      "step": 1289728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.88748318715797e-05,
+      "loss": 3.9407,
+      "step": 1290240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.886644592406918e-05,
+      "loss": 3.9504,
+      "step": 1290752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.885805997655866e-05,
+      "loss": 3.9369,
+      "step": 1291264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8849674029048136e-05,
+      "loss": 3.9526,
+      "step": 1291776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8841288081537616e-05,
+      "loss": 3.9431,
+      "step": 1292288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8832902134027096e-05,
+      "loss": 3.9451,
+      "step": 1292800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8824532565320305e-05,
+      "loss": 3.9405,
+      "step": 1293312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.881616299661352e-05,
+      "loss": 3.9518,
+      "step": 1293824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8807777049103e-05,
+      "loss": 3.9415,
+      "step": 1294336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.879939110159248e-05,
+      "loss": 3.931,
+      "step": 1294848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8791005154081957e-05,
+      "loss": 3.9545,
+      "step": 1295360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8782635585375173e-05,
+      "loss": 3.9421,
+      "step": 1295872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8774249637864653e-05,
+      "loss": 3.9419,
+      "step": 1296384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8765863690354133e-05,
+      "loss": 3.9469,
+      "step": 1296896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.875747774284361e-05,
+      "loss": 3.933,
+      "step": 1297408
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.012454032897949,
+      "eval_runtime": 621.4897,
+      "eval_samples_per_second": 613.994,
+      "eval_steps_per_second": 19.188,
+      "step": 1297440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.874909179533309e-05,
+      "loss": 3.935,
+      "step": 1297920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.874070584782257e-05,
+      "loss": 3.9397,
+      "step": 1298432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.873233627911578e-05,
+      "loss": 3.9428,
+      "step": 1298944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.872395033160526e-05,
+      "loss": 3.9414,
+      "step": 1299456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.871556438409474e-05,
+      "loss": 3.95,
+      "step": 1299968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.870717843658422e-05,
+      "loss": 3.9387,
+      "step": 1300480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8698792489073702e-05,
+      "loss": 3.9388,
+      "step": 1300992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8690406541563182e-05,
+      "loss": 3.9369,
+      "step": 1301504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.868203697285639e-05,
+      "loss": 3.953,
+      "step": 1302016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.867365102534587e-05,
+      "loss": 3.9367,
+      "step": 1302528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.866526507783535e-05,
+      "loss": 3.9383,
+      "step": 1303040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.865687913032483e-05,
+      "loss": 3.9426,
+      "step": 1303552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8648509561618043e-05,
+      "loss": 3.9394,
+      "step": 1304064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8640123614107523e-05,
+      "loss": 3.9332,
+      "step": 1304576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8631737666597003e-05,
+      "loss": 3.9369,
+      "step": 1305088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8623351719086483e-05,
+      "loss": 3.9228,
+      "step": 1305600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8614965771575963e-05,
+      "loss": 3.9353,
+      "step": 1306112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8606596202869172e-05,
+      "loss": 3.9311,
+      "step": 1306624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8598210255358655e-05,
+      "loss": 3.9285,
+      "step": 1307136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8589824307848135e-05,
+      "loss": 3.9561,
+      "step": 1307648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8581438360337615e-05,
+      "loss": 3.9342,
+      "step": 1308160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8573052412827095e-05,
+      "loss": 3.9409,
+      "step": 1308672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8564666465316575e-05,
+      "loss": 3.9399,
+      "step": 1309184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8556280517806055e-05,
+      "loss": 3.9511,
+      "step": 1309696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8547894570295535e-05,
+      "loss": 3.9292,
+      "step": 1310208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8539525001588748e-05,
+      "loss": 3.9319,
+      "step": 1310720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8531139054078228e-05,
+      "loss": 3.9362,
+      "step": 1311232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8522753106567708e-05,
+      "loss": 3.938,
+      "step": 1311744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8514367159057188e-05,
+      "loss": 3.9288,
+      "step": 1312256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8505981211546667e-05,
+      "loss": 3.93,
+      "step": 1312768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.849759526403614e-05,
+      "loss": 3.9395,
+      "step": 1313280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8489209316525624e-05,
+      "loss": 3.9423,
+      "step": 1313792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8480823369015104e-05,
+      "loss": 3.9475,
+      "step": 1314304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.847245380030832e-05,
+      "loss": 3.9402,
+      "step": 1314816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8464067852797793e-05,
+      "loss": 3.9348,
+      "step": 1315328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8455681905287273e-05,
+      "loss": 3.9434,
+      "step": 1315840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8447295957776753e-05,
+      "loss": 3.922,
+      "step": 1316352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8438910010266233e-05,
+      "loss": 3.9453,
+      "step": 1316864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8430524062755716e-05,
+      "loss": 3.9254,
+      "step": 1317376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8422138115245196e-05,
+      "loss": 3.9341,
+      "step": 1317888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8413752167734676e-05,
+      "loss": 3.9326,
+      "step": 1318400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8405366220224156e-05,
+      "loss": 3.9439,
+      "step": 1318912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8396980272713636e-05,
+      "loss": 3.9352,
+      "step": 1319424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8388610704006845e-05,
+      "loss": 3.9314,
+      "step": 1319936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8380241135300058e-05,
+      "loss": 3.9397,
+      "step": 1320448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8371855187789538e-05,
+      "loss": 3.9307,
+      "step": 1320960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8363469240279018e-05,
+      "loss": 3.9445,
+      "step": 1321472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8355083292768497e-05,
+      "loss": 3.9187,
+      "step": 1321984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8346697345257977e-05,
+      "loss": 3.9249,
+      "step": 1322496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8338311397747457e-05,
+      "loss": 3.9428,
+      "step": 1323008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8329925450236937e-05,
+      "loss": 3.9438,
+      "step": 1323520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8321539502726417e-05,
+      "loss": 3.9284,
+      "step": 1324032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.831316993401963e-05,
+      "loss": 3.9268,
+      "step": 1324544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.830478398650911e-05,
+      "loss": 3.9348,
+      "step": 1325056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.829641441780232e-05,
+      "loss": 3.9166,
+      "step": 1325568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.82880284702918e-05,
+      "loss": 3.9475,
+      "step": 1326080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.827964252278128e-05,
+      "loss": 3.9276,
+      "step": 1326592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.827125657527076e-05,
+      "loss": 3.9377,
+      "step": 1327104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8262870627760242e-05,
+      "loss": 3.9341,
+      "step": 1327616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8254484680249722e-05,
+      "loss": 3.9246,
+      "step": 1328128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8246098732739202e-05,
+      "loss": 3.9365,
+      "step": 1328640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8237712785228682e-05,
+      "loss": 3.9287,
+      "step": 1329152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.822934321652189e-05,
+      "loss": 3.9226,
+      "step": 1329664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.822095726901137e-05,
+      "loss": 3.9267,
+      "step": 1330176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.821257132150085e-05,
+      "loss": 3.9458,
+      "step": 1330688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8204185373990328e-05,
+      "loss": 3.9307,
+      "step": 1331200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8195799426479807e-05,
+      "loss": 3.9288,
+      "step": 1331712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8187413478969287e-05,
+      "loss": 3.9213,
+      "step": 1332224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8179043910262503e-05,
+      "loss": 3.9287,
+      "step": 1332736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.817065796275198e-05,
+      "loss": 3.9441,
+      "step": 1333248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8162288394045196e-05,
+      "loss": 3.9438,
+      "step": 1333760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8153902446534676e-05,
+      "loss": 3.9317,
+      "step": 1334272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8145516499024156e-05,
+      "loss": 3.942,
+      "step": 1334784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.813713055151363e-05,
+      "loss": 3.9437,
+      "step": 1335296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.812874460400311e-05,
+      "loss": 3.9344,
+      "step": 1335808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.812035865649259e-05,
+      "loss": 3.9403,
+      "step": 1336320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8111972708982072e-05,
+      "loss": 3.9314,
+      "step": 1336832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8103586761471552e-05,
+      "loss": 3.9238,
+      "step": 1337344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.809521719276476e-05,
+      "loss": 3.9301,
+      "step": 1337856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.808683124525424e-05,
+      "loss": 3.9312,
+      "step": 1338368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.807844529774372e-05,
+      "loss": 3.9351,
+      "step": 1338880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.80700593502332e-05,
+      "loss": 3.9536,
+      "step": 1339392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.806170616033015e-05,
+      "loss": 3.9424,
+      "step": 1339904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.805332021281963e-05,
+      "loss": 3.9211,
+      "step": 1340416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8044934265309103e-05,
+      "loss": 3.9341,
+      "step": 1340928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8036548317798582e-05,
+      "loss": 3.9362,
+      "step": 1341440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8028162370288062e-05,
+      "loss": 3.9346,
+      "step": 1341952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8019776422777542e-05,
+      "loss": 3.9295,
+      "step": 1342464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8011390475267026e-05,
+      "loss": 3.9305,
+      "step": 1342976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8003004527756506e-05,
+      "loss": 3.9307,
+      "step": 1343488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7994618580245986e-05,
+      "loss": 3.9345,
+      "step": 1344000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7986249011539195e-05,
+      "loss": 3.9256,
+      "step": 1344512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7977863064028675e-05,
+      "loss": 3.9396,
+      "step": 1345024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7969477116518155e-05,
+      "loss": 3.9314,
+      "step": 1345536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7961091169007635e-05,
+      "loss": 3.9404,
+      "step": 1346048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7952705221497118e-05,
+      "loss": 3.9223,
+      "step": 1346560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7944335652790327e-05,
+      "loss": 3.9334,
+      "step": 1347072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7935949705279807e-05,
+      "loss": 3.9319,
+      "step": 1347584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7927563757769287e-05,
+      "loss": 3.9361,
+      "step": 1348096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7919177810258767e-05,
+      "loss": 3.9374,
+      "step": 1348608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.791080824155198e-05,
+      "loss": 3.926,
+      "step": 1349120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.790242229404146e-05,
+      "loss": 3.93,
+      "step": 1349632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.789403634653094e-05,
+      "loss": 3.9304,
+      "step": 1350144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.788565039902042e-05,
+      "loss": 3.934,
+      "step": 1350656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.78772644515099e-05,
+      "loss": 3.9254,
+      "step": 1351168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.786889488280311e-05,
+      "loss": 3.925,
+      "step": 1351680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7860508935292588e-05,
+      "loss": 3.937,
+      "step": 1352192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.78521393665858e-05,
+      "loss": 3.9246,
+      "step": 1352704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.784375341907528e-05,
+      "loss": 3.9258,
+      "step": 1353216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.783536747156476e-05,
+      "loss": 3.9213,
+      "step": 1353728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.782698152405424e-05,
+      "loss": 3.92,
+      "step": 1354240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.781859557654372e-05,
+      "loss": 3.9402,
+      "step": 1354752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.78102096290332e-05,
+      "loss": 3.924,
+      "step": 1355264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.780182368152268e-05,
+      "loss": 3.9349,
+      "step": 1355776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.779343773401216e-05,
+      "loss": 3.9321,
+      "step": 1356288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7785068165305373e-05,
+      "loss": 3.9477,
+      "step": 1356800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7776682217794853e-05,
+      "loss": 3.9344,
+      "step": 1357312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7768296270284333e-05,
+      "loss": 3.9209,
+      "step": 1357824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7759910322773813e-05,
+      "loss": 3.9332,
+      "step": 1358336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.775152437526329e-05,
+      "loss": 3.9363,
+      "step": 1358848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.774313842775277e-05,
+      "loss": 3.9246,
+      "step": 1359360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7734785237849714e-05,
+      "loss": 3.9304,
+      "step": 1359872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7726399290339194e-05,
+      "loss": 3.9258,
+      "step": 1360384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7718013342828674e-05,
+      "loss": 3.9399,
+      "step": 1360896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7709627395318154e-05,
+      "loss": 3.9259,
+      "step": 1361408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7701241447807634e-05,
+      "loss": 3.932,
+      "step": 1361920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.769285550029711e-05,
+      "loss": 3.9265,
+      "step": 1362432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.768446955278659e-05,
+      "loss": 3.9318,
+      "step": 1362944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.767608360527607e-05,
+      "loss": 3.9233,
+      "step": 1363456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.766769765776555e-05,
+      "loss": 3.946,
+      "step": 1363968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.765931171025503e-05,
+      "loss": 3.9376,
+      "step": 1364480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.765092576274451e-05,
+      "loss": 3.925,
+      "step": 1364992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.764253981523399e-05,
+      "loss": 3.9307,
+      "step": 1365504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7634170246527203e-05,
+      "loss": 3.935,
+      "step": 1366016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7625784299016683e-05,
+      "loss": 3.9342,
+      "step": 1366528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7617398351506163e-05,
+      "loss": 3.942,
+      "step": 1367040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7609012403995643e-05,
+      "loss": 3.9257,
+      "step": 1367552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7600642835288852e-05,
+      "loss": 3.9392,
+      "step": 1368064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7592256887778335e-05,
+      "loss": 3.9329,
+      "step": 1368576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7583870940267815e-05,
+      "loss": 3.9377,
+      "step": 1369088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7575484992757295e-05,
+      "loss": 3.9319,
+      "step": 1369600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7567099045246775e-05,
+      "loss": 3.9372,
+      "step": 1370112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7558729476539984e-05,
+      "loss": 3.9334,
+      "step": 1370624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7550359907833197e-05,
+      "loss": 3.9242,
+      "step": 1371136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7541973960322677e-05,
+      "loss": 3.9417,
+      "step": 1371648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7533588012812157e-05,
+      "loss": 3.9352,
+      "step": 1372160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7525202065301637e-05,
+      "loss": 3.9327,
+      "step": 1372672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7516816117791117e-05,
+      "loss": 3.9341,
+      "step": 1373184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7508430170280596e-05,
+      "loss": 3.9239,
+      "step": 1373696
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.007392406463623,
+      "eval_runtime": 600.3355,
+      "eval_samples_per_second": 635.63,
+      "eval_steps_per_second": 19.864,
+      "step": 1373760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7500044222770076e-05,
+      "loss": 3.9266,
+      "step": 1374208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7491658275259556e-05,
+      "loss": 3.9256,
+      "step": 1374720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7483272327749036e-05,
+      "loss": 3.9374,
+      "step": 1375232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.747488638023852e-05,
+      "loss": 3.9271,
+      "step": 1375744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7466500432728e-05,
+      "loss": 3.943,
+      "step": 1376256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7458114485217473e-05,
+      "loss": 3.9273,
+      "step": 1376768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7449728537706953e-05,
+      "loss": 3.9297,
+      "step": 1377280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7441342590196433e-05,
+      "loss": 3.9253,
+      "step": 1377792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7432956642685913e-05,
+      "loss": 3.9426,
+      "step": 1378304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7424570695175396e-05,
+      "loss": 3.9293,
+      "step": 1378816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7416184747664876e-05,
+      "loss": 3.9284,
+      "step": 1379328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7407798800154356e-05,
+      "loss": 3.9294,
+      "step": 1379840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7399429231447565e-05,
+      "loss": 3.9308,
+      "step": 1380352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7391043283937045e-05,
+      "loss": 3.9237,
+      "step": 1380864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7382657336426525e-05,
+      "loss": 3.9275,
+      "step": 1381376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7374271388916005e-05,
+      "loss": 3.9179,
+      "step": 1381888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7365885441405488e-05,
+      "loss": 3.9226,
+      "step": 1382400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7357499493894968e-05,
+      "loss": 3.9212,
+      "step": 1382912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7349113546384448e-05,
+      "loss": 3.9196,
+      "step": 1383424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7340727598873928e-05,
+      "loss": 3.9491,
+      "step": 1383936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7332341651363408e-05,
+      "loss": 3.9253,
+      "step": 1384448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7323955703852888e-05,
+      "loss": 3.926,
+      "step": 1384960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7315569756342365e-05,
+      "loss": 3.9324,
+      "step": 1385472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7307183808831845e-05,
+      "loss": 3.9406,
+      "step": 1385984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.729881424012506e-05,
+      "loss": 3.9177,
+      "step": 1386496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.729044467141827e-05,
+      "loss": 3.9247,
+      "step": 1387008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.728205872390775e-05,
+      "loss": 3.9253,
+      "step": 1387520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.727367277639723e-05,
+      "loss": 3.9302,
+      "step": 1388032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.726528682888671e-05,
+      "loss": 3.9182,
+      "step": 1388544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7256917260179922e-05,
+      "loss": 3.9227,
+      "step": 1389056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7248531312669402e-05,
+      "loss": 3.9253,
+      "step": 1389568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7240145365158882e-05,
+      "loss": 3.9337,
+      "step": 1390080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7231759417648362e-05,
+      "loss": 3.942,
+      "step": 1390592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7223373470137835e-05,
+      "loss": 3.9305,
+      "step": 1391104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7214987522627315e-05,
+      "loss": 3.9243,
+      "step": 1391616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7206601575116798e-05,
+      "loss": 3.932,
+      "step": 1392128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7198232006410014e-05,
+      "loss": 3.9166,
+      "step": 1392640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7189846058899487e-05,
+      "loss": 3.9327,
+      "step": 1393152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7181460111388967e-05,
+      "loss": 3.9153,
+      "step": 1393664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7173074163878447e-05,
+      "loss": 3.9209,
+      "step": 1394176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7164688216367927e-05,
+      "loss": 3.9267,
+      "step": 1394688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7156302268857407e-05,
+      "loss": 3.9355,
+      "step": 1395200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.714791632134689e-05,
+      "loss": 3.9275,
+      "step": 1395712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.713953037383637e-05,
+      "loss": 3.9221,
+      "step": 1396224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.713116080512958e-05,
+      "loss": 3.9276,
+      "step": 1396736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.712277485761906e-05,
+      "loss": 3.9243,
+      "step": 1397248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.711438891010854e-05,
+      "loss": 3.9316,
+      "step": 1397760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.710600296259802e-05,
+      "loss": 3.9133,
+      "step": 1398272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.70976170150875e-05,
+      "loss": 3.9126,
+      "step": 1398784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7089247446380712e-05,
+      "loss": 3.9338,
+      "step": 1399296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7080861498870192e-05,
+      "loss": 3.9311,
+      "step": 1399808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7072475551359672e-05,
+      "loss": 3.9207,
+      "step": 1400320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.706408960384915e-05,
+      "loss": 3.9216,
+      "step": 1400832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.705572003514236e-05,
+      "loss": 3.9243,
+      "step": 1401344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7047334087631844e-05,
+      "loss": 3.9073,
+      "step": 1401856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7038948140121324e-05,
+      "loss": 3.9347,
+      "step": 1402368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7030562192610804e-05,
+      "loss": 3.9194,
+      "step": 1402880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7022176245100284e-05,
+      "loss": 3.9281,
+      "step": 1403392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7013790297589764e-05,
+      "loss": 3.9229,
+      "step": 1403904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7005404350079244e-05,
+      "loss": 3.9227,
+      "step": 1404416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6997018402568724e-05,
+      "loss": 3.9211,
+      "step": 1404928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6988648833861936e-05,
+      "loss": 3.9213,
+      "step": 1405440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6980262886351416e-05,
+      "loss": 3.9151,
+      "step": 1405952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6971876938840896e-05,
+      "loss": 3.9176,
+      "step": 1406464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.696349099133037e-05,
+      "loss": 3.9356,
+      "step": 1406976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6955121422623585e-05,
+      "loss": 3.9231,
+      "step": 1407488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6946735475113065e-05,
+      "loss": 3.9208,
+      "step": 1408000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6938349527602545e-05,
+      "loss": 3.9116,
+      "step": 1408512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6929963580092022e-05,
+      "loss": 3.9185,
+      "step": 1409024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6921594011385238e-05,
+      "loss": 3.9359,
+      "step": 1409536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6913208063874718e-05,
+      "loss": 3.9316,
+      "step": 1410048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6904822116364198e-05,
+      "loss": 3.9231,
+      "step": 1410560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6896436168853674e-05,
+      "loss": 3.9344,
+      "step": 1411072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6888050221343154e-05,
+      "loss": 3.9351,
+      "step": 1411584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.687968065263637e-05,
+      "loss": 3.9214,
+      "step": 1412096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6871294705125843e-05,
+      "loss": 3.9305,
+      "step": 1412608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6862908757615323e-05,
+      "loss": 3.9264,
+      "step": 1413120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6854522810104803e-05,
+      "loss": 3.9162,
+      "step": 1413632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.684615324139802e-05,
+      "loss": 3.9242,
+      "step": 1414144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.683778367269123e-05,
+      "loss": 3.9188,
+      "step": 1414656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.682939772518071e-05,
+      "loss": 3.9261,
+      "step": 1415168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.682101177767019e-05,
+      "loss": 3.9395,
+      "step": 1415680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.681262583015967e-05,
+      "loss": 3.9363,
+      "step": 1416192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.680425626145288e-05,
+      "loss": 3.9106,
+      "step": 1416704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.679587031394236e-05,
+      "loss": 3.9282,
+      "step": 1417216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6787484366431844e-05,
+      "loss": 3.9259,
+      "step": 1417728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6779098418921317e-05,
+      "loss": 3.9254,
+      "step": 1418240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6770712471410797e-05,
+      "loss": 3.9233,
+      "step": 1418752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6762342902704013e-05,
+      "loss": 3.9173,
+      "step": 1419264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6753956955193493e-05,
+      "loss": 3.9246,
+      "step": 1419776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.674557100768297e-05,
+      "loss": 3.9199,
+      "step": 1420288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.673718506017245e-05,
+      "loss": 3.9187,
+      "step": 1420800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6728815491465665e-05,
+      "loss": 3.9303,
+      "step": 1421312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6720429543955145e-05,
+      "loss": 3.9231,
+      "step": 1421824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6712043596444618e-05,
+      "loss": 3.9326,
+      "step": 1422336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6703657648934098e-05,
+      "loss": 3.919,
+      "step": 1422848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.669527170142358e-05,
+      "loss": 3.9216,
+      "step": 1423360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.668690213271679e-05,
+      "loss": 3.9227,
+      "step": 1423872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.667851618520627e-05,
+      "loss": 3.9282,
+      "step": 1424384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.667013023769575e-05,
+      "loss": 3.9291,
+      "step": 1424896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.666174429018523e-05,
+      "loss": 3.9191,
+      "step": 1425408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.665335834267471e-05,
+      "loss": 3.9205,
+      "step": 1425920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.664497239516419e-05,
+      "loss": 3.9211,
+      "step": 1426432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.663658644765367e-05,
+      "loss": 3.9274,
+      "step": 1426944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6628200500143154e-05,
+      "loss": 3.9124,
+      "step": 1427456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6619830931436363e-05,
+      "loss": 3.9148,
+      "step": 1427968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6611444983925843e-05,
+      "loss": 3.9293,
+      "step": 1428480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6603059036415323e-05,
+      "loss": 3.9194,
+      "step": 1428992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6594673088904803e-05,
+      "loss": 3.9138,
+      "step": 1429504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6586287141394283e-05,
+      "loss": 3.9145,
+      "step": 1430016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6577917572687495e-05,
+      "loss": 3.9129,
+      "step": 1430528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6569531625176975e-05,
+      "loss": 3.9287,
+      "step": 1431040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6561145677666455e-05,
+      "loss": 3.9219,
+      "step": 1431552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6552759730155935e-05,
+      "loss": 3.9194,
+      "step": 1432064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6544390161449144e-05,
+      "loss": 3.9221,
+      "step": 1432576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6536004213938624e-05,
+      "loss": 3.9392,
+      "step": 1433088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6527618266428107e-05,
+      "loss": 3.9253,
+      "step": 1433600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6519232318917587e-05,
+      "loss": 3.9128,
+      "step": 1434112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6510862750210796e-05,
+      "loss": 3.9234,
+      "step": 1434624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6502493181504005e-05,
+      "loss": 3.9265,
+      "step": 1435136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6494107233993485e-05,
+      "loss": 3.9177,
+      "step": 1435648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6485737665286698e-05,
+      "loss": 3.921,
+      "step": 1436160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6477351717776178e-05,
+      "loss": 3.9156,
+      "step": 1436672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6468965770265658e-05,
+      "loss": 3.9293,
+      "step": 1437184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6460579822755138e-05,
+      "loss": 3.9156,
+      "step": 1437696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6452193875244618e-05,
+      "loss": 3.9287,
+      "step": 1438208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6443807927734098e-05,
+      "loss": 3.912,
+      "step": 1438720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6435421980223578e-05,
+      "loss": 3.9266,
+      "step": 1439232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.642703603271306e-05,
+      "loss": 3.9132,
+      "step": 1439744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.641865008520254e-05,
+      "loss": 3.9344,
+      "step": 1440256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.641026413769202e-05,
+      "loss": 3.9291,
+      "step": 1440768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.64018781901815e-05,
+      "loss": 3.9212,
+      "step": 1441280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.639350862147471e-05,
+      "loss": 3.9175,
+      "step": 1441792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.638512267396419e-05,
+      "loss": 3.9292,
+      "step": 1442304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.637673672645367e-05,
+      "loss": 3.9227,
+      "step": 1442816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6368350778943153e-05,
+      "loss": 3.933,
+      "step": 1443328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6359981210236362e-05,
+      "loss": 3.9163,
+      "step": 1443840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6351595262725842e-05,
+      "loss": 3.9327,
+      "step": 1444352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6343209315215322e-05,
+      "loss": 3.924,
+      "step": 1444864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6334823367704802e-05,
+      "loss": 3.9355,
+      "step": 1445376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.632643742019428e-05,
+      "loss": 3.9186,
+      "step": 1445888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6318067851487495e-05,
+      "loss": 3.9266,
+      "step": 1446400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6309681903976975e-05,
+      "loss": 3.9244,
+      "step": 1446912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6301295956466448e-05,
+      "loss": 3.9167,
+      "step": 1447424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6292910008955928e-05,
+      "loss": 3.931,
+      "step": 1447936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6284540440249144e-05,
+      "loss": 3.9303,
+      "step": 1448448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6276154492738624e-05,
+      "loss": 3.9171,
+      "step": 1448960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.62677685452281e-05,
+      "loss": 3.9275,
+      "step": 1449472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.625938259771758e-05,
+      "loss": 3.9167,
+      "step": 1449984
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.003355026245117,
+      "eval_runtime": 292.7765,
+      "eval_samples_per_second": 1303.353,
+      "eval_steps_per_second": 40.731,
+      "step": 1450080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.625099665020706e-05,
+      "loss": 3.9201,
+      "step": 1450496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.624261070269654e-05,
+      "loss": 3.9182,
+      "step": 1451008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.623422475518602e-05,
+      "loss": 3.9247,
+      "step": 1451520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.62258388076755e-05,
+      "loss": 3.9235,
+      "step": 1452032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6217469238968712e-05,
+      "loss": 3.9385,
+      "step": 1452544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6209083291458192e-05,
+      "loss": 3.9162,
+      "step": 1453056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6200697343947672e-05,
+      "loss": 3.9204,
+      "step": 1453568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6192311396437152e-05,
+      "loss": 3.9169,
+      "step": 1454080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6183925448926632e-05,
+      "loss": 3.9355,
+      "step": 1454592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6175555880219845e-05,
+      "loss": 3.918,
+      "step": 1455104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6167169932709325e-05,
+      "loss": 3.9178,
+      "step": 1455616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6158783985198805e-05,
+      "loss": 3.9248,
+      "step": 1456128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.615043079529575e-05,
+      "loss": 3.9183,
+      "step": 1456640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6142044847785223e-05,
+      "loss": 3.9162,
+      "step": 1457152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6133658900274706e-05,
+      "loss": 3.918,
+      "step": 1457664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6125272952764186e-05,
+      "loss": 3.9129,
+      "step": 1458176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6116887005253666e-05,
+      "loss": 3.9102,
+      "step": 1458688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6108501057743146e-05,
+      "loss": 3.9122,
+      "step": 1459200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6100115110232626e-05,
+      "loss": 3.9115,
+      "step": 1459712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6091729162722106e-05,
+      "loss": 3.9388,
+      "step": 1460224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6083343215211586e-05,
+      "loss": 3.9159,
+      "step": 1460736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6074973646504798e-05,
+      "loss": 3.9179,
+      "step": 1461248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6066587698994278e-05,
+      "loss": 3.9234,
+      "step": 1461760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6058201751483758e-05,
+      "loss": 3.9292,
+      "step": 1462272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6049815803973238e-05,
+      "loss": 3.9135,
+      "step": 1462784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6041446235266447e-05,
+      "loss": 3.9168,
+      "step": 1463296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6033060287755927e-05,
+      "loss": 3.9135,
+      "step": 1463808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6024674340245407e-05,
+      "loss": 3.9214,
+      "step": 1464320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6016288392734887e-05,
+      "loss": 3.9103,
+      "step": 1464832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.600790244522437e-05,
+      "loss": 3.9147,
+      "step": 1465344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.599951649771385e-05,
+      "loss": 3.92,
+      "step": 1465856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.599113055020333e-05,
+      "loss": 3.9254,
+      "step": 1466368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.598274460269281e-05,
+      "loss": 3.9328,
+      "step": 1466880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.597437503398602e-05,
+      "loss": 3.923,
+      "step": 1467392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.59659890864755e-05,
+      "loss": 3.9155,
+      "step": 1467904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.595760313896498e-05,
+      "loss": 3.9236,
+      "step": 1468416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5949233570258192e-05,
+      "loss": 3.911,
+      "step": 1468928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5940847622747672e-05,
+      "loss": 3.9207,
+      "step": 1469440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5932461675237152e-05,
+      "loss": 3.908,
+      "step": 1469952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5924075727726632e-05,
+      "loss": 3.9109,
+      "step": 1470464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.591568978021611e-05,
+      "loss": 3.9147,
+      "step": 1470976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5907303832705588e-05,
+      "loss": 3.9305,
+      "step": 1471488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5898917885195068e-05,
+      "loss": 3.9225,
+      "step": 1472000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5890531937684548e-05,
+      "loss": 3.9132,
+      "step": 1472512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5882162368977757e-05,
+      "loss": 3.9189,
+      "step": 1473024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5873776421467237e-05,
+      "loss": 3.9148,
+      "step": 1473536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5865390473956717e-05,
+      "loss": 3.9247,
+      "step": 1474048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.58570045264462e-05,
+      "loss": 3.9089,
+      "step": 1474560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.584861857893568e-05,
+      "loss": 3.9006,
+      "step": 1475072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.584024901022889e-05,
+      "loss": 3.9267,
+      "step": 1475584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5831879441522105e-05,
+      "loss": 3.9225,
+      "step": 1476096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.582349349401158e-05,
+      "loss": 3.9081,
+      "step": 1476608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5815107546501062e-05,
+      "loss": 3.9151,
+      "step": 1477120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5806721598990542e-05,
+      "loss": 3.9186,
+      "step": 1477632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5798352030283758e-05,
+      "loss": 3.9003,
+      "step": 1478144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.578996608277323e-05,
+      "loss": 3.9214,
+      "step": 1478656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.578158013526271e-05,
+      "loss": 3.9101,
+      "step": 1479168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.577319418775219e-05,
+      "loss": 3.9236,
+      "step": 1479680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.576480824024167e-05,
+      "loss": 3.9083,
+      "step": 1480192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5756438671534883e-05,
+      "loss": 3.9194,
+      "step": 1480704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5748052724024363e-05,
+      "loss": 3.9095,
+      "step": 1481216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5739666776513843e-05,
+      "loss": 3.9139,
+      "step": 1481728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5731280829003323e-05,
+      "loss": 3.9021,
+      "step": 1482240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5722894881492803e-05,
+      "loss": 3.9119,
+      "step": 1482752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5714525312786016e-05,
+      "loss": 3.9288,
+      "step": 1483264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5706139365275496e-05,
+      "loss": 3.913,
+      "step": 1483776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5697753417764976e-05,
+      "loss": 3.9151,
+      "step": 1484288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5689367470254455e-05,
+      "loss": 3.9019,
+      "step": 1484800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5680981522743935e-05,
+      "loss": 3.9113,
+      "step": 1485312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5672595575233415e-05,
+      "loss": 3.9266,
+      "step": 1485824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5664209627722895e-05,
+      "loss": 3.9217,
+      "step": 1486336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5655823680212375e-05,
+      "loss": 3.9197,
+      "step": 1486848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5647454111505588e-05,
+      "loss": 3.9261,
+      "step": 1487360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5639068163995068e-05,
+      "loss": 3.9241,
+      "step": 1487872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5630682216484548e-05,
+      "loss": 3.9169,
+      "step": 1488384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5622296268974028e-05,
+      "loss": 3.9188,
+      "step": 1488896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5613910321463508e-05,
+      "loss": 3.9185,
+      "step": 1489408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5605540752756717e-05,
+      "loss": 3.9115,
+      "step": 1489920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.55971548052462e-05,
+      "loss": 3.9124,
+      "step": 1490432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.558876885773568e-05,
+      "loss": 3.9108,
+      "step": 1490944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.558038291022516e-05,
+      "loss": 3.9159,
+      "step": 1491456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.557201334151837e-05,
+      "loss": 3.9292,
+      "step": 1491968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.556362739400785e-05,
+      "loss": 3.9267,
+      "step": 1492480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.555524144649733e-05,
+      "loss": 3.9017,
+      "step": 1492992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.554685549898681e-05,
+      "loss": 3.9249,
+      "step": 1493504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5538469551476292e-05,
+      "loss": 3.9165,
+      "step": 1494016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.553011636157323e-05,
+      "loss": 3.9164,
+      "step": 1494528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.552173041406271e-05,
+      "loss": 3.9162,
+      "step": 1495040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.551334446655219e-05,
+      "loss": 3.9094,
+      "step": 1495552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.550495851904167e-05,
+      "loss": 3.9166,
+      "step": 1496064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5496572571531154e-05,
+      "loss": 3.9118,
+      "step": 1496576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5488219381628092e-05,
+      "loss": 3.9121,
+      "step": 1497088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5479833434117572e-05,
+      "loss": 3.9184,
+      "step": 1497600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5471447486607052e-05,
+      "loss": 3.9133,
+      "step": 1498112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5463061539096532e-05,
+      "loss": 3.927,
+      "step": 1498624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5454675591586015e-05,
+      "loss": 3.9105,
+      "step": 1499136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5446289644075495e-05,
+      "loss": 3.9106,
+      "step": 1499648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5437903696564975e-05,
+      "loss": 3.9138,
+      "step": 1500160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5429517749054455e-05,
+      "loss": 3.9236,
+      "step": 1500672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5421131801543935e-05,
+      "loss": 3.9165,
+      "step": 1501184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5412745854033415e-05,
+      "loss": 3.9104,
+      "step": 1501696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.540435990652289e-05,
+      "loss": 3.9095,
+      "step": 1502208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.539597395901237e-05,
+      "loss": 3.9137,
+      "step": 1502720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5387620769109316e-05,
+      "loss": 3.9233,
+      "step": 1503232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5379234821598796e-05,
+      "loss": 3.9071,
+      "step": 1503744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5370848874088276e-05,
+      "loss": 3.9033,
+      "step": 1504256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5362462926577756e-05,
+      "loss": 3.921,
+      "step": 1504768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5354076979067236e-05,
+      "loss": 3.9094,
+      "step": 1505280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5345691031556716e-05,
+      "loss": 3.9049,
+      "step": 1505792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5337305084046193e-05,
+      "loss": 3.9048,
+      "step": 1506304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5328919136535673e-05,
+      "loss": 3.9071,
+      "step": 1506816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5320533189025153e-05,
+      "loss": 3.921,
+      "step": 1507328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5312163620318362e-05,
+      "loss": 3.9063,
+      "step": 1507840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5303777672807845e-05,
+      "loss": 3.9119,
+      "step": 1508352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5295391725297325e-05,
+      "loss": 3.9178,
+      "step": 1508864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5287005777786805e-05,
+      "loss": 3.9266,
+      "step": 1509376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5278619830276285e-05,
+      "loss": 3.9236,
+      "step": 1509888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5270250261569494e-05,
+      "loss": 3.9021,
+      "step": 1510400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5261864314058974e-05,
+      "loss": 3.9129,
+      "step": 1510912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5253478366548454e-05,
+      "loss": 3.9157,
+      "step": 1511424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5245092419037934e-05,
+      "loss": 3.9143,
+      "step": 1511936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5236722850331146e-05,
+      "loss": 3.9148,
+      "step": 1512448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5228336902820626e-05,
+      "loss": 3.9058,
+      "step": 1512960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5219950955310106e-05,
+      "loss": 3.9207,
+      "step": 1513472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5211565007799586e-05,
+      "loss": 3.9085,
+      "step": 1513984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5203179060289066e-05,
+      "loss": 3.9163,
+      "step": 1514496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5194793112778546e-05,
+      "loss": 3.9058,
+      "step": 1515008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5186407165268026e-05,
+      "loss": 3.9174,
+      "step": 1515520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.517802121775751e-05,
+      "loss": 3.9039,
+      "step": 1516032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.516965164905072e-05,
+      "loss": 3.9313,
+      "step": 1516544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5161282080343928e-05,
+      "loss": 3.915,
+      "step": 1517056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5152896132833408e-05,
+      "loss": 3.9165,
+      "step": 1517568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5144510185322888e-05,
+      "loss": 3.9083,
+      "step": 1518080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.513612423781237e-05,
+      "loss": 3.9213,
+      "step": 1518592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.512773829030185e-05,
+      "loss": 3.9151,
+      "step": 1519104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.511935234279133e-05,
+      "loss": 3.9245,
+      "step": 1519616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.511096639528081e-05,
+      "loss": 3.909,
+      "step": 1520128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.510258044777029e-05,
+      "loss": 3.9204,
+      "step": 1520640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.50942108790635e-05,
+      "loss": 3.9156,
+      "step": 1521152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.508582493155298e-05,
+      "loss": 3.9275,
+      "step": 1521664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5077438984042463e-05,
+      "loss": 3.9077,
+      "step": 1522176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5069053036531943e-05,
+      "loss": 3.9198,
+      "step": 1522688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5060683467825152e-05,
+      "loss": 3.9134,
+      "step": 1523200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5052297520314632e-05,
+      "loss": 3.9126,
+      "step": 1523712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.504392795160784e-05,
+      "loss": 3.9184,
+      "step": 1524224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5035542004097325e-05,
+      "loss": 3.919,
+      "step": 1524736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5027156056586805e-05,
+      "loss": 3.9121,
+      "step": 1525248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5018770109076285e-05,
+      "loss": 3.9165,
+      "step": 1525760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5010384161565765e-05,
+      "loss": 3.9121,
+      "step": 1526272
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9993996620178223,
+      "eval_runtime": 293.8629,
+      "eval_samples_per_second": 1298.534,
+      "eval_steps_per_second": 40.58,
+      "step": 1526400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5001998214055244e-05,
+      "loss": 3.9112,
+      "step": 1526784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.499361226654472e-05,
+      "loss": 3.9067,
+      "step": 1527296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.49852263190342e-05,
+      "loss": 3.9164,
+      "step": 1527808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.497684037152368e-05,
+      "loss": 3.9151,
+      "step": 1528320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4968454424013164e-05,
+      "loss": 3.9309,
+      "step": 1528832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.496006847650264e-05,
+      "loss": 3.9128,
+      "step": 1529344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.495168252899212e-05,
+      "loss": 3.9104,
+      "step": 1529856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.49432965814816e-05,
+      "loss": 3.9097,
+      "step": 1530368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.493491063397108e-05,
+      "loss": 3.9242,
+      "step": 1530880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.492652468646056e-05,
+      "loss": 3.9143,
+      "step": 1531392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.491813873895004e-05,
+      "loss": 3.908,
+      "step": 1531904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4909752791439524e-05,
+      "loss": 3.9137,
+      "step": 1532416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4901383222732733e-05,
+      "loss": 3.9099,
+      "step": 1532928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4892997275222213e-05,
+      "loss": 3.9083,
+      "step": 1533440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4884611327711693e-05,
+      "loss": 3.9141,
+      "step": 1533952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4876225380201173e-05,
+      "loss": 3.9,
+      "step": 1534464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.486783943269065e-05,
+      "loss": 3.9042,
+      "step": 1534976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4859453485180133e-05,
+      "loss": 3.9031,
+      "step": 1535488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4851067537669613e-05,
+      "loss": 3.9025,
+      "step": 1536000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4842681590159093e-05,
+      "loss": 3.9278,
+      "step": 1536512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4834295642648573e-05,
+      "loss": 3.9115,
+      "step": 1537024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4825909695138053e-05,
+      "loss": 3.9082,
+      "step": 1537536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.481752374762753e-05,
+      "loss": 3.913,
+      "step": 1538048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.480913780011701e-05,
+      "loss": 3.9229,
+      "step": 1538560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4800784610213954e-05,
+      "loss": 3.906,
+      "step": 1539072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4792398662703434e-05,
+      "loss": 3.9117,
+      "step": 1539584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4784012715192914e-05,
+      "loss": 3.9051,
+      "step": 1540096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4775626767682394e-05,
+      "loss": 3.9088,
+      "step": 1540608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4767257198975603e-05,
+      "loss": 3.9033,
+      "step": 1541120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4758871251465087e-05,
+      "loss": 3.9097,
+      "step": 1541632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4750485303954567e-05,
+      "loss": 3.9098,
+      "step": 1542144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4742099356444046e-05,
+      "loss": 3.917,
+      "step": 1542656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4733713408933526e-05,
+      "loss": 3.9227,
+      "step": 1543168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4725327461423003e-05,
+      "loss": 3.9164,
+      "step": 1543680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4716941513912483e-05,
+      "loss": 3.9049,
+      "step": 1544192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4708555566401963e-05,
+      "loss": 3.9134,
+      "step": 1544704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.470018599769518e-05,
+      "loss": 3.9074,
+      "step": 1545216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4691800050184655e-05,
+      "loss": 3.9098,
+      "step": 1545728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4683414102674135e-05,
+      "loss": 3.902,
+      "step": 1546240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4675028155163615e-05,
+      "loss": 3.9035,
+      "step": 1546752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4666658586456828e-05,
+      "loss": 3.9066,
+      "step": 1547264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4658272638946304e-05,
+      "loss": 3.9195,
+      "step": 1547776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4649886691435788e-05,
+      "loss": 3.9108,
+      "step": 1548288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4641500743925268e-05,
+      "loss": 3.9072,
+      "step": 1548800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4633131175218477e-05,
+      "loss": 3.9101,
+      "step": 1549312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.462476160651169e-05,
+      "loss": 3.9073,
+      "step": 1549824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.461637565900117e-05,
+      "loss": 3.9167,
+      "step": 1550336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.460798971149065e-05,
+      "loss": 3.9011,
+      "step": 1550848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.459960376398013e-05,
+      "loss": 3.8935,
+      "step": 1551360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.459123419527334e-05,
+      "loss": 3.917,
+      "step": 1551872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.458284824776282e-05,
+      "loss": 3.9141,
+      "step": 1552384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4574462300252298e-05,
+      "loss": 3.8989,
+      "step": 1552896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4566076352741778e-05,
+      "loss": 3.9065,
+      "step": 1553408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4557690405231258e-05,
+      "loss": 3.9117,
+      "step": 1553920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.454930445772074e-05,
+      "loss": 3.8927,
+      "step": 1554432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.454091851021022e-05,
+      "loss": 3.9135,
+      "step": 1554944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.45325325626997e-05,
+      "loss": 3.9017,
+      "step": 1555456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.452416299399291e-05,
+      "loss": 3.9121,
+      "step": 1555968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.451577704648239e-05,
+      "loss": 3.9003,
+      "step": 1556480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.450739109897187e-05,
+      "loss": 3.9112,
+      "step": 1556992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.449900515146135e-05,
+      "loss": 3.9018,
+      "step": 1557504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4490635582754563e-05,
+      "loss": 3.9085,
+      "step": 1558016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4482266014047772e-05,
+      "loss": 3.8943,
+      "step": 1558528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4473880066537252e-05,
+      "loss": 3.9058,
+      "step": 1559040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.446549411902673e-05,
+      "loss": 3.917,
+      "step": 1559552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.445710817151621e-05,
+      "loss": 3.9069,
+      "step": 1560064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4448722224005695e-05,
+      "loss": 3.9024,
+      "step": 1560576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4440336276495175e-05,
+      "loss": 3.8946,
+      "step": 1561088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4431950328984655e-05,
+      "loss": 3.9014,
+      "step": 1561600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.442356438147413e-05,
+      "loss": 3.9162,
+      "step": 1562112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.441517843396361e-05,
+      "loss": 3.9173,
+      "step": 1562624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4406808865256824e-05,
+      "loss": 3.9117,
+      "step": 1563136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4398422917746304e-05,
+      "loss": 3.9177,
+      "step": 1563648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4390036970235784e-05,
+      "loss": 3.9169,
+      "step": 1564160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4381651022725264e-05,
+      "loss": 3.9074,
+      "step": 1564672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4373281454018476e-05,
+      "loss": 3.9085,
+      "step": 1565184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4364895506507953e-05,
+      "loss": 3.9131,
+      "step": 1565696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4356509558997433e-05,
+      "loss": 3.9027,
+      "step": 1566208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4348123611486913e-05,
+      "loss": 3.9037,
+      "step": 1566720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4339770421583858e-05,
+      "loss": 3.9052,
+      "step": 1567232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4331384474073338e-05,
+      "loss": 3.9066,
+      "step": 1567744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4322998526562818e-05,
+      "loss": 3.9226,
+      "step": 1568256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4314612579052298e-05,
+      "loss": 3.9195,
+      "step": 1568768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4306226631541778e-05,
+      "loss": 3.8958,
+      "step": 1569280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4297840684031258e-05,
+      "loss": 3.9126,
+      "step": 1569792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4289454736520737e-05,
+      "loss": 3.9111,
+      "step": 1570304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.428108516781395e-05,
+      "loss": 3.9079,
+      "step": 1570816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4272699220303427e-05,
+      "loss": 3.9089,
+      "step": 1571328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4264313272792906e-05,
+      "loss": 3.8987,
+      "step": 1571840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4255927325282386e-05,
+      "loss": 3.9081,
+      "step": 1572352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4247541377771866e-05,
+      "loss": 3.9041,
+      "step": 1572864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.423917180906508e-05,
+      "loss": 3.902,
+      "step": 1573376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.423078586155456e-05,
+      "loss": 3.9044,
+      "step": 1573888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.422239991404404e-05,
+      "loss": 3.9078,
+      "step": 1574400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.421401396653352e-05,
+      "loss": 3.9129,
+      "step": 1574912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4205628019023e-05,
+      "loss": 3.9108,
+      "step": 1575424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.419724207151248e-05,
+      "loss": 3.8986,
+      "step": 1575936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.418885612400196e-05,
+      "loss": 3.9054,
+      "step": 1576448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.418047017649144e-05,
+      "loss": 3.9093,
+      "step": 1576960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.417210060778465e-05,
+      "loss": 3.9133,
+      "step": 1577472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.416373103907786e-05,
+      "loss": 3.9007,
+      "step": 1577984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.415534509156734e-05,
+      "loss": 3.9028,
+      "step": 1578496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.414695914405682e-05,
+      "loss": 3.9062,
+      "step": 1579008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4138573196546303e-05,
+      "loss": 3.9131,
+      "step": 1579520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4130187249035783e-05,
+      "loss": 3.9006,
+      "step": 1580032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.412180130152526e-05,
+      "loss": 3.899,
+      "step": 1580544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.411341535401474e-05,
+      "loss": 3.9081,
+      "step": 1581056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.410502940650422e-05,
+      "loss": 3.9032,
+      "step": 1581568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4096659837797432e-05,
+      "loss": 3.8972,
+      "step": 1582080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4088290269090645e-05,
+      "loss": 3.8989,
+      "step": 1582592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4079904321580125e-05,
+      "loss": 3.8981,
+      "step": 1583104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4071518374069605e-05,
+      "loss": 3.9137,
+      "step": 1583616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.406313242655908e-05,
+      "loss": 3.9032,
+      "step": 1584128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.405474647904856e-05,
+      "loss": 3.9017,
+      "step": 1584640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.404636053153804e-05,
+      "loss": 3.9095,
+      "step": 1585152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.403797458402752e-05,
+      "loss": 3.9214,
+      "step": 1585664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4029588636517004e-05,
+      "loss": 3.9155,
+      "step": 1586176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4021219067810214e-05,
+      "loss": 3.8954,
+      "step": 1586688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4012833120299694e-05,
+      "loss": 3.904,
+      "step": 1587200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4004447172789174e-05,
+      "loss": 3.9076,
+      "step": 1587712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3996061225278653e-05,
+      "loss": 3.904,
+      "step": 1588224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3987691656571866e-05,
+      "loss": 3.9088,
+      "step": 1588736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3979305709061346e-05,
+      "loss": 3.8972,
+      "step": 1589248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3970919761550826e-05,
+      "loss": 3.9103,
+      "step": 1589760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3962533814040306e-05,
+      "loss": 3.907,
+      "step": 1590272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3954147866529786e-05,
+      "loss": 3.9065,
+      "step": 1590784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3945778297822995e-05,
+      "loss": 3.8967,
+      "step": 1591296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3937392350312475e-05,
+      "loss": 3.9077,
+      "step": 1591808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3929006402801958e-05,
+      "loss": 3.9026,
+      "step": 1592320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3920620455291438e-05,
+      "loss": 3.918,
+      "step": 1592832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3912234507780915e-05,
+      "loss": 3.9049,
+      "step": 1593344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3903848560270395e-05,
+      "loss": 3.9127,
+      "step": 1593856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3895478991563607e-05,
+      "loss": 3.8989,
+      "step": 1594368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3887093044053087e-05,
+      "loss": 3.9155,
+      "step": 1594880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3878707096542567e-05,
+      "loss": 3.9086,
+      "step": 1595392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3870321149032047e-05,
+      "loss": 3.9138,
+      "step": 1595904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3861935201521527e-05,
+      "loss": 3.8998,
+      "step": 1596416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3853549254011007e-05,
+      "loss": 3.9165,
+      "step": 1596928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3845163306500487e-05,
+      "loss": 3.9048,
+      "step": 1597440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3836777358989967e-05,
+      "loss": 3.9217,
+      "step": 1597952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3828391411479443e-05,
+      "loss": 3.9022,
+      "step": 1598464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.382002184277266e-05,
+      "loss": 3.912,
+      "step": 1598976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.381163589526214e-05,
+      "loss": 3.907,
+      "step": 1599488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.380324994775162e-05,
+      "loss": 3.9012,
+      "step": 1600000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3794880379044828e-05,
+      "loss": 3.9121,
+      "step": 1600512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3786494431534308e-05,
+      "loss": 3.9137,
+      "step": 1601024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3778108484023788e-05,
+      "loss": 3.9034,
+      "step": 1601536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3769738915317e-05,
+      "loss": 3.907,
+      "step": 1602048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.376135296780648e-05,
+      "loss": 3.9081,
+      "step": 1602560
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9954891204833984,
+      "eval_runtime": 299.5483,
+      "eval_samples_per_second": 1273.888,
+      "eval_steps_per_second": 39.81,
+      "step": 1602720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.375296702029596e-05,
+      "loss": 3.9038,
+      "step": 1603072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.374458107278544e-05,
+      "loss": 3.8998,
+      "step": 1603584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3736195125274917e-05,
+      "loss": 3.9095,
+      "step": 1604096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3727809177764397e-05,
+      "loss": 3.907,
+      "step": 1604608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.371942323025388e-05,
+      "loss": 3.9177,
+      "step": 1605120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.371103728274336e-05,
+      "loss": 3.9059,
+      "step": 1605632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.370265133523284e-05,
+      "loss": 3.9028,
+      "step": 1606144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.369426538772232e-05,
+      "loss": 3.9037,
+      "step": 1606656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.36858794402118e-05,
+      "loss": 3.9152,
+      "step": 1607168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3677493492701277e-05,
+      "loss": 3.9055,
+      "step": 1607680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.366912392399449e-05,
+      "loss": 3.9029,
+      "step": 1608192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3660737976483973e-05,
+      "loss": 3.9036,
+      "step": 1608704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3652368407777182e-05,
+      "loss": 3.9037,
+      "step": 1609216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.364398246026666e-05,
+      "loss": 3.9027,
+      "step": 1609728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.363559651275614e-05,
+      "loss": 3.9078,
+      "step": 1610240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.362721056524562e-05,
+      "loss": 3.8928,
+      "step": 1610752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3618824617735098e-05,
+      "loss": 3.8949,
+      "step": 1611264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.361043867022458e-05,
+      "loss": 3.8988,
+      "step": 1611776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.360205272271406e-05,
+      "loss": 3.8939,
+      "step": 1612288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.359366677520354e-05,
+      "loss": 3.9164,
+      "step": 1612800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.358528082769302e-05,
+      "loss": 3.9083,
+      "step": 1613312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.35768948801825e-05,
+      "loss": 3.9013,
+      "step": 1613824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.356850893267198e-05,
+      "loss": 3.9026,
+      "step": 1614336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3560122985161458e-05,
+      "loss": 3.9142,
+      "step": 1614848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3551753416454674e-05,
+      "loss": 3.9002,
+      "step": 1615360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3543383847747883e-05,
+      "loss": 3.9046,
+      "step": 1615872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3535014279041095e-05,
+      "loss": 3.8942,
+      "step": 1616384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3526628331530572e-05,
+      "loss": 3.9039,
+      "step": 1616896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3518242384020052e-05,
+      "loss": 3.8997,
+      "step": 1617408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3509856436509535e-05,
+      "loss": 3.8986,
+      "step": 1617920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3501470488999015e-05,
+      "loss": 3.9013,
+      "step": 1618432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3493084541488495e-05,
+      "loss": 3.9085,
+      "step": 1618944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3484698593977975e-05,
+      "loss": 3.9123,
+      "step": 1619456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3476312646467455e-05,
+      "loss": 3.9096,
+      "step": 1619968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.346792669895693e-05,
+      "loss": 3.8958,
+      "step": 1620480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.345954075144641e-05,
+      "loss": 3.9092,
+      "step": 1620992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3451171182739627e-05,
+      "loss": 3.902,
+      "step": 1621504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3442785235229104e-05,
+      "loss": 3.8947,
+      "step": 1622016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3434399287718584e-05,
+      "loss": 3.8974,
+      "step": 1622528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3426013340208064e-05,
+      "loss": 3.8956,
+      "step": 1623040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3417627392697544e-05,
+      "loss": 3.901,
+      "step": 1623552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3409241445187024e-05,
+      "loss": 3.9156,
+      "step": 1624064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3400855497676504e-05,
+      "loss": 3.9003,
+      "step": 1624576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3392469550165984e-05,
+      "loss": 3.8986,
+      "step": 1625088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3384099981459196e-05,
+      "loss": 3.9085,
+      "step": 1625600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3375714033948676e-05,
+      "loss": 3.8998,
+      "step": 1626112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3367344465241885e-05,
+      "loss": 3.9046,
+      "step": 1626624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3358958517731365e-05,
+      "loss": 3.8966,
+      "step": 1627136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3350572570220845e-05,
+      "loss": 3.8847,
+      "step": 1627648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.334218662271033e-05,
+      "loss": 3.9142,
+      "step": 1628160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.333380067519981e-05,
+      "loss": 3.9045,
+      "step": 1628672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3325431106493018e-05,
+      "loss": 3.8976,
+      "step": 1629184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3317045158982497e-05,
+      "loss": 3.8975,
+      "step": 1629696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3308659211471977e-05,
+      "loss": 3.8996,
+      "step": 1630208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3300273263961457e-05,
+      "loss": 3.8843,
+      "step": 1630720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3291887316450937e-05,
+      "loss": 3.9026,
+      "step": 1631232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3283501368940417e-05,
+      "loss": 3.901,
+      "step": 1631744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3275115421429897e-05,
+      "loss": 3.9045,
+      "step": 1632256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3266729473919377e-05,
+      "loss": 3.8974,
+      "step": 1632768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3258359905212586e-05,
+      "loss": 3.9012,
+      "step": 1633280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3249973957702066e-05,
+      "loss": 3.8956,
+      "step": 1633792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3241588010191546e-05,
+      "loss": 3.9042,
+      "step": 1634304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.323320206268103e-05,
+      "loss": 3.8864,
+      "step": 1634816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.322481611517051e-05,
+      "loss": 3.8962,
+      "step": 1635328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.321644654646372e-05,
+      "loss": 3.9084,
+      "step": 1635840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.320807697775693e-05,
+      "loss": 3.9017,
+      "step": 1636352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3199691030246408e-05,
+      "loss": 3.8939,
+      "step": 1636864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.319130508273589e-05,
+      "loss": 3.8905,
+      "step": 1637376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.318291913522537e-05,
+      "loss": 3.8919,
+      "step": 1637888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.317453318771485e-05,
+      "loss": 3.9052,
+      "step": 1638400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.316614724020433e-05,
+      "loss": 3.9117,
+      "step": 1638912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.315776129269381e-05,
+      "loss": 3.904,
+      "step": 1639424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3149375345183287e-05,
+      "loss": 3.9113,
+      "step": 1639936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.31410057764765e-05,
+      "loss": 3.9076,
+      "step": 1640448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3132619828965983e-05,
+      "loss": 3.9098,
+      "step": 1640960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3124233881455463e-05,
+      "loss": 3.8987,
+      "step": 1641472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.311584793394494e-05,
+      "loss": 3.9032,
+      "step": 1641984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3107478365238152e-05,
+      "loss": 3.9012,
+      "step": 1642496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3099092417727632e-05,
+      "loss": 3.8939,
+      "step": 1643008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3090706470217112e-05,
+      "loss": 3.9028,
+      "step": 1643520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3082320522706592e-05,
+      "loss": 3.8945,
+      "step": 1644032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3073934575196072e-05,
+      "loss": 3.911,
+      "step": 1644544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3065581385293014e-05,
+      "loss": 3.9163,
+      "step": 1645056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3057195437782494e-05,
+      "loss": 3.8857,
+      "step": 1645568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3048809490271974e-05,
+      "loss": 3.9087,
+      "step": 1646080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3040423542761454e-05,
+      "loss": 3.8992,
+      "step": 1646592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3032053974054666e-05,
+      "loss": 3.9037,
+      "step": 1647104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3023668026544146e-05,
+      "loss": 3.9017,
+      "step": 1647616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3015282079033626e-05,
+      "loss": 3.893,
+      "step": 1648128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3006896131523106e-05,
+      "loss": 3.9021,
+      "step": 1648640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2998510184012586e-05,
+      "loss": 3.8987,
+      "step": 1649152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.29901406153058e-05,
+      "loss": 3.8986,
+      "step": 1649664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2981754667795278e-05,
+      "loss": 3.8946,
+      "step": 1650176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2973368720284758e-05,
+      "loss": 3.9041,
+      "step": 1650688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2964982772774235e-05,
+      "loss": 3.9062,
+      "step": 1651200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2956596825263715e-05,
+      "loss": 3.9038,
+      "step": 1651712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2948210877753195e-05,
+      "loss": 3.8909,
+      "step": 1652224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2939824930242675e-05,
+      "loss": 3.9011,
+      "step": 1652736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2931438982732155e-05,
+      "loss": 3.8995,
+      "step": 1653248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2923069414025367e-05,
+      "loss": 3.9099,
+      "step": 1653760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2914683466514847e-05,
+      "loss": 3.8947,
+      "step": 1654272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2906297519004327e-05,
+      "loss": 3.9002,
+      "step": 1654784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2897911571493807e-05,
+      "loss": 3.8949,
+      "step": 1655296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2889542002787016e-05,
+      "loss": 3.9077,
+      "step": 1655808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.28811560552765e-05,
+      "loss": 3.8909,
+      "step": 1656320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.287277010776598e-05,
+      "loss": 3.8948,
+      "step": 1656832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.286438416025546e-05,
+      "loss": 3.8981,
+      "step": 1657344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.285601459154867e-05,
+      "loss": 3.8987,
+      "step": 1657856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.284764502284188e-05,
+      "loss": 3.8876,
+      "step": 1658368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.283925907533136e-05,
+      "loss": 3.8963,
+      "step": 1658880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.283087312782084e-05,
+      "loss": 3.8917,
+      "step": 1659392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.282248718031032e-05,
+      "loss": 3.9012,
+      "step": 1659904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2814117611603533e-05,
+      "loss": 3.9012,
+      "step": 1660416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.280573166409301e-05,
+      "loss": 3.8902,
+      "step": 1660928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.279734571658249e-05,
+      "loss": 3.9039,
+      "step": 1661440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.278895976907197e-05,
+      "loss": 3.9151,
+      "step": 1661952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2780573821561453e-05,
+      "loss": 3.9092,
+      "step": 1662464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2772187874050933e-05,
+      "loss": 3.8884,
+      "step": 1662976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2763801926540413e-05,
+      "loss": 3.8983,
+      "step": 1663488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.275541597902989e-05,
+      "loss": 3.9014,
+      "step": 1664000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2747046410323102e-05,
+      "loss": 3.8959,
+      "step": 1664512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2738676841616315e-05,
+      "loss": 3.904,
+      "step": 1665024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2730290894105795e-05,
+      "loss": 3.8901,
+      "step": 1665536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2721904946595274e-05,
+      "loss": 3.9041,
+      "step": 1666048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2713518999084754e-05,
+      "loss": 3.8987,
+      "step": 1666560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2705133051574234e-05,
+      "loss": 3.9049,
+      "step": 1667072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.269674710406371e-05,
+      "loss": 3.8921,
+      "step": 1667584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2688377535356923e-05,
+      "loss": 3.8946,
+      "step": 1668096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2679991587846407e-05,
+      "loss": 3.9008,
+      "step": 1668608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2671605640335887e-05,
+      "loss": 3.9075,
+      "step": 1669120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2663219692825363e-05,
+      "loss": 3.9,
+      "step": 1669632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2654833745314843e-05,
+      "loss": 3.9027,
+      "step": 1670144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2646447797804323e-05,
+      "loss": 3.8952,
+      "step": 1670656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2638078229097536e-05,
+      "loss": 3.9062,
+      "step": 1671168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2629692281587016e-05,
+      "loss": 3.9006,
+      "step": 1671680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2621306334076496e-05,
+      "loss": 3.9077,
+      "step": 1672192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2612920386565976e-05,
+      "loss": 3.8965,
+      "step": 1672704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2604534439055456e-05,
+      "loss": 3.9055,
+      "step": 1673216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2596148491544935e-05,
+      "loss": 3.8985,
+      "step": 1673728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2587762544034415e-05,
+      "loss": 3.92,
+      "step": 1674240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2579376596523892e-05,
+      "loss": 3.8926,
+      "step": 1674752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2571007027817108e-05,
+      "loss": 3.9068,
+      "step": 1675264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2562621080306588e-05,
+      "loss": 3.8995,
+      "step": 1675776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2554235132796068e-05,
+      "loss": 3.8965,
+      "step": 1676288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2545865564089277e-05,
+      "loss": 3.9021,
+      "step": 1676800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2537479616578757e-05,
+      "loss": 3.9019,
+      "step": 1677312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2529093669068237e-05,
+      "loss": 3.9035,
+      "step": 1677824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.252072410036145e-05,
+      "loss": 3.8959,
+      "step": 1678336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.251233815285093e-05,
+      "loss": 3.9053,
+      "step": 1678848
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9926960468292236,
+      "eval_runtime": 310.4718,
+      "eval_samples_per_second": 1229.068,
+      "eval_steps_per_second": 38.409,
+      "step": 1679040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.250395220534041e-05,
+      "loss": 3.9041,
+      "step": 1679360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.249556625782989e-05,
+      "loss": 3.8941,
+      "step": 1679872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2487180310319366e-05,
+      "loss": 3.8985,
+      "step": 1680384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2478794362808846e-05,
+      "loss": 3.9004,
+      "step": 1680896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.247040841529833e-05,
+      "loss": 3.9114,
+      "step": 1681408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.246202246778781e-05,
+      "loss": 3.9017,
+      "step": 1681920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.245363652027729e-05,
+      "loss": 3.8915,
+      "step": 1682432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.244525057276677e-05,
+      "loss": 3.8977,
+      "step": 1682944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.243686462525625e-05,
+      "loss": 3.9078,
+      "step": 1683456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2428478677745725e-05,
+      "loss": 3.9,
+      "step": 1683968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2420092730235205e-05,
+      "loss": 3.8976,
+      "step": 1684480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2411706782724685e-05,
+      "loss": 3.8976,
+      "step": 1684992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.24033372140179e-05,
+      "loss": 3.8996,
+      "step": 1685504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2394951266507378e-05,
+      "loss": 3.8944,
+      "step": 1686016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2386565318996858e-05,
+      "loss": 3.9015,
+      "step": 1686528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2378179371486338e-05,
+      "loss": 3.8864,
+      "step": 1687040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2369793423975818e-05,
+      "loss": 3.8881,
+      "step": 1687552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2361407476465298e-05,
+      "loss": 3.8918,
+      "step": 1688064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2353021528954778e-05,
+      "loss": 3.8881,
+      "step": 1688576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2344635581444257e-05,
+      "loss": 3.9107,
+      "step": 1689088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2336249633933737e-05,
+      "loss": 3.9029,
+      "step": 1689600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2327863686423217e-05,
+      "loss": 3.8968,
+      "step": 1690112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2319477738912697e-05,
+      "loss": 3.8954,
+      "step": 1690624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2311091791402177e-05,
+      "loss": 3.9037,
+      "step": 1691136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2302722222695386e-05,
+      "loss": 3.893,
+      "step": 1691648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2294352653988602e-05,
+      "loss": 3.8994,
+      "step": 1692160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2285966706478082e-05,
+      "loss": 3.8891,
+      "step": 1692672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.227758075896756e-05,
+      "loss": 3.8969,
+      "step": 1693184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.226919481145704e-05,
+      "loss": 3.8965,
+      "step": 1693696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.226080886394652e-05,
+      "loss": 3.8893,
+      "step": 1694208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2252422916436e-05,
+      "loss": 3.8958,
+      "step": 1694720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.224403696892548e-05,
+      "loss": 3.9043,
+      "step": 1695232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2235651021414962e-05,
+      "loss": 3.9013,
+      "step": 1695744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222726507390444e-05,
+      "loss": 3.9046,
+      "step": 1696256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.221887912639392e-05,
+      "loss": 3.8903,
+      "step": 1696768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.22104931788834e-05,
+      "loss": 3.9077,
+      "step": 1697280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.220212361017661e-05,
+      "loss": 3.8961,
+      "step": 1697792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.219373766266609e-05,
+      "loss": 3.8872,
+      "step": 1698304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.218535171515557e-05,
+      "loss": 3.8925,
+      "step": 1698816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.217696576764505e-05,
+      "loss": 3.8896,
+      "step": 1699328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.216857982013453e-05,
+      "loss": 3.8943,
+      "step": 1699840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.216019387262401e-05,
+      "loss": 3.9088,
+      "step": 1700352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.215180792511349e-05,
+      "loss": 3.8938,
+      "step": 1700864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2143421977602967e-05,
+      "loss": 3.8939,
+      "step": 1701376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.213505240889618e-05,
+      "loss": 3.9022,
+      "step": 1701888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2126666461385663e-05,
+      "loss": 3.8938,
+      "step": 1702400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2118296892678872e-05,
+      "loss": 3.9007,
+      "step": 1702912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2109910945168352e-05,
+      "loss": 3.8871,
+      "step": 1703424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2101524997657832e-05,
+      "loss": 3.8768,
+      "step": 1703936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2093139050147312e-05,
+      "loss": 3.9066,
+      "step": 1704448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2084753102636792e-05,
+      "loss": 3.9001,
+      "step": 1704960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2076367155126272e-05,
+      "loss": 3.8881,
+      "step": 1705472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2067981207615752e-05,
+      "loss": 3.897,
+      "step": 1705984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2059595260105232e-05,
+      "loss": 3.8918,
+      "step": 1706496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.205122569139844e-05,
+      "loss": 3.8808,
+      "step": 1707008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.204283974388792e-05,
+      "loss": 3.893,
+      "step": 1707520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.20344537963774e-05,
+      "loss": 3.8952,
+      "step": 1708032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2026084227670617e-05,
+      "loss": 3.898,
+      "step": 1708544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2017714658963826e-05,
+      "loss": 3.8902,
+      "step": 1709056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2009328711453306e-05,
+      "loss": 3.8965,
+      "step": 1709568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000942763942786e-05,
+      "loss": 3.8876,
+      "step": 1710080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1992556816432266e-05,
+      "loss": 3.8984,
+      "step": 1710592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1984170868921746e-05,
+      "loss": 3.8809,
+      "step": 1711104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1975784921411226e-05,
+      "loss": 3.8874,
+      "step": 1711616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1967398973900706e-05,
+      "loss": 3.9029,
+      "step": 1712128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1959013026390185e-05,
+      "loss": 3.8965,
+      "step": 1712640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1950643457683395e-05,
+      "loss": 3.8919,
+      "step": 1713152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1942257510172875e-05,
+      "loss": 3.8868,
+      "step": 1713664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1933871562662355e-05,
+      "loss": 3.8845,
+      "step": 1714176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1925485615151838e-05,
+      "loss": 3.8945,
+      "step": 1714688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1917116046445047e-05,
+      "loss": 3.9087,
+      "step": 1715200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1908730098934527e-05,
+      "loss": 3.8963,
+      "step": 1715712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1900344151424007e-05,
+      "loss": 3.9067,
+      "step": 1716224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1891958203913487e-05,
+      "loss": 3.9014,
+      "step": 1716736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.18835886352067e-05,
+      "loss": 3.9047,
+      "step": 1717248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.187520268769618e-05,
+      "loss": 3.8913,
+      "step": 1717760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.186681674018566e-05,
+      "loss": 3.8942,
+      "step": 1718272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.185843079267514e-05,
+      "loss": 3.8945,
+      "step": 1718784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.185004484516462e-05,
+      "loss": 3.8831,
+      "step": 1719296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1841658897654096e-05,
+      "loss": 3.898,
+      "step": 1719808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1833289328947308e-05,
+      "loss": 3.8923,
+      "step": 1720320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1824903381436788e-05,
+      "loss": 3.907,
+      "step": 1720832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.181651743392627e-05,
+      "loss": 3.9076,
+      "step": 1721344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1808131486415748e-05,
+      "loss": 3.8813,
+      "step": 1721856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1799745538905228e-05,
+      "loss": 3.9059,
+      "step": 1722368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1791359591394708e-05,
+      "loss": 3.892,
+      "step": 1722880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1782973643884188e-05,
+      "loss": 3.8967,
+      "step": 1723392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1774587696373668e-05,
+      "loss": 3.8937,
+      "step": 1723904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.176621812766688e-05,
+      "loss": 3.8875,
+      "step": 1724416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1757848558960093e-05,
+      "loss": 3.8979,
+      "step": 1724928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.174946261144957e-05,
+      "loss": 3.8907,
+      "step": 1725440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.174107666393905e-05,
+      "loss": 3.8925,
+      "step": 1725952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.173269071642853e-05,
+      "loss": 3.8918,
+      "step": 1726464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.172430476891801e-05,
+      "loss": 3.8972,
+      "step": 1726976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1715918821407493e-05,
+      "loss": 3.9028,
+      "step": 1727488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1707532873896973e-05,
+      "loss": 3.8888,
+      "step": 1728000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1699146926386453e-05,
+      "loss": 3.8894,
+      "step": 1728512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.169077735767966e-05,
+      "loss": 3.8932,
+      "step": 1729024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.168239141016914e-05,
+      "loss": 3.8907,
+      "step": 1729536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.167400546265862e-05,
+      "loss": 3.9062,
+      "step": 1730048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.16656195151481e-05,
+      "loss": 3.8893,
+      "step": 1730560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1657249946441314e-05,
+      "loss": 3.8932,
+      "step": 1731072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1648863998930794e-05,
+      "loss": 3.8904,
+      "step": 1731584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1640478051420274e-05,
+      "loss": 3.9032,
+      "step": 1732096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.163209210390975e-05,
+      "loss": 3.88,
+      "step": 1732608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.162370615639923e-05,
+      "loss": 3.8908,
+      "step": 1733120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1615336587692446e-05,
+      "loss": 3.8907,
+      "step": 1733632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1606950640181926e-05,
+      "loss": 3.8925,
+      "step": 1734144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1598564692671403e-05,
+      "loss": 3.8823,
+      "step": 1734656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1590178745160883e-05,
+      "loss": 3.8896,
+      "step": 1735168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1581809176454095e-05,
+      "loss": 3.8842,
+      "step": 1735680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1573423228943572e-05,
+      "loss": 3.896,
+      "step": 1736192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1565037281433055e-05,
+      "loss": 3.8933,
+      "step": 1736704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1556651333922535e-05,
+      "loss": 3.8872,
+      "step": 1737216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1548265386412015e-05,
+      "loss": 3.894,
+      "step": 1737728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1539895817705224e-05,
+      "loss": 3.9082,
+      "step": 1738240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1531509870194704e-05,
+      "loss": 3.906,
+      "step": 1738752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1523123922684184e-05,
+      "loss": 3.8831,
+      "step": 1739264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1514737975173664e-05,
+      "loss": 3.8895,
+      "step": 1739776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1506368406466877e-05,
+      "loss": 3.8964,
+      "step": 1740288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1497982458956356e-05,
+      "loss": 3.888,
+      "step": 1740800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.148961289024957e-05,
+      "loss": 3.8994,
+      "step": 1741312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1481226942739046e-05,
+      "loss": 3.8796,
+      "step": 1741824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1472840995228525e-05,
+      "loss": 3.9039,
+      "step": 1742336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.146445504771801e-05,
+      "loss": 3.8885,
+      "step": 1742848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.145606910020749e-05,
+      "loss": 3.9001,
+      "step": 1743360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.144768315269697e-05,
+      "loss": 3.8862,
+      "step": 1743872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1439313583990178e-05,
+      "loss": 3.8916,
+      "step": 1744384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1430927636479658e-05,
+      "loss": 3.8919,
+      "step": 1744896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1422541688969138e-05,
+      "loss": 3.9033,
+      "step": 1745408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1414155741458618e-05,
+      "loss": 3.8909,
+      "step": 1745920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.14057697939481e-05,
+      "loss": 3.8986,
+      "step": 1746432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.139738384643758e-05,
+      "loss": 3.889,
+      "step": 1746944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.138901427773079e-05,
+      "loss": 3.9032,
+      "step": 1747456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.138062833022027e-05,
+      "loss": 3.8935,
+      "step": 1747968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.137224238270975e-05,
+      "loss": 3.8975,
+      "step": 1748480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1363856435199227e-05,
+      "loss": 3.8937,
+      "step": 1748992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.135547048768871e-05,
+      "loss": 3.9008,
+      "step": 1749504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.134708454017819e-05,
+      "loss": 3.8941,
+      "step": 1750016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.133869859266767e-05,
+      "loss": 3.9117,
+      "step": 1750528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.133032902396088e-05,
+      "loss": 3.8882,
+      "step": 1751040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.132194307645036e-05,
+      "loss": 3.8994,
+      "step": 1751552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.131355712893984e-05,
+      "loss": 3.8945,
+      "step": 1752064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1305187560233055e-05,
+      "loss": 3.8931,
+      "step": 1752576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.129680161272253e-05,
+      "loss": 3.8921,
+      "step": 1753088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.128841566521201e-05,
+      "loss": 3.8953,
+      "step": 1753600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.128002971770149e-05,
+      "loss": 3.8985,
+      "step": 1754112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.127164377019097e-05,
+      "loss": 3.8922,
+      "step": 1754624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.126325782268045e-05,
+      "loss": 3.9027,
+      "step": 1755136
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9906952381134033,
+      "eval_runtime": 311.5103,
+      "eval_samples_per_second": 1224.971,
+      "eval_steps_per_second": 38.281,
+      "step": 1755360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.125487187516993e-05,
+      "loss": 3.8875,
+      "step": 1755648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.124648592765941e-05,
+      "loss": 3.8871,
+      "step": 1756160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1238116358952623e-05,
+      "loss": 3.8937,
+      "step": 1756672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1229746790245833e-05,
+      "loss": 3.8945,
+      "step": 1757184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1221360842735313e-05,
+      "loss": 3.9044,
+      "step": 1757696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1212974895224792e-05,
+      "loss": 3.8981,
+      "step": 1758208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1204588947714272e-05,
+      "loss": 3.8907,
+      "step": 1758720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1196203000203756e-05,
+      "loss": 3.8892,
+      "step": 1759232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1187817052693236e-05,
+      "loss": 3.9028,
+      "step": 1759744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1179431105182712e-05,
+      "loss": 3.8963,
+      "step": 1760256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1171045157672192e-05,
+      "loss": 3.8895,
+      "step": 1760768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1162675588965405e-05,
+      "loss": 3.8899,
+      "step": 1761280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1154306020258617e-05,
+      "loss": 3.8961,
+      "step": 1761792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1145920072748097e-05,
+      "loss": 3.8887,
+      "step": 1762304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1137534125237577e-05,
+      "loss": 3.8949,
+      "step": 1762816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1129148177727057e-05,
+      "loss": 3.8797,
+      "step": 1763328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1120762230216534e-05,
+      "loss": 3.8852,
+      "step": 1763840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1112376282706014e-05,
+      "loss": 3.8866,
+      "step": 1764352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1103990335195494e-05,
+      "loss": 3.8827,
+      "step": 1764864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1095604387684974e-05,
+      "loss": 3.9022,
+      "step": 1765376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1087234818978186e-05,
+      "loss": 3.8978,
+      "step": 1765888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1078848871467666e-05,
+      "loss": 3.8914,
+      "step": 1766400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1070462923957146e-05,
+      "loss": 3.8886,
+      "step": 1766912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1062076976446626e-05,
+      "loss": 3.8978,
+      "step": 1767424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1053691028936106e-05,
+      "loss": 3.8901,
+      "step": 1767936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.104533783903305e-05,
+      "loss": 3.8924,
+      "step": 1768448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.103695189152253e-05,
+      "loss": 3.882,
+      "step": 1768960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1028565944012007e-05,
+      "loss": 3.893,
+      "step": 1769472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1020179996501487e-05,
+      "loss": 3.8942,
+      "step": 1769984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1011794048990967e-05,
+      "loss": 3.8827,
+      "step": 1770496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1003408101480447e-05,
+      "loss": 3.8879,
+      "step": 1771008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0995022153969927e-05,
+      "loss": 3.8971,
+      "step": 1771520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.098663620645941e-05,
+      "loss": 3.8936,
+      "step": 1772032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0978250258948887e-05,
+      "loss": 3.9032,
+      "step": 1772544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0969864311438367e-05,
+      "loss": 3.8863,
+      "step": 1773056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.096151112153531e-05,
+      "loss": 3.8951,
+      "step": 1773568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.095312517402479e-05,
+      "loss": 3.8918,
+      "step": 1774080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0944739226514272e-05,
+      "loss": 3.8835,
+      "step": 1774592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0936353279003752e-05,
+      "loss": 3.886,
+      "step": 1775104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.092798371029696e-05,
+      "loss": 3.8859,
+      "step": 1775616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.091959776278644e-05,
+      "loss": 3.8858,
+      "step": 1776128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.091121181527592e-05,
+      "loss": 3.8996,
+      "step": 1776640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.09028258677654e-05,
+      "loss": 3.891,
+      "step": 1777152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.089443992025488e-05,
+      "loss": 3.888,
+      "step": 1777664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.088605397274436e-05,
+      "loss": 3.897,
+      "step": 1778176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0877684404037573e-05,
+      "loss": 3.8881,
+      "step": 1778688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0869298456527053e-05,
+      "loss": 3.8889,
+      "step": 1779200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0860912509016533e-05,
+      "loss": 3.8866,
+      "step": 1779712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.085252656150601e-05,
+      "loss": 3.8651,
+      "step": 1780224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0844140613995493e-05,
+      "loss": 3.9087,
+      "step": 1780736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0835754666484973e-05,
+      "loss": 3.8895,
+      "step": 1781248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0827368718974453e-05,
+      "loss": 3.889,
+      "step": 1781760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0818982771463933e-05,
+      "loss": 3.8878,
+      "step": 1782272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0810629581560875e-05,
+      "loss": 3.8862,
+      "step": 1782784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0802243634050355e-05,
+      "loss": 3.8774,
+      "step": 1783296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0793857686539835e-05,
+      "loss": 3.8826,
+      "step": 1783808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0785471739029314e-05,
+      "loss": 3.8899,
+      "step": 1784320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0777085791518794e-05,
+      "loss": 3.8904,
+      "step": 1784832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0768699844008274e-05,
+      "loss": 3.8913,
+      "step": 1785344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0760313896497754e-05,
+      "loss": 3.8826,
+      "step": 1785856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0751927948987234e-05,
+      "loss": 3.8838,
+      "step": 1786368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0743542001476714e-05,
+      "loss": 3.8974,
+      "step": 1786880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0735156053966194e-05,
+      "loss": 3.8732,
+      "step": 1787392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0726786485259407e-05,
+      "loss": 3.8828,
+      "step": 1787904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0718400537748887e-05,
+      "loss": 3.8946,
+      "step": 1788416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0710014590238367e-05,
+      "loss": 3.8919,
+      "step": 1788928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0701628642727843e-05,
+      "loss": 3.8828,
+      "step": 1789440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0693259074021056e-05,
+      "loss": 3.884,
+      "step": 1789952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0684873126510536e-05,
+      "loss": 3.8741,
+      "step": 1790464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0676487179000016e-05,
+      "loss": 3.8894,
+      "step": 1790976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0668101231489496e-05,
+      "loss": 3.9022,
+      "step": 1791488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0659731662782708e-05,
+      "loss": 3.8919,
+      "step": 1792000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0651345715272188e-05,
+      "loss": 3.8991,
+      "step": 1792512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0642959767761665e-05,
+      "loss": 3.8935,
+      "step": 1793024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0634573820251148e-05,
+      "loss": 3.9037,
+      "step": 1793536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.062620425154436e-05,
+      "loss": 3.8857,
+      "step": 1794048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.061781830403384e-05,
+      "loss": 3.8919,
+      "step": 1794560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0609432356523317e-05,
+      "loss": 3.8897,
+      "step": 1795072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0601046409012797e-05,
+      "loss": 3.8754,
+      "step": 1795584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.059267684030601e-05,
+      "loss": 3.8923,
+      "step": 1796096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.058429089279549e-05,
+      "loss": 3.8863,
+      "step": 1796608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.057590494528497e-05,
+      "loss": 3.8957,
+      "step": 1797120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.056751899777445e-05,
+      "loss": 3.903,
+      "step": 1797632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0559149429067662e-05,
+      "loss": 3.8782,
+      "step": 1798144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0550763481557138e-05,
+      "loss": 3.8973,
+      "step": 1798656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0542377534046618e-05,
+      "loss": 3.888,
+      "step": 1799168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.05339915865361e-05,
+      "loss": 3.8925,
+      "step": 1799680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.052560563902558e-05,
+      "loss": 3.8863,
+      "step": 1800192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.051723607031879e-05,
+      "loss": 3.885,
+      "step": 1800704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0508866501612003e-05,
+      "loss": 3.8884,
+      "step": 1801216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0500480554101483e-05,
+      "loss": 3.8847,
+      "step": 1801728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0492110985394696e-05,
+      "loss": 3.8871,
+      "step": 1802240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0483725037884175e-05,
+      "loss": 3.8851,
+      "step": 1802752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0475339090373655e-05,
+      "loss": 3.8955,
+      "step": 1803264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0466953142863135e-05,
+      "loss": 3.8945,
+      "step": 1803776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0458567195352612e-05,
+      "loss": 3.8838,
+      "step": 1804288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0450181247842092e-05,
+      "loss": 3.8822,
+      "step": 1804800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0441795300331572e-05,
+      "loss": 3.8877,
+      "step": 1805312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0433409352821052e-05,
+      "loss": 3.8891,
+      "step": 1805824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0425023405310535e-05,
+      "loss": 3.8953,
+      "step": 1806336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0416637457800015e-05,
+      "loss": 3.8866,
+      "step": 1806848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0408251510289492e-05,
+      "loss": 3.885,
+      "step": 1807360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.039986556277897e-05,
+      "loss": 3.8842,
+      "step": 1807872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0391495994072184e-05,
+      "loss": 3.902,
+      "step": 1808384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0383110046561664e-05,
+      "loss": 3.8726,
+      "step": 1808896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0374724099051144e-05,
+      "loss": 3.8883,
+      "step": 1809408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0366338151540624e-05,
+      "loss": 3.8795,
+      "step": 1809920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0357952204030104e-05,
+      "loss": 3.8878,
+      "step": 1810432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0349582635323316e-05,
+      "loss": 3.8791,
+      "step": 1810944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0341196687812793e-05,
+      "loss": 3.8888,
+      "step": 1811456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0332810740302273e-05,
+      "loss": 3.8771,
+      "step": 1811968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.032444117159549e-05,
+      "loss": 3.89,
+      "step": 1812480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0316055224084965e-05,
+      "loss": 3.889,
+      "step": 1812992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0307669276574445e-05,
+      "loss": 3.8795,
+      "step": 1813504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0299283329063925e-05,
+      "loss": 3.8885,
+      "step": 1814016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0290897381553405e-05,
+      "loss": 3.9055,
+      "step": 1814528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0282511434042885e-05,
+      "loss": 3.8906,
+      "step": 1815040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0274141865336098e-05,
+      "loss": 3.8882,
+      "step": 1815552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0265755917825578e-05,
+      "loss": 3.8777,
+      "step": 1816064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0257369970315058e-05,
+      "loss": 3.8899,
+      "step": 1816576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0248984022804538e-05,
+      "loss": 3.8891,
+      "step": 1817088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0240614454097747e-05,
+      "loss": 3.8914,
+      "step": 1817600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0232228506587227e-05,
+      "loss": 3.8741,
+      "step": 1818112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.022384255907671e-05,
+      "loss": 3.8969,
+      "step": 1818624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.021547299036992e-05,
+      "loss": 3.8846,
+      "step": 1819136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.02070870428594e-05,
+      "loss": 3.8923,
+      "step": 1819648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.019870109534888e-05,
+      "loss": 3.8808,
+      "step": 1820160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.019031514783836e-05,
+      "loss": 3.8881,
+      "step": 1820672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.018192920032784e-05,
+      "loss": 3.8835,
+      "step": 1821184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.017354325281732e-05,
+      "loss": 3.9005,
+      "step": 1821696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.01651573053068e-05,
+      "loss": 3.8893,
+      "step": 1822208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.015677135779628e-05,
+      "loss": 3.8919,
+      "step": 1822720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.014838541028576e-05,
+      "loss": 3.8807,
+      "step": 1823232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.013999946277524e-05,
+      "loss": 3.8998,
+      "step": 1823744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.013161351526472e-05,
+      "loss": 3.889,
+      "step": 1824256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.01232275677542e-05,
+      "loss": 3.8934,
+      "step": 1824768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0114874377851144e-05,
+      "loss": 3.8892,
+      "step": 1825280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.010648843034062e-05,
+      "loss": 3.8946,
+      "step": 1825792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.00981024828301e-05,
+      "loss": 3.8853,
+      "step": 1826304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.008971653531958e-05,
+      "loss": 3.9057,
+      "step": 1826816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.008133058780906e-05,
+      "loss": 3.8829,
+      "step": 1827328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.007294464029854e-05,
+      "loss": 3.8955,
+      "step": 1827840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.006455869278802e-05,
+      "loss": 3.8925,
+      "step": 1828352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0056172745277503e-05,
+      "loss": 3.8857,
+      "step": 1828864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.004778679776698e-05,
+      "loss": 3.8847,
+      "step": 1829376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0039417229060192e-05,
+      "loss": 3.8923,
+      "step": 1829888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0031031281549672e-05,
+      "loss": 3.893,
+      "step": 1830400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.002266171284288e-05,
+      "loss": 3.8877,
+      "step": 1830912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0014275765332365e-05,
+      "loss": 3.8966,
+      "step": 1831424
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.988779306411743,
+      "eval_runtime": 299.4996,
+      "eval_samples_per_second": 1274.095,
+      "eval_steps_per_second": 39.816,
+      "step": 1831680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0005889817821845e-05,
+      "loss": 3.8849,
+      "step": 1831936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9997503870311325e-05,
+      "loss": 3.8866,
+      "step": 1832448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99891179228008e-05,
+      "loss": 3.8861,
+      "step": 1832960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.998073197529028e-05,
+      "loss": 3.8904,
+      "step": 1833472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.997234602777976e-05,
+      "loss": 3.9007,
+      "step": 1833984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.996396008026924e-05,
+      "loss": 3.8883,
+      "step": 1834496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.995557413275872e-05,
+      "loss": 3.8913,
+      "step": 1835008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9947188185248204e-05,
+      "loss": 3.8834,
+      "step": 1835520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9938802237737684e-05,
+      "loss": 3.8924,
+      "step": 1836032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.993041629022716e-05,
+      "loss": 3.8944,
+      "step": 1836544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.992203034271664e-05,
+      "loss": 3.8848,
+      "step": 1837056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.991364439520612e-05,
+      "loss": 3.8815,
+      "step": 1837568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9905274826499333e-05,
+      "loss": 3.8941,
+      "step": 1838080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9896905257792546e-05,
+      "loss": 3.883,
+      "step": 1838592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9888519310282026e-05,
+      "loss": 3.8827,
+      "step": 1839104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9880133362771506e-05,
+      "loss": 3.8773,
+      "step": 1839616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9871747415260982e-05,
+      "loss": 3.8764,
+      "step": 1840128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9863361467750462e-05,
+      "loss": 3.8835,
+      "step": 1840640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9854975520239942e-05,
+      "loss": 3.8761,
+      "step": 1841152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9846589572729422e-05,
+      "loss": 3.8961,
+      "step": 1841664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9838203625218905e-05,
+      "loss": 3.8935,
+      "step": 1842176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9829817677708385e-05,
+      "loss": 3.8902,
+      "step": 1842688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9821431730197865e-05,
+      "loss": 3.8828,
+      "step": 1843200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9813045782687342e-05,
+      "loss": 3.8865,
+      "step": 1843712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9804659835176822e-05,
+      "loss": 3.8887,
+      "step": 1844224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9796290266470034e-05,
+      "loss": 3.8886,
+      "step": 1844736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9787904318959514e-05,
+      "loss": 3.8788,
+      "step": 1845248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9779518371448994e-05,
+      "loss": 3.8892,
+      "step": 1845760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9771132423938474e-05,
+      "loss": 3.8845,
+      "step": 1846272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9762746476427954e-05,
+      "loss": 3.8785,
+      "step": 1846784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9754376907721163e-05,
+      "loss": 3.8783,
+      "step": 1847296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9745990960210643e-05,
+      "loss": 3.8955,
+      "step": 1847808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9737605012700127e-05,
+      "loss": 3.8926,
+      "step": 1848320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9729219065189607e-05,
+      "loss": 3.8964,
+      "step": 1848832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9720833117679087e-05,
+      "loss": 3.8795,
+      "step": 1849344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9712447170168566e-05,
+      "loss": 3.8849,
+      "step": 1849856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9704077601461776e-05,
+      "loss": 3.8939,
+      "step": 1850368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9695691653951256e-05,
+      "loss": 3.8789,
+      "step": 1850880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9687305706440735e-05,
+      "loss": 3.8789,
+      "step": 1851392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9678919758930215e-05,
+      "loss": 3.8774,
+      "step": 1851904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9670533811419695e-05,
+      "loss": 3.8843,
+      "step": 1852416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9662164242712908e-05,
+      "loss": 3.8871,
+      "step": 1852928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9653778295202388e-05,
+      "loss": 3.8921,
+      "step": 1853440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9645392347691868e-05,
+      "loss": 3.886,
+      "step": 1853952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9637006400181344e-05,
+      "loss": 3.8842,
+      "step": 1854464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9628620452670828e-05,
+      "loss": 3.8864,
+      "step": 1854976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9620234505160308e-05,
+      "loss": 3.8803,
+      "step": 1855488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.961186493645352e-05,
+      "loss": 3.8843,
+      "step": 1856000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9603478988942997e-05,
+      "loss": 3.8566,
+      "step": 1856512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9595093041432477e-05,
+      "loss": 3.9021,
+      "step": 1857024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9586707093921957e-05,
+      "loss": 3.8823,
+      "step": 1857536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9578321146411437e-05,
+      "loss": 3.8898,
+      "step": 1858048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.956993519890092e-05,
+      "loss": 3.8805,
+      "step": 1858560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.95615492513904e-05,
+      "loss": 3.8805,
+      "step": 1859072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9553163303879876e-05,
+      "loss": 3.8712,
+      "step": 1859584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.954479373517309e-05,
+      "loss": 3.8768,
+      "step": 1860096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.953640778766257e-05,
+      "loss": 3.888,
+      "step": 1860608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.952803821895578e-05,
+      "loss": 3.8827,
+      "step": 1861120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.951965227144526e-05,
+      "loss": 3.8877,
+      "step": 1861632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.951126632393474e-05,
+      "loss": 3.8799,
+      "step": 1862144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.950288037642422e-05,
+      "loss": 3.8737,
+      "step": 1862656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94944944289137e-05,
+      "loss": 3.8937,
+      "step": 1863168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9486108481403178e-05,
+      "loss": 3.8657,
+      "step": 1863680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9477722533892658e-05,
+      "loss": 3.8735,
+      "step": 1864192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9469336586382138e-05,
+      "loss": 3.8912,
+      "step": 1864704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.946096701767535e-05,
+      "loss": 3.891,
+      "step": 1865216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.945258107016483e-05,
+      "loss": 3.8725,
+      "step": 1865728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.944419512265431e-05,
+      "loss": 3.8839,
+      "step": 1866240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.943580917514379e-05,
+      "loss": 3.8663,
+      "step": 1866752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9427439606437e-05,
+      "loss": 3.8879,
+      "step": 1867264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9419053658926482e-05,
+      "loss": 3.8918,
+      "step": 1867776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9410667711415962e-05,
+      "loss": 3.8908,
+      "step": 1868288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9402281763905442e-05,
+      "loss": 3.888,
+      "step": 1868800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.939391219519865e-05,
+      "loss": 3.8905,
+      "step": 1869312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.938552624768813e-05,
+      "loss": 3.9001,
+      "step": 1869824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.937714030017761e-05,
+      "loss": 3.8787,
+      "step": 1870336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.936875435266709e-05,
+      "loss": 3.8878,
+      "step": 1870848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9360368405156575e-05,
+      "loss": 3.887,
+      "step": 1871360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9351998836449784e-05,
+      "loss": 3.8652,
+      "step": 1871872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9343629267742996e-05,
+      "loss": 3.8867,
+      "step": 1872384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9335243320232473e-05,
+      "loss": 3.885,
+      "step": 1872896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9326857372721953e-05,
+      "loss": 3.8885,
+      "step": 1873408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9318471425211436e-05,
+      "loss": 3.9032,
+      "step": 1873920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9310085477700916e-05,
+      "loss": 3.8747,
+      "step": 1874432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9301699530190396e-05,
+      "loss": 3.8857,
+      "step": 1874944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9293313582679876e-05,
+      "loss": 3.8846,
+      "step": 1875456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9284927635169356e-05,
+      "loss": 3.8874,
+      "step": 1875968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9276541687658833e-05,
+      "loss": 3.8795,
+      "step": 1876480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9268172118952045e-05,
+      "loss": 3.8821,
+      "step": 1876992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.925978617144153e-05,
+      "loss": 3.8811,
+      "step": 1877504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9251400223931005e-05,
+      "loss": 3.8827,
+      "step": 1878016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9243014276420485e-05,
+      "loss": 3.8789,
+      "step": 1878528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9234644707713697e-05,
+      "loss": 3.883,
+      "step": 1879040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9226258760203177e-05,
+      "loss": 3.8907,
+      "step": 1879552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.921788919149639e-05,
+      "loss": 3.8842,
+      "step": 1880064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.920950324398587e-05,
+      "loss": 3.8868,
+      "step": 1880576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.920111729647535e-05,
+      "loss": 3.8746,
+      "step": 1881088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9192731348964826e-05,
+      "loss": 3.8809,
+      "step": 1881600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9184345401454306e-05,
+      "loss": 3.883,
+      "step": 1882112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9175959453943786e-05,
+      "loss": 3.8931,
+      "step": 1882624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9167573506433266e-05,
+      "loss": 3.8836,
+      "step": 1883136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9159187558922746e-05,
+      "loss": 3.878,
+      "step": 1883648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.915080161141223e-05,
+      "loss": 3.8854,
+      "step": 1884160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.914243204270544e-05,
+      "loss": 3.8884,
+      "step": 1884672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.913404609519492e-05,
+      "loss": 3.8707,
+      "step": 1885184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.91256601476844e-05,
+      "loss": 3.8825,
+      "step": 1885696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.911727420017388e-05,
+      "loss": 3.8704,
+      "step": 1886208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.910888825266336e-05,
+      "loss": 3.8899,
+      "step": 1886720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.910051868395657e-05,
+      "loss": 3.8748,
+      "step": 1887232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.909214911524978e-05,
+      "loss": 3.8798,
+      "step": 1887744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.908376316773926e-05,
+      "loss": 3.8733,
+      "step": 1888256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.907537722022874e-05,
+      "loss": 3.8843,
+      "step": 1888768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.906699127271822e-05,
+      "loss": 3.8863,
+      "step": 1889280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.90586053252077e-05,
+      "loss": 3.8765,
+      "step": 1889792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9050219377697183e-05,
+      "loss": 3.8823,
+      "step": 1890304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.904183343018666e-05,
+      "loss": 3.8979,
+      "step": 1890816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.903344748267614e-05,
+      "loss": 3.8909,
+      "step": 1891328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9025077913969352e-05,
+      "loss": 3.88,
+      "step": 1891840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9016691966458832e-05,
+      "loss": 3.8718,
+      "step": 1892352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9008322397752045e-05,
+      "loss": 3.8855,
+      "step": 1892864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8999936450241525e-05,
+      "loss": 3.8876,
+      "step": 1893376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8991566881534734e-05,
+      "loss": 3.8814,
+      "step": 1893888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8983180934024214e-05,
+      "loss": 3.8717,
+      "step": 1894400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8974794986513694e-05,
+      "loss": 3.8895,
+      "step": 1894912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8966409039003173e-05,
+      "loss": 3.883,
+      "step": 1895424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8958023091492653e-05,
+      "loss": 3.8877,
+      "step": 1895936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8949637143982133e-05,
+      "loss": 3.8738,
+      "step": 1896448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8941251196471613e-05,
+      "loss": 3.8828,
+      "step": 1896960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8932865248961093e-05,
+      "loss": 3.8776,
+      "step": 1897472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8924479301450573e-05,
+      "loss": 3.8922,
+      "step": 1897984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8916093353940053e-05,
+      "loss": 3.8845,
+      "step": 1898496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8907707406429533e-05,
+      "loss": 3.8865,
+      "step": 1899008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8899321458919013e-05,
+      "loss": 3.8739,
+      "step": 1899520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8890951890212226e-05,
+      "loss": 3.8972,
+      "step": 1900032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8882565942701706e-05,
+      "loss": 3.8811,
+      "step": 1900544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8874179995191186e-05,
+      "loss": 3.8897,
+      "step": 1901056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8865794047680662e-05,
+      "loss": 3.8829,
+      "step": 1901568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8857424478973875e-05,
+      "loss": 3.883,
+      "step": 1902080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8849054910267087e-05,
+      "loss": 3.8834,
+      "step": 1902592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8840668962756567e-05,
+      "loss": 3.8986,
+      "step": 1903104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8832283015246047e-05,
+      "loss": 3.8791,
+      "step": 1903616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8823897067735527e-05,
+      "loss": 3.8912,
+      "step": 1904128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8815511120225007e-05,
+      "loss": 3.887,
+      "step": 1904640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8807125172714487e-05,
+      "loss": 3.8785,
+      "step": 1905152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8798739225203967e-05,
+      "loss": 3.8821,
+      "step": 1905664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8790353277693447e-05,
+      "loss": 3.8837,
+      "step": 1906176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8781967330182927e-05,
+      "loss": 3.8871,
+      "step": 1906688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.877361414027987e-05,
+      "loss": 3.8868,
+      "step": 1907200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8765228192769348e-05,
+      "loss": 3.8913,
+      "step": 1907712
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9872899055480957,
+      "eval_runtime": 295.3202,
+      "eval_samples_per_second": 1292.126,
+      "eval_steps_per_second": 40.38,
+      "step": 1908000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8756842245258828e-05,
+      "loss": 3.8795,
+      "step": 1908224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8748456297748308e-05,
+      "loss": 3.8786,
+      "step": 1908736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8740070350237788e-05,
+      "loss": 3.8815,
+      "step": 1909248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8731684402727268e-05,
+      "loss": 3.8863,
+      "step": 1909760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8723298455216748e-05,
+      "loss": 3.8947,
+      "step": 1910272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8714912507706228e-05,
+      "loss": 3.8826,
+      "step": 1910784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8706526560195708e-05,
+      "loss": 3.8854,
+      "step": 1911296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8698140612685188e-05,
+      "loss": 3.8812,
+      "step": 1911808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8689754665174668e-05,
+      "loss": 3.8838,
+      "step": 1912320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8681368717664148e-05,
+      "loss": 3.8932,
+      "step": 1912832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8672982770153628e-05,
+      "loss": 3.8769,
+      "step": 1913344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8664596822643108e-05,
+      "loss": 3.8764,
+      "step": 1913856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8656227253936317e-05,
+      "loss": 3.8916,
+      "step": 1914368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8647841306425797e-05,
+      "loss": 3.875,
+      "step": 1914880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8639455358915277e-05,
+      "loss": 3.8782,
+      "step": 1915392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.863106941140476e-05,
+      "loss": 3.8729,
+      "step": 1915904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.862268346389424e-05,
+      "loss": 3.8747,
+      "step": 1916416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.861429751638372e-05,
+      "loss": 3.8755,
+      "step": 1916928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.86059115688732e-05,
+      "loss": 3.8721,
+      "step": 1917440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8597525621362677e-05,
+      "loss": 3.8856,
+      "step": 1917952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8589139673852156e-05,
+      "loss": 3.8869,
+      "step": 1918464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8580753726341636e-05,
+      "loss": 3.888,
+      "step": 1918976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8572367778831116e-05,
+      "loss": 3.8796,
+      "step": 1919488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.85639818313206e-05,
+      "loss": 3.8806,
+      "step": 1920000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.855561226261381e-05,
+      "loss": 3.886,
+      "step": 1920512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.854722631510329e-05,
+      "loss": 3.8768,
+      "step": 1921024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.853884036759277e-05,
+      "loss": 3.8735,
+      "step": 1921536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8530470798885978e-05,
+      "loss": 3.8891,
+      "step": 1922048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.852208485137546e-05,
+      "loss": 3.8775,
+      "step": 1922560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.851369890386494e-05,
+      "loss": 3.8726,
+      "step": 1923072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.850531295635442e-05,
+      "loss": 3.8742,
+      "step": 1923584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.84969270088439e-05,
+      "loss": 3.8858,
+      "step": 1924096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.848855744013711e-05,
+      "loss": 3.887,
+      "step": 1924608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.848017149262659e-05,
+      "loss": 3.896,
+      "step": 1925120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.847178554511607e-05,
+      "loss": 3.8735,
+      "step": 1925632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8463399597605553e-05,
+      "loss": 3.8823,
+      "step": 1926144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.845501365009503e-05,
+      "loss": 3.89,
+      "step": 1926656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.844662770258451e-05,
+      "loss": 3.8656,
+      "step": 1927168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.843824175507399e-05,
+      "loss": 3.8763,
+      "step": 1927680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.842985580756347e-05,
+      "loss": 3.8746,
+      "step": 1928192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.842146986005295e-05,
+      "loss": 3.8774,
+      "step": 1928704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8413100291346162e-05,
+      "loss": 3.8827,
+      "step": 1929216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8404714343835642e-05,
+      "loss": 3.8828,
+      "step": 1929728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8396328396325122e-05,
+      "loss": 3.8835,
+      "step": 1930240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8387942448814602e-05,
+      "loss": 3.8789,
+      "step": 1930752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.837957288010781e-05,
+      "loss": 3.8795,
+      "step": 1931264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.837118693259729e-05,
+      "loss": 3.8769,
+      "step": 1931776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.836280098508677e-05,
+      "loss": 3.8792,
+      "step": 1932288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8354431416379984e-05,
+      "loss": 3.8496,
+      "step": 1932800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8346045468869464e-05,
+      "loss": 3.8954,
+      "step": 1933312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8337659521358944e-05,
+      "loss": 3.8756,
+      "step": 1933824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8329273573848424e-05,
+      "loss": 3.886,
+      "step": 1934336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8320887626337903e-05,
+      "loss": 3.875,
+      "step": 1934848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8312501678827383e-05,
+      "loss": 3.8783,
+      "step": 1935360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8304132110120596e-05,
+      "loss": 3.8658,
+      "step": 1935872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8295762541413805e-05,
+      "loss": 3.8708,
+      "step": 1936384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8287376593903285e-05,
+      "loss": 3.8866,
+      "step": 1936896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8278990646392765e-05,
+      "loss": 3.871,
+      "step": 1937408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8270604698882245e-05,
+      "loss": 3.8833,
+      "step": 1937920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8262218751371725e-05,
+      "loss": 3.8801,
+      "step": 1938432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8253832803861208e-05,
+      "loss": 3.8648,
+      "step": 1938944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8245446856350685e-05,
+      "loss": 3.8907,
+      "step": 1939456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8237060908840165e-05,
+      "loss": 3.8621,
+      "step": 1939968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8228674961329645e-05,
+      "loss": 3.8714,
+      "step": 1940480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8220289013819125e-05,
+      "loss": 3.8871,
+      "step": 1940992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8211919445112334e-05,
+      "loss": 3.8821,
+      "step": 1941504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8203533497601817e-05,
+      "loss": 3.8671,
+      "step": 1942016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8195147550091297e-05,
+      "loss": 3.8817,
+      "step": 1942528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8186761602580777e-05,
+      "loss": 3.8599,
+      "step": 1943040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8178375655070257e-05,
+      "loss": 3.8843,
+      "step": 1943552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8169989707559737e-05,
+      "loss": 3.8867,
+      "step": 1944064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8161603760049217e-05,
+      "loss": 3.8789,
+      "step": 1944576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8153217812538693e-05,
+      "loss": 3.8833,
+      "step": 1945088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8144831865028177e-05,
+      "loss": 3.8862,
+      "step": 1945600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.813647867512512e-05,
+      "loss": 3.896,
+      "step": 1946112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.81280927276146e-05,
+      "loss": 3.873,
+      "step": 1946624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8119723158907807e-05,
+      "loss": 3.8784,
+      "step": 1947136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8111337211397287e-05,
+      "loss": 3.886,
+      "step": 1947648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.810295126388677e-05,
+      "loss": 3.8576,
+      "step": 1948160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.809456531637625e-05,
+      "loss": 3.8818,
+      "step": 1948672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.808617936886573e-05,
+      "loss": 3.8828,
+      "step": 1949184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.807779342135521e-05,
+      "loss": 3.8817,
+      "step": 1949696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.806940747384469e-05,
+      "loss": 3.8974,
+      "step": 1950208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8061021526334167e-05,
+      "loss": 3.8695,
+      "step": 1950720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8052635578823647e-05,
+      "loss": 3.8786,
+      "step": 1951232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.804424963131313e-05,
+      "loss": 3.8805,
+      "step": 1951744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.803588006260634e-05,
+      "loss": 3.8803,
+      "step": 1952256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.802749411509582e-05,
+      "loss": 3.8795,
+      "step": 1952768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8019124546389032e-05,
+      "loss": 3.8769,
+      "step": 1953280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8010738598878512e-05,
+      "loss": 3.8722,
+      "step": 1953792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8002352651367992e-05,
+      "loss": 3.8808,
+      "step": 1954304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7993966703857472e-05,
+      "loss": 3.8735,
+      "step": 1954816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7985580756346952e-05,
+      "loss": 3.8766,
+      "step": 1955328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7977194808836432e-05,
+      "loss": 3.8851,
+      "step": 1955840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.796880886132591e-05,
+      "loss": 3.878,
+      "step": 1956352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.796042291381539e-05,
+      "loss": 3.8818,
+      "step": 1956864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7952036966304868e-05,
+      "loss": 3.8695,
+      "step": 1957376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.794366739759808e-05,
+      "loss": 3.875,
+      "step": 1957888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7935281450087564e-05,
+      "loss": 3.8761,
+      "step": 1958400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7926895502577044e-05,
+      "loss": 3.8869,
+      "step": 1958912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.791850955506652e-05,
+      "loss": 3.8763,
+      "step": 1959424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7910139986359733e-05,
+      "loss": 3.8741,
+      "step": 1959936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7901754038849213e-05,
+      "loss": 3.8804,
+      "step": 1960448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7893368091338693e-05,
+      "loss": 3.8822,
+      "step": 1960960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7884982143828173e-05,
+      "loss": 3.8649,
+      "step": 1961472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7876596196317653e-05,
+      "loss": 3.8785,
+      "step": 1961984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7868210248807133e-05,
+      "loss": 3.8645,
+      "step": 1962496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7859824301296613e-05,
+      "loss": 3.8828,
+      "step": 1963008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7851454732589822e-05,
+      "loss": 3.8708,
+      "step": 1963520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7843068785079302e-05,
+      "loss": 3.8748,
+      "step": 1964032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7834682837568785e-05,
+      "loss": 3.8643,
+      "step": 1964544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7826296890058265e-05,
+      "loss": 3.8839,
+      "step": 1965056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7817910942547745e-05,
+      "loss": 3.8761,
+      "step": 1965568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7809541373840954e-05,
+      "loss": 3.8699,
+      "step": 1966080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7801155426330434e-05,
+      "loss": 3.881,
+      "step": 1966592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7792769478819914e-05,
+      "loss": 3.8897,
+      "step": 1967104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7784383531309394e-05,
+      "loss": 3.8879,
+      "step": 1967616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7775997583798877e-05,
+      "loss": 3.8734,
+      "step": 1968128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7767644393895816e-05,
+      "loss": 3.8687,
+      "step": 1968640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7759258446385296e-05,
+      "loss": 3.8805,
+      "step": 1969152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7750872498874776e-05,
+      "loss": 3.8842,
+      "step": 1969664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7742502930167988e-05,
+      "loss": 3.8711,
+      "step": 1970176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7734116982657468e-05,
+      "loss": 3.8694,
+      "step": 1970688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7725731035146948e-05,
+      "loss": 3.8857,
+      "step": 1971200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7717345087636428e-05,
+      "loss": 3.8816,
+      "step": 1971712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7708959140125908e-05,
+      "loss": 3.8795,
+      "step": 1972224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7700573192615388e-05,
+      "loss": 3.8707,
+      "step": 1972736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7692187245104868e-05,
+      "loss": 3.8766,
+      "step": 1973248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7683801297594348e-05,
+      "loss": 3.8737,
+      "step": 1973760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7675415350083828e-05,
+      "loss": 3.8892,
+      "step": 1974272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7667029402573308e-05,
+      "loss": 3.876,
+      "step": 1974784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7658643455062788e-05,
+      "loss": 3.8836,
+      "step": 1975296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7650257507552268e-05,
+      "loss": 3.8654,
+      "step": 1975808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7641887938845477e-05,
+      "loss": 3.8924,
+      "step": 1976320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7633501991334957e-05,
+      "loss": 3.8718,
+      "step": 1976832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.762511604382444e-05,
+      "loss": 3.8876,
+      "step": 1977344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.761673009631392e-05,
+      "loss": 3.8788,
+      "step": 1977856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.76083441488034e-05,
+      "loss": 3.8815,
+      "step": 1978368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.759997458009661e-05,
+      "loss": 3.8743,
+      "step": 1978880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.759158863258609e-05,
+      "loss": 3.897,
+      "step": 1979392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.758320268507557e-05,
+      "loss": 3.8752,
+      "step": 1979904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.757481673756505e-05,
+      "loss": 3.8862,
+      "step": 1980416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7566430790054532e-05,
+      "loss": 3.8836,
+      "step": 1980928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.755804484254401e-05,
+      "loss": 3.8721,
+      "step": 1981440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.754965889503349e-05,
+      "loss": 3.8788,
+      "step": 1981952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.754127294752297e-05,
+      "loss": 3.8765,
+      "step": 1982464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7532903378816178e-05,
+      "loss": 3.881,
+      "step": 1982976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7524533810109394e-05,
+      "loss": 3.8828,
+      "step": 1983488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7516147862598874e-05,
+      "loss": 3.8831,
+      "step": 1984000
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9854087829589844,
+      "eval_runtime": 294.6349,
+      "eval_samples_per_second": 1295.132,
+      "eval_steps_per_second": 40.474,
+      "step": 1984320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7507761915088354e-05,
+      "loss": 3.8693,
+      "step": 1984512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749937596757783e-05,
+      "loss": 3.875,
+      "step": 1985024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749099002006731e-05,
+      "loss": 3.8741,
+      "step": 1985536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.748260407255679e-05,
+      "loss": 3.8782,
+      "step": 1986048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.747421812504627e-05,
+      "loss": 3.8903,
+      "step": 1986560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.746583217753575e-05,
+      "loss": 3.8798,
+      "step": 1987072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7457462608828962e-05,
+      "loss": 3.8785,
+      "step": 1987584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7449076661318442e-05,
+      "loss": 3.8734,
+      "step": 1988096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7440690713807922e-05,
+      "loss": 3.8797,
+      "step": 1988608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7432304766297402e-05,
+      "loss": 3.8862,
+      "step": 1989120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7423918818786882e-05,
+      "loss": 3.8743,
+      "step": 1989632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7415532871276362e-05,
+      "loss": 3.8755,
+      "step": 1990144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7407163302569575e-05,
+      "loss": 3.8851,
+      "step": 1990656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7398777355059055e-05,
+      "loss": 3.87,
+      "step": 1991168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7390391407548535e-05,
+      "loss": 3.8738,
+      "step": 1991680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.738200546003801e-05,
+      "loss": 3.8626,
+      "step": 1992192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.737361951252749e-05,
+      "loss": 3.8737,
+      "step": 1992704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.736523356501697e-05,
+      "loss": 3.8716,
+      "step": 1993216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7356863996310184e-05,
+      "loss": 3.8654,
+      "step": 1993728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7348478048799663e-05,
+      "loss": 3.881,
+      "step": 1994240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7340092101289143e-05,
+      "loss": 3.8818,
+      "step": 1994752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7331706153778623e-05,
+      "loss": 3.8803,
+      "step": 1995264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7323320206268103e-05,
+      "loss": 3.8746,
+      "step": 1995776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7314934258757583e-05,
+      "loss": 3.8758,
+      "step": 1996288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7306548311247063e-05,
+      "loss": 3.8816,
+      "step": 1996800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7298162363736543e-05,
+      "loss": 3.8714,
+      "step": 1997312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7289792795029756e-05,
+      "loss": 3.8655,
+      "step": 1997824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7281406847519236e-05,
+      "loss": 3.8849,
+      "step": 1998336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7273020900008716e-05,
+      "loss": 3.8726,
+      "step": 1998848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7264634952498192e-05,
+      "loss": 3.8692,
+      "step": 1999360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7256249004987672e-05,
+      "loss": 3.8673,
+      "step": 1999872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7247863057477155e-05,
+      "loss": 3.8785,
+      "step": 2000384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7239477109966635e-05,
+      "loss": 3.8774,
+      "step": 2000896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7231091162456115e-05,
+      "loss": 3.8942,
+      "step": 2001408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7222721593749324e-05,
+      "loss": 3.8689,
+      "step": 2001920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7214335646238804e-05,
+      "loss": 3.8731,
+      "step": 2002432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7205949698728284e-05,
+      "loss": 3.8899,
+      "step": 2002944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7197563751217764e-05,
+      "loss": 3.8576,
+      "step": 2003456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7189177803707244e-05,
+      "loss": 3.872,
+      "step": 2003968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7180808235000457e-05,
+      "loss": 3.8718,
+      "step": 2004480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7172422287489937e-05,
+      "loss": 3.8675,
+      "step": 2004992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7164036339979417e-05,
+      "loss": 3.8807,
+      "step": 2005504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7155650392468897e-05,
+      "loss": 3.8779,
+      "step": 2006016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7147280823762106e-05,
+      "loss": 3.8811,
+      "step": 2006528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.713889487625159e-05,
+      "loss": 3.8723,
+      "step": 2007040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.713050892874107e-05,
+      "loss": 3.8783,
+      "step": 2007552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7122139360034278e-05,
+      "loss": 3.8672,
+      "step": 2008064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7113753412523758e-05,
+      "loss": 3.8804,
+      "step": 2008576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7105367465013238e-05,
+      "loss": 3.8471,
+      "step": 2009088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7096981517502718e-05,
+      "loss": 3.8867,
+      "step": 2009600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7088595569992198e-05,
+      "loss": 3.872,
+      "step": 2010112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.708022600128541e-05,
+      "loss": 3.8815,
+      "step": 2010624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.707184005377489e-05,
+      "loss": 3.8711,
+      "step": 2011136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.706345410626437e-05,
+      "loss": 3.8719,
+      "step": 2011648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7055068158753847e-05,
+      "loss": 3.8611,
+      "step": 2012160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7046682211243327e-05,
+      "loss": 3.8656,
+      "step": 2012672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.703829626373281e-05,
+      "loss": 3.8817,
+      "step": 2013184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.702992669502602e-05,
+      "loss": 3.8628,
+      "step": 2013696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.70215407475155e-05,
+      "loss": 3.8774,
+      "step": 2014208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.701315480000498e-05,
+      "loss": 3.8783,
+      "step": 2014720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7004785231298192e-05,
+      "loss": 3.8572,
+      "step": 2015232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.699639928378767e-05,
+      "loss": 3.8844,
+      "step": 2015744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.698801333627715e-05,
+      "loss": 3.8612,
+      "step": 2016256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.697962738876663e-05,
+      "loss": 3.8656,
+      "step": 2016768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.697124144125611e-05,
+      "loss": 3.8808,
+      "step": 2017280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.696285549374559e-05,
+      "loss": 3.8749,
+      "step": 2017792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.695446954623507e-05,
+      "loss": 3.8643,
+      "step": 2018304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.694608359872455e-05,
+      "loss": 3.877,
+      "step": 2018816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6937697651214028e-05,
+      "loss": 3.8582,
+      "step": 2019328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.692931170370351e-05,
+      "loss": 3.8768,
+      "step": 2019840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.692092575619299e-05,
+      "loss": 3.8815,
+      "step": 2020352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.691253980868247e-05,
+      "loss": 3.8772,
+      "step": 2020864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.690417023997568e-05,
+      "loss": 3.8783,
+      "step": 2021376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.689578429246516e-05,
+      "loss": 3.8809,
+      "step": 2021888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.688739834495464e-05,
+      "loss": 3.8872,
+      "step": 2022400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.687901239744412e-05,
+      "loss": 3.8696,
+      "step": 2022912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6870626449933604e-05,
+      "loss": 3.8756,
+      "step": 2023424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6862256881226813e-05,
+      "loss": 3.8776,
+      "step": 2023936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6853870933716293e-05,
+      "loss": 3.8577,
+      "step": 2024448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6845484986205773e-05,
+      "loss": 3.8736,
+      "step": 2024960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6837099038695252e-05,
+      "loss": 3.8785,
+      "step": 2025472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6828729469988465e-05,
+      "loss": 3.8747,
+      "step": 2025984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6820359901281674e-05,
+      "loss": 3.8959,
+      "step": 2026496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6811973953771154e-05,
+      "loss": 3.8621,
+      "step": 2027008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6803604385064367e-05,
+      "loss": 3.874,
+      "step": 2027520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6795218437553846e-05,
+      "loss": 3.8778,
+      "step": 2028032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6786832490043326e-05,
+      "loss": 3.8755,
+      "step": 2028544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6778446542532806e-05,
+      "loss": 3.8759,
+      "step": 2029056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6770060595022286e-05,
+      "loss": 3.8733,
+      "step": 2029568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6761674647511766e-05,
+      "loss": 3.8637,
+      "step": 2030080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6753288700001246e-05,
+      "loss": 3.8727,
+      "step": 2030592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6744902752490726e-05,
+      "loss": 3.8748,
+      "step": 2031104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6736516804980203e-05,
+      "loss": 3.8669,
+      "step": 2031616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6728130857469683e-05,
+      "loss": 3.8785,
+      "step": 2032128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6719744909959166e-05,
+      "loss": 3.8788,
+      "step": 2032640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6711358962448646e-05,
+      "loss": 3.871,
+      "step": 2033152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6702989393741855e-05,
+      "loss": 3.8654,
+      "step": 2033664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6694619825035068e-05,
+      "loss": 3.8699,
+      "step": 2034176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6686233877524548e-05,
+      "loss": 3.8705,
+      "step": 2034688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6677847930014028e-05,
+      "loss": 3.8843,
+      "step": 2035200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6669461982503507e-05,
+      "loss": 3.8733,
+      "step": 2035712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6661076034992987e-05,
+      "loss": 3.8676,
+      "step": 2036224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6652690087482467e-05,
+      "loss": 3.8756,
+      "step": 2036736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6644304139971947e-05,
+      "loss": 3.8703,
+      "step": 2037248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6635918192461427e-05,
+      "loss": 3.8687,
+      "step": 2037760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6627532244950907e-05,
+      "loss": 3.8719,
+      "step": 2038272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.661916267624412e-05,
+      "loss": 3.8584,
+      "step": 2038784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.66107767287336e-05,
+      "loss": 3.8799,
+      "step": 2039296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.660239078122308e-05,
+      "loss": 3.8696,
+      "step": 2039808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.659400483371256e-05,
+      "loss": 3.8651,
+      "step": 2040320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.658563526500577e-05,
+      "loss": 3.8591,
+      "step": 2040832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.657724931749525e-05,
+      "loss": 3.877,
+      "step": 2041344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.656887974878846e-05,
+      "loss": 3.8743,
+      "step": 2041856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.656049380127794e-05,
+      "loss": 3.8642,
+      "step": 2042368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.655210785376742e-05,
+      "loss": 3.8762,
+      "step": 2042880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.65437219062569e-05,
+      "loss": 3.8867,
+      "step": 2043392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.653533595874638e-05,
+      "loss": 3.8798,
+      "step": 2043904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6526950011235858e-05,
+      "loss": 3.8715,
+      "step": 2044416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6518564063725337e-05,
+      "loss": 3.8647,
+      "step": 2044928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6510194495018553e-05,
+      "loss": 3.872,
+      "step": 2045440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6501808547508033e-05,
+      "loss": 3.8823,
+      "step": 2045952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6493438978801242e-05,
+      "loss": 3.8664,
+      "step": 2046464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6485053031290722e-05,
+      "loss": 3.864,
+      "step": 2046976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6476667083780202e-05,
+      "loss": 3.8804,
+      "step": 2047488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6468281136269682e-05,
+      "loss": 3.8735,
+      "step": 2048000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6459895188759162e-05,
+      "loss": 3.8744,
+      "step": 2048512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6451509241248642e-05,
+      "loss": 3.8689,
+      "step": 2049024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6443123293738122e-05,
+      "loss": 3.8701,
+      "step": 2049536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6434737346227602e-05,
+      "loss": 3.8686,
+      "step": 2050048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6426351398717082e-05,
+      "loss": 3.8842,
+      "step": 2050560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6417965451206562e-05,
+      "loss": 3.8698,
+      "step": 2051072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6409579503696042e-05,
+      "loss": 3.8797,
+      "step": 2051584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6401193556185522e-05,
+      "loss": 3.8612,
+      "step": 2052096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6392823987478734e-05,
+      "loss": 3.8844,
+      "step": 2052608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6384438039968214e-05,
+      "loss": 3.8734,
+      "step": 2053120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6376068471261423e-05,
+      "loss": 3.8798,
+      "step": 2053632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6367682523750903e-05,
+      "loss": 3.8762,
+      "step": 2054144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6359296576240383e-05,
+      "loss": 3.8733,
+      "step": 2054656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6350910628729867e-05,
+      "loss": 3.875,
+      "step": 2055168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6342524681219343e-05,
+      "loss": 3.887,
+      "step": 2055680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6334138733708823e-05,
+      "loss": 3.8727,
+      "step": 2056192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6325752786198303e-05,
+      "loss": 3.8749,
+      "step": 2056704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6317366838687783e-05,
+      "loss": 3.8816,
+      "step": 2057216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6308997269980996e-05,
+      "loss": 3.8695,
+      "step": 2057728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6300611322470476e-05,
+      "loss": 3.8726,
+      "step": 2058240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6292225374959956e-05,
+      "loss": 3.8747,
+      "step": 2058752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6283839427449436e-05,
+      "loss": 3.874,
+      "step": 2059264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6275469858742645e-05,
+      "loss": 3.8753,
+      "step": 2059776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6267100290035857e-05,
+      "loss": 3.8796,
+      "step": 2060288
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9838733673095703,
+      "eval_runtime": 293.6565,
+      "eval_samples_per_second": 1299.447,
+      "eval_steps_per_second": 40.609,
+      "step": 2060640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6258714342525337e-05,
+      "loss": 3.8618,
+      "step": 2060800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6250328395014817e-05,
+      "loss": 3.8704,
+      "step": 2061312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6241942447504297e-05,
+      "loss": 3.8706,
+      "step": 2061824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6233556499993777e-05,
+      "loss": 3.8741,
+      "step": 2062336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6225170552483257e-05,
+      "loss": 3.8838,
+      "step": 2062848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6216784604972737e-05,
+      "loss": 3.8763,
+      "step": 2063360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6208398657462217e-05,
+      "loss": 3.8735,
+      "step": 2063872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6200012709951697e-05,
+      "loss": 3.872,
+      "step": 2064384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6191626762441177e-05,
+      "loss": 3.8749,
+      "step": 2064896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6183240814930657e-05,
+      "loss": 3.875,
+      "step": 2065408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6174854867420137e-05,
+      "loss": 3.8751,
+      "step": 2065920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6166468919909617e-05,
+      "loss": 3.8696,
+      "step": 2066432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6158099351202826e-05,
+      "loss": 3.88,
+      "step": 2066944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6149713403692306e-05,
+      "loss": 3.866,
+      "step": 2067456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.614132745618179e-05,
+      "loss": 3.8721,
+      "step": 2067968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.613294150867127e-05,
+      "loss": 3.8581,
+      "step": 2068480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.612455556116075e-05,
+      "loss": 3.8719,
+      "step": 2068992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6116169613650225e-05,
+      "loss": 3.8669,
+      "step": 2069504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6107783666139705e-05,
+      "loss": 3.8578,
+      "step": 2070016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6099397718629185e-05,
+      "loss": 3.876,
+      "step": 2070528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6091011771118665e-05,
+      "loss": 3.8758,
+      "step": 2071040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6082625823608145e-05,
+      "loss": 3.8728,
+      "step": 2071552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.607423987609763e-05,
+      "loss": 3.8731,
+      "step": 2072064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.606585392858711e-05,
+      "loss": 3.8707,
+      "step": 2072576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6057467981076585e-05,
+      "loss": 3.8804,
+      "step": 2073088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6049098412369798e-05,
+      "loss": 3.8709,
+      "step": 2073600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6040728843663007e-05,
+      "loss": 3.8547,
+      "step": 2074112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.603234289615249e-05,
+      "loss": 3.8796,
+      "step": 2074624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.602395694864197e-05,
+      "loss": 3.8681,
+      "step": 2075136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.601557100113145e-05,
+      "loss": 3.8647,
+      "step": 2075648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.600720143242466e-05,
+      "loss": 3.8695,
+      "step": 2076160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.599881548491414e-05,
+      "loss": 3.8729,
+      "step": 2076672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.599042953740362e-05,
+      "loss": 3.8712,
+      "step": 2077184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.59820435898931e-05,
+      "loss": 3.8903,
+      "step": 2077696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5973657642382582e-05,
+      "loss": 3.8662,
+      "step": 2078208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.596527169487206e-05,
+      "loss": 3.8709,
+      "step": 2078720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.595688574736154e-05,
+      "loss": 3.8848,
+      "step": 2079232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.594849979985102e-05,
+      "loss": 3.8502,
+      "step": 2079744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.59401138523405e-05,
+      "loss": 3.8684,
+      "step": 2080256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5931744283633708e-05,
+      "loss": 3.8663,
+      "step": 2080768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.592335833612319e-05,
+      "loss": 3.8606,
+      "step": 2081280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.591497238861267e-05,
+      "loss": 3.8755,
+      "step": 2081792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.590658644110215e-05,
+      "loss": 3.8711,
+      "step": 2082304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.589821687239536e-05,
+      "loss": 3.8729,
+      "step": 2082816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.588983092488484e-05,
+      "loss": 3.8684,
+      "step": 2083328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.588144497737432e-05,
+      "loss": 3.8719,
+      "step": 2083840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.58730590298638e-05,
+      "loss": 3.8602,
+      "step": 2084352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5864673082353283e-05,
+      "loss": 3.8798,
+      "step": 2084864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5856303513646492e-05,
+      "loss": 3.8423,
+      "step": 2085376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5847917566135972e-05,
+      "loss": 3.8768,
+      "step": 2085888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5839531618625452e-05,
+      "loss": 3.869,
+      "step": 2086400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.583116204991866e-05,
+      "loss": 3.8799,
+      "step": 2086912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5822776102408145e-05,
+      "loss": 3.8637,
+      "step": 2087424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5814390154897625e-05,
+      "loss": 3.8689,
+      "step": 2087936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5806004207387105e-05,
+      "loss": 3.8581,
+      "step": 2088448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5797651017484046e-05,
+      "loss": 3.8601,
+      "step": 2088960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5789265069973526e-05,
+      "loss": 3.8818,
+      "step": 2089472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5780879122463006e-05,
+      "loss": 3.8547,
+      "step": 2089984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5772493174952486e-05,
+      "loss": 3.8727,
+      "step": 2090496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5764107227441966e-05,
+      "loss": 3.8737,
+      "step": 2091008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5755721279931446e-05,
+      "loss": 3.8508,
+      "step": 2091520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5747335332420926e-05,
+      "loss": 3.883,
+      "step": 2092032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5738949384910406e-05,
+      "loss": 3.8548,
+      "step": 2092544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5730563437399886e-05,
+      "loss": 3.8607,
+      "step": 2093056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5722177489889363e-05,
+      "loss": 3.8744,
+      "step": 2093568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.571380792118258e-05,
+      "loss": 3.8723,
+      "step": 2094080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.570542197367206e-05,
+      "loss": 3.8589,
+      "step": 2094592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5697036026161535e-05,
+      "loss": 3.8725,
+      "step": 2095104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5688650078651015e-05,
+      "loss": 3.8503,
+      "step": 2095616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5680280509944227e-05,
+      "loss": 3.8736,
+      "step": 2096128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5671894562433707e-05,
+      "loss": 3.8778,
+      "step": 2096640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5663508614923187e-05,
+      "loss": 3.8745,
+      "step": 2097152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5655122667412667e-05,
+      "loss": 3.8715,
+      "step": 2097664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5646736719902147e-05,
+      "loss": 3.8737,
+      "step": 2098176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5638350772391627e-05,
+      "loss": 3.886,
+      "step": 2098688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5629964824881107e-05,
+      "loss": 3.8657,
+      "step": 2099200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5621578877370587e-05,
+      "loss": 3.8756,
+      "step": 2099712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5613192929860067e-05,
+      "loss": 3.8675,
+      "step": 2100224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5604806982349547e-05,
+      "loss": 3.8594,
+      "step": 2100736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5596421034839027e-05,
+      "loss": 3.8668,
+      "step": 2101248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.558805146613224e-05,
+      "loss": 3.8724,
+      "step": 2101760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.557968189742545e-05,
+      "loss": 3.8704,
+      "step": 2102272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.557129594991493e-05,
+      "loss": 3.8908,
+      "step": 2102784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.556291000240441e-05,
+      "loss": 3.8559,
+      "step": 2103296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5554524054893892e-05,
+      "loss": 3.8694,
+      "step": 2103808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.554613810738337e-05,
+      "loss": 3.8726,
+      "step": 2104320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.553776853867658e-05,
+      "loss": 3.8698,
+      "step": 2104832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.552938259116606e-05,
+      "loss": 3.8687,
+      "step": 2105344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.552101302245927e-05,
+      "loss": 3.8652,
+      "step": 2105856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5512627074948753e-05,
+      "loss": 3.8629,
+      "step": 2106368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5504241127438233e-05,
+      "loss": 3.868,
+      "step": 2106880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5495855179927713e-05,
+      "loss": 3.8696,
+      "step": 2107392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.548746923241719e-05,
+      "loss": 3.8614,
+      "step": 2107904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.547908328490667e-05,
+      "loss": 3.8747,
+      "step": 2108416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.547069733739615e-05,
+      "loss": 3.8751,
+      "step": 2108928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.546231138988563e-05,
+      "loss": 3.8669,
+      "step": 2109440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.545392544237511e-05,
+      "loss": 3.8629,
+      "step": 2109952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5445539494864593e-05,
+      "loss": 3.8673,
+      "step": 2110464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5437153547354073e-05,
+      "loss": 3.8655,
+      "step": 2110976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.542876759984355e-05,
+      "loss": 3.8772,
+      "step": 2111488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5420398031136762e-05,
+      "loss": 3.8686,
+      "step": 2112000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5412012083626242e-05,
+      "loss": 3.8613,
+      "step": 2112512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5403626136115722e-05,
+      "loss": 3.8713,
+      "step": 2113024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5395240188605202e-05,
+      "loss": 3.8688,
+      "step": 2113536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5386870619898414e-05,
+      "loss": 3.8675,
+      "step": 2114048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5378484672387894e-05,
+      "loss": 3.8655,
+      "step": 2114560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.537009872487737e-05,
+      "loss": 3.8559,
+      "step": 2115072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.536171277736685e-05,
+      "loss": 3.8734,
+      "step": 2115584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5353343208660063e-05,
+      "loss": 3.8683,
+      "step": 2116096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5344957261149547e-05,
+      "loss": 3.8579,
+      "step": 2116608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5336571313639023e-05,
+      "loss": 3.8541,
+      "step": 2117120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5328185366128503e-05,
+      "loss": 3.8716,
+      "step": 2117632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5319815797421716e-05,
+      "loss": 3.8716,
+      "step": 2118144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5311429849911192e-05,
+      "loss": 3.859,
+      "step": 2118656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5303043902400675e-05,
+      "loss": 3.8737,
+      "step": 2119168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5294657954890155e-05,
+      "loss": 3.8816,
+      "step": 2119680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5286272007379635e-05,
+      "loss": 3.8722,
+      "step": 2120192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5277886059869115e-05,
+      "loss": 3.8678,
+      "step": 2120704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5269500112358595e-05,
+      "loss": 3.8617,
+      "step": 2121216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5261114164848075e-05,
+      "loss": 3.8688,
+      "step": 2121728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5252744596141286e-05,
+      "loss": 3.8762,
+      "step": 2122240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5244375027434497e-05,
+      "loss": 3.8631,
+      "step": 2122752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5235989079923977e-05,
+      "loss": 3.8586,
+      "step": 2123264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5227603132413457e-05,
+      "loss": 3.8718,
+      "step": 2123776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5219217184902937e-05,
+      "loss": 3.8689,
+      "step": 2124288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5210831237392417e-05,
+      "loss": 3.8704,
+      "step": 2124800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5202461668685627e-05,
+      "loss": 3.8661,
+      "step": 2125312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5194075721175107e-05,
+      "loss": 3.8633,
+      "step": 2125824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5185689773664587e-05,
+      "loss": 3.8649,
+      "step": 2126336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5177303826154069e-05,
+      "loss": 3.8781,
+      "step": 2126848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5168917878643549e-05,
+      "loss": 3.8682,
+      "step": 2127360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5160531931133026e-05,
+      "loss": 3.8757,
+      "step": 2127872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5152145983622507e-05,
+      "loss": 3.862,
+      "step": 2128384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5143760036111987e-05,
+      "loss": 3.8752,
+      "step": 2128896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.51353904674052e-05,
+      "loss": 3.8722,
+      "step": 2129408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5127004519894678e-05,
+      "loss": 3.8713,
+      "step": 2129920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.511863495118789e-05,
+      "loss": 3.8764,
+      "step": 2130432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.511024900367737e-05,
+      "loss": 3.8685,
+      "step": 2130944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5101863056166849e-05,
+      "loss": 3.8699,
+      "step": 2131456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5093477108656329e-05,
+      "loss": 3.8831,
+      "step": 2131968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5085091161145808e-05,
+      "loss": 3.8686,
+      "step": 2132480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.507670521363529e-05,
+      "loss": 3.8688,
+      "step": 2132992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.506831926612477e-05,
+      "loss": 3.8766,
+      "step": 2133504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.505993331861425e-05,
+      "loss": 3.8642,
+      "step": 2134016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.505154737110373e-05,
+      "loss": 3.8685,
+      "step": 2134528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.504317780239694e-05,
+      "loss": 3.8742,
+      "step": 2135040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5034808233690152e-05,
+      "loss": 3.8648,
+      "step": 2135552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5026422286179632e-05,
+      "loss": 3.8699,
+      "step": 2136064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5018036338669111e-05,
+      "loss": 3.8751,
+      "step": 2136576
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9823312759399414,
+      "eval_runtime": 295.8205,
+      "eval_samples_per_second": 1289.941,
+      "eval_steps_per_second": 40.312,
+      "step": 2136960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5009650391158591e-05,
+      "loss": 3.8643,
+      "step": 2137088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5001264443648071e-05,
+      "loss": 3.8641,
+      "step": 2137600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4992878496137553e-05,
+      "loss": 3.866,
+      "step": 2138112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4984508927430762e-05,
+      "loss": 3.8725,
+      "step": 2138624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4976122979920242e-05,
+      "loss": 3.8739,
+      "step": 2139136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4967737032409724e-05,
+      "loss": 3.8734,
+      "step": 2139648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4959351084899204e-05,
+      "loss": 3.8726,
+      "step": 2140160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.495096513738868e-05,
+      "loss": 3.8661,
+      "step": 2140672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4942579189878162e-05,
+      "loss": 3.8671,
+      "step": 2141184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4934193242367642e-05,
+      "loss": 3.8723,
+      "step": 2141696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4925807294857122e-05,
+      "loss": 3.8718,
+      "step": 2142208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4917437726150333e-05,
+      "loss": 3.86,
+      "step": 2142720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4909068157443545e-05,
+      "loss": 3.8789,
+      "step": 2143232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4900682209933025e-05,
+      "loss": 3.8613,
+      "step": 2143744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4892296262422503e-05,
+      "loss": 3.8624,
+      "step": 2144256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4883910314911983e-05,
+      "loss": 3.8565,
+      "step": 2144768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4875524367401463e-05,
+      "loss": 3.8666,
+      "step": 2145280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4867138419890945e-05,
+      "loss": 3.861,
+      "step": 2145792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4858752472380425e-05,
+      "loss": 3.8574,
+      "step": 2146304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4850366524869905e-05,
+      "loss": 3.8661,
+      "step": 2146816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4841980577359385e-05,
+      "loss": 3.8773,
+      "step": 2147328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4833611008652596e-05,
+      "loss": 3.8678,
+      "step": 2147840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4825225061142076e-05,
+      "loss": 3.8662,
+      "step": 2148352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4816839113631555e-05,
+      "loss": 3.8694,
+      "step": 2148864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4808453166121034e-05,
+      "loss": 3.8744,
+      "step": 2149376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4800083597414246e-05,
+      "loss": 3.8625,
+      "step": 2149888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4791714028707457e-05,
+      "loss": 3.855,
+      "step": 2150400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4783328081196937e-05,
+      "loss": 3.8742,
+      "step": 2150912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4774942133686417e-05,
+      "loss": 3.863,
+      "step": 2151424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4766556186175899e-05,
+      "loss": 3.8584,
+      "step": 2151936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4758186617469108e-05,
+      "loss": 3.8638,
+      "step": 2152448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4749800669958588e-05,
+      "loss": 3.8647,
+      "step": 2152960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.474141472244807e-05,
+      "loss": 3.8723,
+      "step": 2153472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.473302877493755e-05,
+      "loss": 3.8846,
+      "step": 2153984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.472464282742703e-05,
+      "loss": 3.8576,
+      "step": 2154496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4716256879916507e-05,
+      "loss": 3.8695,
+      "step": 2155008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4707870932405987e-05,
+      "loss": 3.8794,
+      "step": 2155520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.46995013636992e-05,
+      "loss": 3.8457,
+      "step": 2156032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.469111541618868e-05,
+      "loss": 3.8651,
+      "step": 2156544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4682729468678158e-05,
+      "loss": 3.8635,
+      "step": 2157056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4674343521167638e-05,
+      "loss": 3.8519,
+      "step": 2157568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4665957573657118e-05,
+      "loss": 3.8786,
+      "step": 2158080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.46575716261466e-05,
+      "loss": 3.864,
+      "step": 2158592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.464918567863608e-05,
+      "loss": 3.8701,
+      "step": 2159104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.464079973112556e-05,
+      "loss": 3.8641,
+      "step": 2159616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.463243016241877e-05,
+      "loss": 3.8686,
+      "step": 2160128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.462404421490825e-05,
+      "loss": 3.8547,
+      "step": 2160640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.461565826739773e-05,
+      "loss": 3.8763,
+      "step": 2161152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4607288698690941e-05,
+      "loss": 3.8413,
+      "step": 2161664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4598902751180421e-05,
+      "loss": 3.8696,
+      "step": 2162176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4590516803669901e-05,
+      "loss": 3.8644,
+      "step": 2162688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4582130856159381e-05,
+      "loss": 3.8774,
+      "step": 2163200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4573744908648863e-05,
+      "loss": 3.8561,
+      "step": 2163712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4565375339942072e-05,
+      "loss": 3.8611,
+      "step": 2164224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4556989392431553e-05,
+      "loss": 3.8592,
+      "step": 2164736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4548619823724762e-05,
+      "loss": 3.8556,
+      "step": 2165248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4540233876214242e-05,
+      "loss": 3.8773,
+      "step": 2165760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4531847928703724e-05,
+      "loss": 3.8466,
+      "step": 2166272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4523461981193204e-05,
+      "loss": 3.8715,
+      "step": 2166784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4515076033682684e-05,
+      "loss": 3.871,
+      "step": 2167296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4506690086172162e-05,
+      "loss": 3.8464,
+      "step": 2167808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4498304138661642e-05,
+      "loss": 3.8765,
+      "step": 2168320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4489918191151122e-05,
+      "loss": 3.8515,
+      "step": 2168832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4481532243640602e-05,
+      "loss": 3.8551,
+      "step": 2169344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4473146296130084e-05,
+      "loss": 3.8701,
+      "step": 2169856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4464776727423293e-05,
+      "loss": 3.8701,
+      "step": 2170368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4456390779912773e-05,
+      "loss": 3.8547,
+      "step": 2170880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4448004832402254e-05,
+      "loss": 3.8677,
+      "step": 2171392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4439618884891734e-05,
+      "loss": 3.8473,
+      "step": 2171904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4431249316184945e-05,
+      "loss": 3.8655,
+      "step": 2172416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4422863368674425e-05,
+      "loss": 3.8756,
+      "step": 2172928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4414477421163905e-05,
+      "loss": 3.8676,
+      "step": 2173440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4406091473653385e-05,
+      "loss": 3.8702,
+      "step": 2173952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4397705526142865e-05,
+      "loss": 3.8674,
+      "step": 2174464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4389319578632343e-05,
+      "loss": 3.8808,
+      "step": 2174976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4380933631121823e-05,
+      "loss": 3.8624,
+      "step": 2175488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4372547683611303e-05,
+      "loss": 3.8711,
+      "step": 2176000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4364161736100785e-05,
+      "loss": 3.8619,
+      "step": 2176512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4355792167393994e-05,
+      "loss": 3.8512,
+      "step": 2177024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4347406219883474e-05,
+      "loss": 3.8628,
+      "step": 2177536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4339020272372955e-05,
+      "loss": 3.8689,
+      "step": 2178048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4330634324862435e-05,
+      "loss": 3.8686,
+      "step": 2178560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4322264756155646e-05,
+      "loss": 3.884,
+      "step": 2179072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4313895187448859e-05,
+      "loss": 3.858,
+      "step": 2179584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4305509239938339e-05,
+      "loss": 3.8622,
+      "step": 2180096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4297123292427817e-05,
+      "loss": 3.8668,
+      "step": 2180608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4288737344917297e-05,
+      "loss": 3.8722,
+      "step": 2181120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4280351397406777e-05,
+      "loss": 3.8636,
+      "step": 2181632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4271965449896257e-05,
+      "loss": 3.8602,
+      "step": 2182144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4263579502385738e-05,
+      "loss": 3.8616,
+      "step": 2182656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4255193554875218e-05,
+      "loss": 3.8621,
+      "step": 2183168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4246840364972162e-05,
+      "loss": 3.8645,
+      "step": 2183680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4238454417461638e-05,
+      "loss": 3.8591,
+      "step": 2184192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4230068469951118e-05,
+      "loss": 3.8713,
+      "step": 2184704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.42216825224406e-05,
+      "loss": 3.8716,
+      "step": 2185216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.421329657493008e-05,
+      "loss": 3.8633,
+      "step": 2185728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.420491062741956e-05,
+      "loss": 3.86,
+      "step": 2186240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.419652467990904e-05,
+      "loss": 3.8604,
+      "step": 2186752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.418813873239852e-05,
+      "loss": 3.8619,
+      "step": 2187264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4179752784887998e-05,
+      "loss": 3.8762,
+      "step": 2187776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.417138321618121e-05,
+      "loss": 3.862,
+      "step": 2188288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4162997268670692e-05,
+      "loss": 3.8571,
+      "step": 2188800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4154611321160172e-05,
+      "loss": 3.8678,
+      "step": 2189312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4146225373649649e-05,
+      "loss": 3.862,
+      "step": 2189824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.413783942613913e-05,
+      "loss": 3.865,
+      "step": 2190336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4129486236236072e-05,
+      "loss": 3.8617,
+      "step": 2190848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4121100288725554e-05,
+      "loss": 3.8498,
+      "step": 2191360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4112714341215034e-05,
+      "loss": 3.869,
+      "step": 2191872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4104328393704513e-05,
+      "loss": 3.8614,
+      "step": 2192384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4095942446193993e-05,
+      "loss": 3.8567,
+      "step": 2192896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4087572877487204e-05,
+      "loss": 3.8485,
+      "step": 2193408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4079186929976684e-05,
+      "loss": 3.8666,
+      "step": 2193920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4070800982466164e-05,
+      "loss": 3.8683,
+      "step": 2194432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4062415034955644e-05,
+      "loss": 3.8508,
+      "step": 2194944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4054029087445122e-05,
+      "loss": 3.8703,
+      "step": 2195456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4045643139934602e-05,
+      "loss": 3.8743,
+      "step": 2195968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4037257192424082e-05,
+      "loss": 3.875,
+      "step": 2196480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4028871244913564e-05,
+      "loss": 3.8612,
+      "step": 2196992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4020501676206773e-05,
+      "loss": 3.857,
+      "step": 2197504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4012115728696255e-05,
+      "loss": 3.8628,
+      "step": 2198016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4003729781185735e-05,
+      "loss": 3.8759,
+      "step": 2198528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3995360212478945e-05,
+      "loss": 3.8573,
+      "step": 2199040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3986974264968425e-05,
+      "loss": 3.8601,
+      "step": 2199552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3978588317457905e-05,
+      "loss": 3.8597,
+      "step": 2200064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3970202369947385e-05,
+      "loss": 3.8698,
+      "step": 2200576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3961816422436865e-05,
+      "loss": 3.8656,
+      "step": 2201088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3953446853730076e-05,
+      "loss": 3.8615,
+      "step": 2201600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3945060906219556e-05,
+      "loss": 3.8555,
+      "step": 2202112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3936674958709036e-05,
+      "loss": 3.8584,
+      "step": 2202624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3928289011198518e-05,
+      "loss": 3.8762,
+      "step": 2203136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3919903063687998e-05,
+      "loss": 3.8611,
+      "step": 2203648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3911517116177474e-05,
+      "loss": 3.8724,
+      "step": 2204160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3903147547470688e-05,
+      "loss": 3.8526,
+      "step": 2204672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3894761599960168e-05,
+      "loss": 3.8736,
+      "step": 2205184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3886375652449648e-05,
+      "loss": 3.8684,
+      "step": 2205696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3877989704939126e-05,
+      "loss": 3.8662,
+      "step": 2206208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3869603757428606e-05,
+      "loss": 3.8716,
+      "step": 2206720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3861217809918086e-05,
+      "loss": 3.861,
+      "step": 2207232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3852848241211297e-05,
+      "loss": 3.8708,
+      "step": 2207744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.384447867250451e-05,
+      "loss": 3.8767,
+      "step": 2208256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.383609272499399e-05,
+      "loss": 3.8656,
+      "step": 2208768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3827706777483471e-05,
+      "loss": 3.865,
+      "step": 2209280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3819320829972948e-05,
+      "loss": 3.872,
+      "step": 2209792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3810934882462428e-05,
+      "loss": 3.8574,
+      "step": 2210304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.380254893495191e-05,
+      "loss": 3.8622,
+      "step": 2210816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.379416298744139e-05,
+      "loss": 3.8721,
+      "step": 2211328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.378577703993087e-05,
+      "loss": 3.8616,
+      "step": 2211840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.377740747122408e-05,
+      "loss": 3.8668,
+      "step": 2212352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.376902152371356e-05,
+      "loss": 3.8685,
+      "step": 2212864
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.98100209236145,
+      "eval_runtime": 296.2814,
+      "eval_samples_per_second": 1287.935,
+      "eval_steps_per_second": 40.249,
+      "step": 2213280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.376063557620304e-05,
+      "loss": 3.8658,
+      "step": 2213376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.375224962869252e-05,
+      "loss": 3.8572,
+      "step": 2213888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3743863681182002e-05,
+      "loss": 3.8618,
+      "step": 2214400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3735477733671478e-05,
+      "loss": 3.8682,
+      "step": 2214912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3727091786160958e-05,
+      "loss": 3.8711,
+      "step": 2215424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.371870583865044e-05,
+      "loss": 3.8685,
+      "step": 2215936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.371031989113992e-05,
+      "loss": 3.8619,
+      "step": 2216448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.37019339436294e-05,
+      "loss": 3.8644,
+      "step": 2216960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.369354799611888e-05,
+      "loss": 3.8619,
+      "step": 2217472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.368516204860836e-05,
+      "loss": 3.8744,
+      "step": 2217984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3676776101097838e-05,
+      "loss": 3.8629,
+      "step": 2218496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3668390153587318e-05,
+      "loss": 3.8589,
+      "step": 2219008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3660020584880532e-05,
+      "loss": 3.8759,
+      "step": 2219520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3651634637370012e-05,
+      "loss": 3.8545,
+      "step": 2220032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3643248689859489e-05,
+      "loss": 3.861,
+      "step": 2220544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.363486274234897e-05,
+      "loss": 3.8524,
+      "step": 2221056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.362647679483845e-05,
+      "loss": 3.8609,
+      "step": 2221568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.361809084732793e-05,
+      "loss": 3.8574,
+      "step": 2222080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.360970489981741e-05,
+      "loss": 3.8542,
+      "step": 2222592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.360131895230689e-05,
+      "loss": 3.8563,
+      "step": 2223104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3592933004796368e-05,
+      "loss": 3.8746,
+      "step": 2223616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3584547057285848e-05,
+      "loss": 3.8641,
+      "step": 2224128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3576161109775328e-05,
+      "loss": 3.8602,
+      "step": 2224640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.356777516226481e-05,
+      "loss": 3.8658,
+      "step": 2225152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3559405593558019e-05,
+      "loss": 3.873,
+      "step": 2225664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3551019646047499e-05,
+      "loss": 3.8545,
+      "step": 2226176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3542650077340713e-05,
+      "loss": 3.8542,
+      "step": 2226688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3534264129830193e-05,
+      "loss": 3.8699,
+      "step": 2227200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3525878182319671e-05,
+      "loss": 3.8571,
+      "step": 2227712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3517492234809151e-05,
+      "loss": 3.8558,
+      "step": 2228224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3509106287298631e-05,
+      "loss": 3.8595,
+      "step": 2228736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3500720339788111e-05,
+      "loss": 3.856,
+      "step": 2229248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3492334392277591e-05,
+      "loss": 3.8714,
+      "step": 2229760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3483948444767073e-05,
+      "loss": 3.8752,
+      "step": 2230272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3475578876060282e-05,
+      "loss": 3.859,
+      "step": 2230784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3467209307353493e-05,
+      "loss": 3.8649,
+      "step": 2231296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3458823359842973e-05,
+      "loss": 3.8746,
+      "step": 2231808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3450437412332453e-05,
+      "loss": 3.842,
+      "step": 2232320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3442051464821934e-05,
+      "loss": 3.8656,
+      "step": 2232832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3433665517311414e-05,
+      "loss": 3.8545,
+      "step": 2233344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3425279569800894e-05,
+      "loss": 3.8497,
+      "step": 2233856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3416893622290374e-05,
+      "loss": 3.8685,
+      "step": 2234368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3408524053583585e-05,
+      "loss": 3.8678,
+      "step": 2234880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3400138106073065e-05,
+      "loss": 3.8628,
+      "step": 2235392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3391752158562545e-05,
+      "loss": 3.8578,
+      "step": 2235904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3383366211052023e-05,
+      "loss": 3.8627,
+      "step": 2236416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3374980263541503e-05,
+      "loss": 3.8564,
+      "step": 2236928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3366594316030983e-05,
+      "loss": 3.8679,
+      "step": 2237440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3358208368520465e-05,
+      "loss": 3.8412,
+      "step": 2237952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3349822421009945e-05,
+      "loss": 3.8633,
+      "step": 2238464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3341452852303155e-05,
+      "loss": 3.8633,
+      "step": 2238976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3333066904792635e-05,
+      "loss": 3.8731,
+      "step": 2239488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3324680957282115e-05,
+      "loss": 3.8535,
+      "step": 2240000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3316311388575326e-05,
+      "loss": 3.8568,
+      "step": 2240512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3307925441064806e-05,
+      "loss": 3.8503,
+      "step": 2241024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3299539493554286e-05,
+      "loss": 3.8535,
+      "step": 2241536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3291153546043766e-05,
+      "loss": 3.8734,
+      "step": 2242048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3282767598533246e-05,
+      "loss": 3.8368,
+      "step": 2242560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.327441440863019e-05,
+      "loss": 3.8707,
+      "step": 2243072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3266028461119671e-05,
+      "loss": 3.867,
+      "step": 2243584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3257642513609147e-05,
+      "loss": 3.8468,
+      "step": 2244096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3249256566098627e-05,
+      "loss": 3.8712,
+      "step": 2244608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3240870618588107e-05,
+      "loss": 3.8448,
+      "step": 2245120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3232484671077589e-05,
+      "loss": 3.8493,
+      "step": 2245632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3224098723567069e-05,
+      "loss": 3.8634,
+      "step": 2246144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3215712776056549e-05,
+      "loss": 3.8667,
+      "step": 2246656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.320734320734976e-05,
+      "loss": 3.8481,
+      "step": 2247168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.319895725983924e-05,
+      "loss": 3.8708,
+      "step": 2247680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.319057131232872e-05,
+      "loss": 3.8406,
+      "step": 2248192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.31821853648182e-05,
+      "loss": 3.8576,
+      "step": 2248704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3173799417307678e-05,
+      "loss": 3.8727,
+      "step": 2249216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.316542984860089e-05,
+      "loss": 3.8642,
+      "step": 2249728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3157043901090372e-05,
+      "loss": 3.865,
+      "step": 2250240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3148657953579852e-05,
+      "loss": 3.8673,
+      "step": 2250752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3140272006069328e-05,
+      "loss": 3.8775,
+      "step": 2251264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.313188605855881e-05,
+      "loss": 3.8571,
+      "step": 2251776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.312350011104829e-05,
+      "loss": 3.8668,
+      "step": 2252288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.31151305423415e-05,
+      "loss": 3.8528,
+      "step": 2252800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.310674459483098e-05,
+      "loss": 3.8511,
+      "step": 2253312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.309835864732046e-05,
+      "loss": 3.8601,
+      "step": 2253824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.308997269980994e-05,
+      "loss": 3.8594,
+      "step": 2254336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.308158675229942e-05,
+      "loss": 3.8664,
+      "step": 2254848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3073200804788902e-05,
+      "loss": 3.8808,
+      "step": 2255360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3064814857278382e-05,
+      "loss": 3.8612,
+      "step": 2255872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3056445288571591e-05,
+      "loss": 3.8536,
+      "step": 2256384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3048059341061073e-05,
+      "loss": 3.858,
+      "step": 2256896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3039673393550553e-05,
+      "loss": 3.8691,
+      "step": 2257408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3031287446040033e-05,
+      "loss": 3.857,
+      "step": 2257920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3022901498529511e-05,
+      "loss": 3.8578,
+      "step": 2258432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3014515551018991e-05,
+      "loss": 3.8592,
+      "step": 2258944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3006129603508471e-05,
+      "loss": 3.8566,
+      "step": 2259456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2997743655997951e-05,
+      "loss": 3.8621,
+      "step": 2259968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2989374087291162e-05,
+      "loss": 3.8531,
+      "step": 2260480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2980988139780642e-05,
+      "loss": 3.8658,
+      "step": 2260992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2972602192270122e-05,
+      "loss": 3.8671,
+      "step": 2261504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2964216244759603e-05,
+      "loss": 3.8559,
+      "step": 2262016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2955846676052813e-05,
+      "loss": 3.8563,
+      "step": 2262528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2947460728542292e-05,
+      "loss": 3.856,
+      "step": 2263040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2939074781031774e-05,
+      "loss": 3.8593,
+      "step": 2263552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2930688833521254e-05,
+      "loss": 3.8712,
+      "step": 2264064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2922319264814465e-05,
+      "loss": 3.8607,
+      "step": 2264576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2913933317303945e-05,
+      "loss": 3.8492,
+      "step": 2265088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2905547369793425e-05,
+      "loss": 3.866,
+      "step": 2265600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2897161422282905e-05,
+      "loss": 3.857,
+      "step": 2266112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2888775474772385e-05,
+      "loss": 3.8553,
+      "step": 2266624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2880405906065595e-05,
+      "loss": 3.8585,
+      "step": 2267136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2872019958555075e-05,
+      "loss": 3.8475,
+      "step": 2267648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2863634011044557e-05,
+      "loss": 3.8636,
+      "step": 2268160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2855248063534037e-05,
+      "loss": 3.8586,
+      "step": 2268672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2846878494827246e-05,
+      "loss": 3.8512,
+      "step": 2269184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2838492547316728e-05,
+      "loss": 3.846,
+      "step": 2269696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2830106599806208e-05,
+      "loss": 3.8603,
+      "step": 2270208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2821720652295688e-05,
+      "loss": 3.8656,
+      "step": 2270720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2813351083588899e-05,
+      "loss": 3.8505,
+      "step": 2271232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2804965136078378e-05,
+      "loss": 3.863,
+      "step": 2271744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2796579188567858e-05,
+      "loss": 3.8663,
+      "step": 2272256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2788193241057337e-05,
+      "loss": 3.871,
+      "step": 2272768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.277982367235055e-05,
+      "loss": 3.8615,
+      "step": 2273280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2771437724840029e-05,
+      "loss": 3.85,
+      "step": 2273792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.276305177732951e-05,
+      "loss": 3.8583,
+      "step": 2274304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2754665829818987e-05,
+      "loss": 3.8704,
+      "step": 2274816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.274631263991593e-05,
+      "loss": 3.8546,
+      "step": 2275328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.273792669240541e-05,
+      "loss": 3.8568,
+      "step": 2275840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.272954074489489e-05,
+      "loss": 3.8578,
+      "step": 2276352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2721154797384372e-05,
+      "loss": 3.8607,
+      "step": 2276864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2712768849873852e-05,
+      "loss": 3.8586,
+      "step": 2277376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2704399281167061e-05,
+      "loss": 3.8629,
+      "step": 2277888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2696013333656543e-05,
+      "loss": 3.848,
+      "step": 2278400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2687627386146023e-05,
+      "loss": 3.8556,
+      "step": 2278912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2679241438635503e-05,
+      "loss": 3.8715,
+      "step": 2279424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2670871869928714e-05,
+      "loss": 3.8614,
+      "step": 2279936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2662485922418194e-05,
+      "loss": 3.8644,
+      "step": 2280448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2654099974907674e-05,
+      "loss": 3.8505,
+      "step": 2280960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2645714027397153e-05,
+      "loss": 3.8676,
+      "step": 2281472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2637328079886632e-05,
+      "loss": 3.8639,
+      "step": 2281984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2628942132376112e-05,
+      "loss": 3.8615,
+      "step": 2282496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2620556184865592e-05,
+      "loss": 3.8608,
+      "step": 2283008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2612186616158806e-05,
+      "loss": 3.8582,
+      "step": 2283520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2603800668648282e-05,
+      "loss": 3.8703,
+      "step": 2284032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2595431099941497e-05,
+      "loss": 3.8675,
+      "step": 2284544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2587045152430977e-05,
+      "loss": 3.8656,
+      "step": 2285056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2578659204920456e-05,
+      "loss": 3.8573,
+      "step": 2285568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2570273257409935e-05,
+      "loss": 3.8693,
+      "step": 2286080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2561887309899415e-05,
+      "loss": 3.8581,
+      "step": 2286592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2553501362388895e-05,
+      "loss": 3.8558,
+      "step": 2287104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2545115414878375e-05,
+      "loss": 3.8704,
+      "step": 2287616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2536729467367855e-05,
+      "loss": 3.8559,
+      "step": 2288128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2528343519857336e-05,
+      "loss": 3.868,
+      "step": 2288640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2519973951150545e-05,
+      "loss": 3.86,
+      "step": 2289152
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.97961688041687,
+      "eval_runtime": 293.5936,
+      "eval_samples_per_second": 1299.725,
+      "eval_steps_per_second": 40.617,
+      "step": 2289600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2511588003640027e-05,
+      "loss": 3.8547,
+      "step": 2289664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2503202056129507e-05,
+      "loss": 3.8574,
+      "step": 2290176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2494816108618985e-05,
+      "loss": 3.8598,
+      "step": 2290688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2486446539912198e-05,
+      "loss": 3.8617,
+      "step": 2291200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2478060592401678e-05,
+      "loss": 3.8681,
+      "step": 2291712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2469674644891156e-05,
+      "loss": 3.8655,
+      "step": 2292224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2461305076184368e-05,
+      "loss": 3.8591,
+      "step": 2292736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2452919128673848e-05,
+      "loss": 3.8587,
+      "step": 2293248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2444533181163328e-05,
+      "loss": 3.8579,
+      "step": 2293760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2436147233652808e-05,
+      "loss": 3.8706,
+      "step": 2294272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2427761286142288e-05,
+      "loss": 3.8579,
+      "step": 2294784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2419375338631768e-05,
+      "loss": 3.8605,
+      "step": 2295296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2411005769924979e-05,
+      "loss": 3.8702,
+      "step": 2295808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2402619822414459e-05,
+      "loss": 3.8487,
+      "step": 2296320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2394233874903939e-05,
+      "loss": 3.8557,
+      "step": 2296832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2385847927393419e-05,
+      "loss": 3.8503,
+      "step": 2297344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2377461979882899e-05,
+      "loss": 3.8595,
+      "step": 2297856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2369076032372379e-05,
+      "loss": 3.8476,
+      "step": 2298368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2360690084861859e-05,
+      "loss": 3.8511,
+      "step": 2298880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2352304137351339e-05,
+      "loss": 3.856,
+      "step": 2299392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2343918189840819e-05,
+      "loss": 3.8669,
+      "step": 2299904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2335532242330299e-05,
+      "loss": 3.8618,
+      "step": 2300416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.232716267362351e-05,
+      "loss": 3.8525,
+      "step": 2300928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.231877672611299e-05,
+      "loss": 3.8671,
+      "step": 2301440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.231039077860247e-05,
+      "loss": 3.8687,
+      "step": 2301952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.230200483109195e-05,
+      "loss": 3.8518,
+      "step": 2302464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.229363526238516e-05,
+      "loss": 3.8477,
+      "step": 2302976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2285265693678372e-05,
+      "loss": 3.8644,
+      "step": 2303488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2276879746167852e-05,
+      "loss": 3.8485,
+      "step": 2304000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2268493798657332e-05,
+      "loss": 3.8541,
+      "step": 2304512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.226010785114681e-05,
+      "loss": 3.8545,
+      "step": 2305024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2251721903636292e-05,
+      "loss": 3.856,
+      "step": 2305536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2243335956125772e-05,
+      "loss": 3.8632,
+      "step": 2306048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.223495000861525e-05,
+      "loss": 3.8729,
+      "step": 2306560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.222656406110473e-05,
+      "loss": 3.8589,
+      "step": 2307072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2218178113594212e-05,
+      "loss": 3.858,
+      "step": 2307584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2209792166083692e-05,
+      "loss": 3.8678,
+      "step": 2308096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2201422597376901e-05,
+      "loss": 3.8419,
+      "step": 2308608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2193036649866383e-05,
+      "loss": 3.8626,
+      "step": 2309120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2184650702355863e-05,
+      "loss": 3.8479,
+      "step": 2309632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2176264754845341e-05,
+      "loss": 3.8505,
+      "step": 2310144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2167878807334821e-05,
+      "loss": 3.8608,
+      "step": 2310656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2159509238628033e-05,
+      "loss": 3.8663,
+      "step": 2311168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2151123291117513e-05,
+      "loss": 3.8531,
+      "step": 2311680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2142737343606993e-05,
+      "loss": 3.857,
+      "step": 2312192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2134351396096473e-05,
+      "loss": 3.8595,
+      "step": 2312704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2125965448585953e-05,
+      "loss": 3.8489,
+      "step": 2313216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2117579501075432e-05,
+      "loss": 3.8646,
+      "step": 2313728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2109209932368644e-05,
+      "loss": 3.8398,
+      "step": 2314240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2100823984858124e-05,
+      "loss": 3.8576,
+      "step": 2314752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2092438037347604e-05,
+      "loss": 3.8612,
+      "step": 2315264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2084052089837084e-05,
+      "loss": 3.8682,
+      "step": 2315776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2075666142326564e-05,
+      "loss": 3.8488,
+      "step": 2316288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2067280194816044e-05,
+      "loss": 3.8501,
+      "step": 2316800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2058910626109255e-05,
+      "loss": 3.8477,
+      "step": 2317312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2050524678598735e-05,
+      "loss": 3.8458,
+      "step": 2317824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2042138731088215e-05,
+      "loss": 3.872,
+      "step": 2318336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2033769162381427e-05,
+      "loss": 3.8379,
+      "step": 2318848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2025383214870905e-05,
+      "loss": 3.8638,
+      "step": 2319360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2016997267360385e-05,
+      "loss": 3.8601,
+      "step": 2319872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2008611319849867e-05,
+      "loss": 3.8459,
+      "step": 2320384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2000225372339345e-05,
+      "loss": 3.8646,
+      "step": 2320896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1991839424828825e-05,
+      "loss": 3.8422,
+      "step": 2321408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1983453477318305e-05,
+      "loss": 3.8468,
+      "step": 2321920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1975067529807787e-05,
+      "loss": 3.8578,
+      "step": 2322432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1966681582297265e-05,
+      "loss": 3.8634,
+      "step": 2322944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1958295634786745e-05,
+      "loss": 3.8455,
+      "step": 2323456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1949909687276225e-05,
+      "loss": 3.8661,
+      "step": 2323968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1941540118569436e-05,
+      "loss": 3.8374,
+      "step": 2324480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1933154171058916e-05,
+      "loss": 3.8529,
+      "step": 2324992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1924768223548397e-05,
+      "loss": 3.8695,
+      "step": 2325504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1916382276037877e-05,
+      "loss": 3.8632,
+      "step": 2326016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1907996328527355e-05,
+      "loss": 3.8574,
+      "step": 2326528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1899610381016835e-05,
+      "loss": 3.8627,
+      "step": 2327040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1891224433506317e-05,
+      "loss": 3.8702,
+      "step": 2327552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1882838485995795e-05,
+      "loss": 3.855,
+      "step": 2328064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1874468917289006e-05,
+      "loss": 3.8674,
+      "step": 2328576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1866082969778488e-05,
+      "loss": 3.8478,
+      "step": 2329088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1857697022267968e-05,
+      "loss": 3.8462,
+      "step": 2329600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1849311074757446e-05,
+      "loss": 3.8538,
+      "step": 2330112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1840941506050658e-05,
+      "loss": 3.8565,
+      "step": 2330624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1832555558540138e-05,
+      "loss": 3.8571,
+      "step": 2331136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1824169611029617e-05,
+      "loss": 3.8783,
+      "step": 2331648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1815783663519098e-05,
+      "loss": 3.8589,
+      "step": 2332160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1807397716008578e-05,
+      "loss": 3.8465,
+      "step": 2332672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1799011768498058e-05,
+      "loss": 3.8573,
+      "step": 2333184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1790625820987537e-05,
+      "loss": 3.8611,
+      "step": 2333696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1782239873477018e-05,
+      "loss": 3.8529,
+      "step": 2334208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1773870304770229e-05,
+      "loss": 3.853,
+      "step": 2334720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1765484357259709e-05,
+      "loss": 3.858,
+      "step": 2335232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1757098409749189e-05,
+      "loss": 3.8505,
+      "step": 2335744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.17487288410424e-05,
+      "loss": 3.8596,
+      "step": 2336256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.174034289353188e-05,
+      "loss": 3.8459,
+      "step": 2336768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.173197332482509e-05,
+      "loss": 3.8619,
+      "step": 2337280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.172358737731457e-05,
+      "loss": 3.8654,
+      "step": 2337792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1715201429804052e-05,
+      "loss": 3.8549,
+      "step": 2338304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1706815482293532e-05,
+      "loss": 3.8493,
+      "step": 2338816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.169842953478301e-05,
+      "loss": 3.8548,
+      "step": 2339328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.169004358727249e-05,
+      "loss": 3.8527,
+      "step": 2339840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1681657639761972e-05,
+      "loss": 3.8661,
+      "step": 2340352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.167327169225145e-05,
+      "loss": 3.8585,
+      "step": 2340864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.166488574474093e-05,
+      "loss": 3.8483,
+      "step": 2341376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.165649979723041e-05,
+      "loss": 3.8581,
+      "step": 2341888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.164811384971989e-05,
+      "loss": 3.8544,
+      "step": 2342400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.163972790220937e-05,
+      "loss": 3.8564,
+      "step": 2342912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1631358333502582e-05,
+      "loss": 3.854,
+      "step": 2343424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1622972385992062e-05,
+      "loss": 3.8475,
+      "step": 2343936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.161458643848154e-05,
+      "loss": 3.8565,
+      "step": 2344448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1606216869774753e-05,
+      "loss": 3.8552,
+      "step": 2344960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1597830922264233e-05,
+      "loss": 3.846,
+      "step": 2345472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1589444974753713e-05,
+      "loss": 3.8432,
+      "step": 2345984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1581059027243191e-05,
+      "loss": 3.8509,
+      "step": 2346496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1572673079732673e-05,
+      "loss": 3.866,
+      "step": 2347008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1564287132222153e-05,
+      "loss": 3.8477,
+      "step": 2347520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1555901184711631e-05,
+      "loss": 3.8569,
+      "step": 2348032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1547531616004844e-05,
+      "loss": 3.8558,
+      "step": 2348544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1539145668494324e-05,
+      "loss": 3.875,
+      "step": 2349056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1530759720983804e-05,
+      "loss": 3.8568,
+      "step": 2349568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1522373773473284e-05,
+      "loss": 3.8458,
+      "step": 2350080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1513987825962763e-05,
+      "loss": 3.8569,
+      "step": 2350592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1505601878452243e-05,
+      "loss": 3.8612,
+      "step": 2351104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1497232309745454e-05,
+      "loss": 3.8493,
+      "step": 2351616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1488846362234934e-05,
+      "loss": 3.8521,
+      "step": 2352128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1480460414724414e-05,
+      "loss": 3.8554,
+      "step": 2352640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1472074467213894e-05,
+      "loss": 3.8591,
+      "step": 2353152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1463688519703374e-05,
+      "loss": 3.8507,
+      "step": 2353664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1455302572192854e-05,
+      "loss": 3.8595,
+      "step": 2354176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1446933003486065e-05,
+      "loss": 3.8508,
+      "step": 2354688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1438563434779276e-05,
+      "loss": 3.8548,
+      "step": 2355200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1430177487268756e-05,
+      "loss": 3.8634,
+      "step": 2355712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1421791539758237e-05,
+      "loss": 3.8584,
+      "step": 2356224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1413405592247717e-05,
+      "loss": 3.8588,
+      "step": 2356736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1405019644737195e-05,
+      "loss": 3.85,
+      "step": 2357248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1396633697226675e-05,
+      "loss": 3.8623,
+      "step": 2357760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1388247749716157e-05,
+      "loss": 3.8616,
+      "step": 2358272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1379861802205635e-05,
+      "loss": 3.8573,
+      "step": 2358784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1371475854695115e-05,
+      "loss": 3.8616,
+      "step": 2359296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1363089907184595e-05,
+      "loss": 3.8535,
+      "step": 2359808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1354720338477808e-05,
+      "loss": 3.8681,
+      "step": 2360320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1346334390967286e-05,
+      "loss": 3.8647,
+      "step": 2360832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1337948443456768e-05,
+      "loss": 3.8617,
+      "step": 2361344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1329562495946248e-05,
+      "loss": 3.8554,
+      "step": 2361856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1321176548435726e-05,
+      "loss": 3.8634,
+      "step": 2362368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1312790600925206e-05,
+      "loss": 3.8533,
+      "step": 2362880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1304421032218418e-05,
+      "loss": 3.8519,
+      "step": 2363392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1296035084707898e-05,
+      "loss": 3.8677,
+      "step": 2363904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1287649137197376e-05,
+      "loss": 3.8534,
+      "step": 2364416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1279263189686858e-05,
+      "loss": 3.8624,
+      "step": 2364928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1270893620980069e-05,
+      "loss": 3.8605,
+      "step": 2365440
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9779810905456543,
+      "eval_runtime": 290.6758,
+      "eval_samples_per_second": 1312.772,
+      "eval_steps_per_second": 41.025,
+      "step": 2365920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1262507673469549e-05,
+      "loss": 3.8383,
+      "step": 2365952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1254121725959029e-05,
+      "loss": 3.8511,
+      "step": 2366464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1245735778448509e-05,
+      "loss": 3.8533,
+      "step": 2366976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1237349830937989e-05,
+      "loss": 3.8601,
+      "step": 2367488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1228963883427469e-05,
+      "loss": 3.8623,
+      "step": 2368000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1220577935916949e-05,
+      "loss": 3.864,
+      "step": 2368512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1212191988406429e-05,
+      "loss": 3.8545,
+      "step": 2369024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1203806040895907e-05,
+      "loss": 3.8529,
+      "step": 2369536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1195420093385388e-05,
+      "loss": 3.8564,
+      "step": 2370048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1187034145874868e-05,
+      "loss": 3.8679,
+      "step": 2370560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1178648198364347e-05,
+      "loss": 3.8488,
+      "step": 2371072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1170262250853827e-05,
+      "loss": 3.8597,
+      "step": 2371584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1161892682147039e-05,
+      "loss": 3.8643,
+      "step": 2372096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1153506734636519e-05,
+      "loss": 3.85,
+      "step": 2372608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1145120787125999e-05,
+      "loss": 3.8499,
+      "step": 2373120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1136734839615479e-05,
+      "loss": 3.8506,
+      "step": 2373632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1128348892104959e-05,
+      "loss": 3.848,
+      "step": 2374144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1119962944594437e-05,
+      "loss": 3.8494,
+      "step": 2374656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1111576997083919e-05,
+      "loss": 3.8479,
+      "step": 2375168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1103191049573399e-05,
+      "loss": 3.8489,
+      "step": 2375680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1094805102062879e-05,
+      "loss": 3.8647,
+      "step": 2376192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.108643553335609e-05,
+      "loss": 3.8563,
+      "step": 2376704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.107804958584557e-05,
+      "loss": 3.8525,
+      "step": 2377216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.106966363833505e-05,
+      "loss": 3.8603,
+      "step": 2377728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1061277690824528e-05,
+      "loss": 3.8689,
+      "step": 2378240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.105289174331401e-05,
+      "loss": 3.8446,
+      "step": 2378752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.104452217460722e-05,
+      "loss": 3.8449,
+      "step": 2379264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.10361362270967e-05,
+      "loss": 3.8585,
+      "step": 2379776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.102775027958618e-05,
+      "loss": 3.8488,
+      "step": 2380288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.101936433207566e-05,
+      "loss": 3.8503,
+      "step": 2380800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.101097838456514e-05,
+      "loss": 3.848,
+      "step": 2381312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.100259243705462e-05,
+      "loss": 3.8545,
+      "step": 2381824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.099422286834783e-05,
+      "loss": 3.8597,
+      "step": 2382336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.098583692083731e-05,
+      "loss": 3.8653,
+      "step": 2382848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0977450973326792e-05,
+      "loss": 3.8587,
+      "step": 2383360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.096906502581627e-05,
+      "loss": 3.8523,
+      "step": 2383872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.096067907830575e-05,
+      "loss": 3.8627,
+      "step": 2384384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.095229313079523e-05,
+      "loss": 3.8398,
+      "step": 2384896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.094390718328471e-05,
+      "loss": 3.8623,
+      "step": 2385408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.093552123577419e-05,
+      "loss": 3.8436,
+      "step": 2385920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0927151667067401e-05,
+      "loss": 3.8479,
+      "step": 2386432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0918765719556883e-05,
+      "loss": 3.855,
+      "step": 2386944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0910379772046361e-05,
+      "loss": 3.8609,
+      "step": 2387456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0902010203339574e-05,
+      "loss": 3.8524,
+      "step": 2387968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0893624255829054e-05,
+      "loss": 3.8543,
+      "step": 2388480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0885238308318532e-05,
+      "loss": 3.8544,
+      "step": 2388992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0876852360808012e-05,
+      "loss": 3.849,
+      "step": 2389504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0868466413297493e-05,
+      "loss": 3.8627,
+      "step": 2390016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0860096844590704e-05,
+      "loss": 3.8315,
+      "step": 2390528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0851727275883915e-05,
+      "loss": 3.8496,
+      "step": 2391040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0843341328373395e-05,
+      "loss": 3.865,
+      "step": 2391552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0834955380862875e-05,
+      "loss": 3.8588,
+      "step": 2392064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0826569433352355e-05,
+      "loss": 3.8432,
+      "step": 2392576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0818183485841835e-05,
+      "loss": 3.8496,
+      "step": 2393088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0809797538331315e-05,
+      "loss": 3.8455,
+      "step": 2393600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0801411590820795e-05,
+      "loss": 3.8405,
+      "step": 2394112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0793025643310275e-05,
+      "loss": 3.8714,
+      "step": 2394624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0784639695799755e-05,
+      "loss": 3.8321,
+      "step": 2395136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0776270127092965e-05,
+      "loss": 3.8598,
+      "step": 2395648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0767884179582447e-05,
+      "loss": 3.8502,
+      "step": 2396160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0759514610875656e-05,
+      "loss": 3.8448,
+      "step": 2396672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0751128663365136e-05,
+      "loss": 3.8569,
+      "step": 2397184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0742742715854618e-05,
+      "loss": 3.8406,
+      "step": 2397696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0734356768344096e-05,
+      "loss": 3.8445,
+      "step": 2398208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0725970820833576e-05,
+      "loss": 3.8489,
+      "step": 2398720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0717584873323058e-05,
+      "loss": 3.8657,
+      "step": 2399232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0709198925812538e-05,
+      "loss": 3.8405,
+      "step": 2399744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0700812978302016e-05,
+      "loss": 3.86,
+      "step": 2400256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0692427030791496e-05,
+      "loss": 3.8358,
+      "step": 2400768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0684041083280978e-05,
+      "loss": 3.8465,
+      "step": 2401280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0675655135770456e-05,
+      "loss": 3.8652,
+      "step": 2401792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0667269188259936e-05,
+      "loss": 3.8632,
+      "step": 2402304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0658899619553148e-05,
+      "loss": 3.8521,
+      "step": 2402816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0650513672042628e-05,
+      "loss": 3.8584,
+      "step": 2403328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0642127724532106e-05,
+      "loss": 3.8669,
+      "step": 2403840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0633741777021586e-05,
+      "loss": 3.8491,
+      "step": 2404352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0625372208314799e-05,
+      "loss": 3.8634,
+      "step": 2404864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0616986260804277e-05,
+      "loss": 3.8483,
+      "step": 2405376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.060861669209749e-05,
+      "loss": 3.8433,
+      "step": 2405888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.060023074458697e-05,
+      "loss": 3.847,
+      "step": 2406400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.059184479707645e-05,
+      "loss": 3.8515,
+      "step": 2406912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.058345884956593e-05,
+      "loss": 3.8575,
+      "step": 2407424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.057507290205541e-05,
+      "loss": 3.8705,
+      "step": 2407936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.056668695454489e-05,
+      "loss": 3.8628,
+      "step": 2408448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0558317385838102e-05,
+      "loss": 3.8438,
+      "step": 2408960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.054993143832758e-05,
+      "loss": 3.8506,
+      "step": 2409472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.054154549081706e-05,
+      "loss": 3.8599,
+      "step": 2409984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.053315954330654e-05,
+      "loss": 3.852,
+      "step": 2410496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0524789974599751e-05,
+      "loss": 3.8505,
+      "step": 2411008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.051640402708923e-05,
+      "loss": 3.8535,
+      "step": 2411520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0508018079578712e-05,
+      "loss": 3.8443,
+      "step": 2412032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0499632132068192e-05,
+      "loss": 3.8555,
+      "step": 2412544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.049124618455767e-05,
+      "loss": 3.8457,
+      "step": 2413056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.048286023704715e-05,
+      "loss": 3.8582,
+      "step": 2413568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0474474289536632e-05,
+      "loss": 3.8585,
+      "step": 2414080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.046608834202611e-05,
+      "loss": 3.8548,
+      "step": 2414592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0457718773319321e-05,
+      "loss": 3.8481,
+      "step": 2415104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0449332825808803e-05,
+      "loss": 3.8502,
+      "step": 2415616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0440946878298283e-05,
+      "loss": 3.8498,
+      "step": 2416128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0432560930787761e-05,
+      "loss": 3.8608,
+      "step": 2416640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0424191362080974e-05,
+      "loss": 3.8544,
+      "step": 2417152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0415805414570454e-05,
+      "loss": 3.8434,
+      "step": 2417664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0407419467059932e-05,
+      "loss": 3.8539,
+      "step": 2418176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0399033519549414e-05,
+      "loss": 3.8512,
+      "step": 2418688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0390647572038894e-05,
+      "loss": 3.8543,
+      "step": 2419200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0382278003332104e-05,
+      "loss": 3.8492,
+      "step": 2419712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0373892055821584e-05,
+      "loss": 3.8428,
+      "step": 2420224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0365506108311064e-05,
+      "loss": 3.8567,
+      "step": 2420736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0357120160800544e-05,
+      "loss": 3.853,
+      "step": 2421248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0348750592093757e-05,
+      "loss": 3.8383,
+      "step": 2421760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0340381023386966e-05,
+      "loss": 3.8411,
+      "step": 2422272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0331995075876447e-05,
+      "loss": 3.8441,
+      "step": 2422784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0323609128365927e-05,
+      "loss": 3.8607,
+      "step": 2423296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0315223180855406e-05,
+      "loss": 3.8481,
+      "step": 2423808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0306837233344886e-05,
+      "loss": 3.8521,
+      "step": 2424320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0298451285834367e-05,
+      "loss": 3.8556,
+      "step": 2424832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0290081717127578e-05,
+      "loss": 3.8722,
+      "step": 2425344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0281695769617056e-05,
+      "loss": 3.8524,
+      "step": 2425856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0273309822106538e-05,
+      "loss": 3.841,
+      "step": 2426368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0264923874596018e-05,
+      "loss": 3.8545,
+      "step": 2426880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0256537927085496e-05,
+      "loss": 3.8567,
+      "step": 2427392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0248151979574978e-05,
+      "loss": 3.8474,
+      "step": 2427904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0239782410868189e-05,
+      "loss": 3.8485,
+      "step": 2428416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0231396463357669e-05,
+      "loss": 3.8525,
+      "step": 2428928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0223010515847148e-05,
+      "loss": 3.854,
+      "step": 2429440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0214624568336628e-05,
+      "loss": 3.8491,
+      "step": 2429952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0206238620826108e-05,
+      "loss": 3.8568,
+      "step": 2430464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0197852673315587e-05,
+      "loss": 3.847,
+      "step": 2430976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0189483104608799e-05,
+      "loss": 3.853,
+      "step": 2431488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0181097157098279e-05,
+      "loss": 3.8518,
+      "step": 2432000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0172711209587759e-05,
+      "loss": 3.8617,
+      "step": 2432512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0164325262077239e-05,
+      "loss": 3.8561,
+      "step": 2433024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0155939314566719e-05,
+      "loss": 3.8435,
+      "step": 2433536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.014756974585993e-05,
+      "loss": 3.8579,
+      "step": 2434048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.013918379834941e-05,
+      "loss": 3.8592,
+      "step": 2434560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.013079785083889e-05,
+      "loss": 3.8524,
+      "step": 2435072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.012241190332837e-05,
+      "loss": 3.8607,
+      "step": 2435584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0114025955817851e-05,
+      "loss": 3.8488,
+      "step": 2436096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.010564000830733e-05,
+      "loss": 3.8651,
+      "step": 2436608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.009725406079681e-05,
+      "loss": 3.8573,
+      "step": 2437120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.008886811328629e-05,
+      "loss": 3.8589,
+      "step": 2437632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.00804985445795e-05,
+      "loss": 3.85,
+      "step": 2438144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0072128975872713e-05,
+      "loss": 3.8622,
+      "step": 2438656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0063743028362193e-05,
+      "loss": 3.8509,
+      "step": 2439168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0055357080851673e-05,
+      "loss": 3.846,
+      "step": 2439680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0046971133341151e-05,
+      "loss": 3.8674,
+      "step": 2440192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0038585185830633e-05,
+      "loss": 3.8536,
+      "step": 2440704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0030215617123843e-05,
+      "loss": 3.8565,
+      "step": 2441216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0021829669613323e-05,
+      "loss": 3.8582,
+      "step": 2441728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0013443722102803e-05,
+      "loss": 3.8465,
+      "step": 2442240
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.977142572402954,
+      "eval_runtime": 287.122,
+      "eval_samples_per_second": 1329.021,
+      "eval_steps_per_second": 41.533,
+      "step": 2442240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0005057774592283e-05,
+      "loss": 3.8483,
+      "step": 2442752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.996671827081763e-06,
+      "loss": 3.8494,
+      "step": 2443264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.988285879571241e-06,
+      "loss": 3.8567,
+      "step": 2443776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.979916310864454e-06,
+      "loss": 3.8571,
+      "step": 2444288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.971530363353934e-06,
+      "loss": 3.8618,
+      "step": 2444800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.963144415843414e-06,
+      "loss": 3.8516,
+      "step": 2445312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.954758468332894e-06,
+      "loss": 3.8529,
+      "step": 2445824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.946372520822374e-06,
+      "loss": 3.8479,
+      "step": 2446336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.937986573311854e-06,
+      "loss": 3.8661,
+      "step": 2446848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.929600625801334e-06,
+      "loss": 3.8454,
+      "step": 2447360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.921214678290814e-06,
+      "loss": 3.8562,
+      "step": 2447872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.912877867191488e-06,
+      "loss": 3.8606,
+      "step": 2448384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.904491919680968e-06,
+      "loss": 3.851,
+      "step": 2448896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.896105972170448e-06,
+      "loss": 3.846,
+      "step": 2449408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.887720024659928e-06,
+      "loss": 3.845,
+      "step": 2449920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.879334077149408e-06,
+      "loss": 3.8427,
+      "step": 2450432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.870948129638888e-06,
+      "loss": 3.847,
+      "step": 2450944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.862562182128367e-06,
+      "loss": 3.8461,
+      "step": 2451456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.854176234617847e-06,
+      "loss": 3.8418,
+      "step": 2451968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.845790287107327e-06,
+      "loss": 3.8655,
+      "step": 2452480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.837404339596806e-06,
+      "loss": 3.8516,
+      "step": 2452992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.829018392086287e-06,
+      "loss": 3.8483,
+      "step": 2453504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.820632444575767e-06,
+      "loss": 3.8567,
+      "step": 2454016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.812246497065246e-06,
+      "loss": 3.8651,
+      "step": 2454528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.803876928358458e-06,
+      "loss": 3.8406,
+      "step": 2455040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.795507359651669e-06,
+      "loss": 3.8461,
+      "step": 2455552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.787121412141149e-06,
+      "loss": 3.849,
+      "step": 2456064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.778735464630629e-06,
+      "loss": 3.852,
+      "step": 2456576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.770349517120109e-06,
+      "loss": 3.8433,
+      "step": 2457088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.761979948413321e-06,
+      "loss": 3.8422,
+      "step": 2457600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.753594000902801e-06,
+      "loss": 3.8533,
+      "step": 2458112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.74520805339228e-06,
+      "loss": 3.8556,
+      "step": 2458624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.73682210588176e-06,
+      "loss": 3.8603,
+      "step": 2459136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.728436158371241e-06,
+      "loss": 3.8556,
+      "step": 2459648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.720066589664452e-06,
+      "loss": 3.8505,
+      "step": 2460160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.71168064215393e-06,
+      "loss": 3.8552,
+      "step": 2460672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.703294694643412e-06,
+      "loss": 3.8372,
+      "step": 2461184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.694908747132892e-06,
+      "loss": 3.857,
+      "step": 2461696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.68652279962237e-06,
+      "loss": 3.8429,
+      "step": 2462208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.67813685211185e-06,
+      "loss": 3.8448,
+      "step": 2462720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.669750904601331e-06,
+      "loss": 3.8484,
+      "step": 2463232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.66136495709081e-06,
+      "loss": 3.861,
+      "step": 2463744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.652995388384022e-06,
+      "loss": 3.8485,
+      "step": 2464256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.644609440873502e-06,
+      "loss": 3.8479,
+      "step": 2464768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.636223493362982e-06,
+      "loss": 3.8563,
+      "step": 2465280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.62783754585246e-06,
+      "loss": 3.8414,
+      "step": 2465792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.619467977145673e-06,
+      "loss": 3.8586,
+      "step": 2466304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.611082029635153e-06,
+      "loss": 3.8316,
+      "step": 2466816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.602696082124633e-06,
+      "loss": 3.8452,
+      "step": 2467328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.594310134614113e-06,
+      "loss": 3.8598,
+      "step": 2467840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.585940565907324e-06,
+      "loss": 3.8588,
+      "step": 2468352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.577554618396804e-06,
+      "loss": 3.8432,
+      "step": 2468864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.569168670886283e-06,
+      "loss": 3.8437,
+      "step": 2469376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.560782723375763e-06,
+      "loss": 3.8467,
+      "step": 2469888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.552413154668976e-06,
+      "loss": 3.8308,
+      "step": 2470400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.544027207158456e-06,
+      "loss": 3.8653,
+      "step": 2470912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.535641259647934e-06,
+      "loss": 3.838,
+      "step": 2471424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.527255312137414e-06,
+      "loss": 3.8551,
+      "step": 2471936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.518869364626896e-06,
+      "loss": 3.8484,
+      "step": 2472448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.510483417116374e-06,
+      "loss": 3.8409,
+      "step": 2472960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.502113848409586e-06,
+      "loss": 3.8498,
+      "step": 2473472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.493727900899066e-06,
+      "loss": 3.8412,
+      "step": 2473984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.485341953388546e-06,
+      "loss": 3.8432,
+      "step": 2474496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.476972384681757e-06,
+      "loss": 3.8427,
+      "step": 2475008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.468586437171237e-06,
+      "loss": 3.8622,
+      "step": 2475520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.460200489660717e-06,
+      "loss": 3.8433,
+      "step": 2476032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.451814542150195e-06,
+      "loss": 3.8487,
+      "step": 2476544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.443428594639677e-06,
+      "loss": 3.8333,
+      "step": 2477056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.435059025932888e-06,
+      "loss": 3.8475,
+      "step": 2477568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.426673078422368e-06,
+      "loss": 3.8605,
+      "step": 2478080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.418287130911848e-06,
+      "loss": 3.8572,
+      "step": 2478592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.409901183401328e-06,
+      "loss": 3.8491,
+      "step": 2479104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.401515235890808e-06,
+      "loss": 3.8608,
+      "step": 2479616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.393129288380288e-06,
+      "loss": 3.8626,
+      "step": 2480128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.384759719673498e-06,
+      "loss": 3.8464,
+      "step": 2480640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.376373772162978e-06,
+      "loss": 3.8571,
+      "step": 2481152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.367987824652458e-06,
+      "loss": 3.8473,
+      "step": 2481664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.359601877141938e-06,
+      "loss": 3.8413,
+      "step": 2482176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.351215929631418e-06,
+      "loss": 3.8449,
+      "step": 2482688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.342829982120898e-06,
+      "loss": 3.8455,
+      "step": 2483200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.33446041341411e-06,
+      "loss": 3.8543,
+      "step": 2483712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.326074465903589e-06,
+      "loss": 3.8697,
+      "step": 2484224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.317704897196801e-06,
+      "loss": 3.8591,
+      "step": 2484736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.309318949686281e-06,
+      "loss": 3.8375,
+      "step": 2485248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.30093300217576e-06,
+      "loss": 3.8509,
+      "step": 2485760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.292547054665241e-06,
+      "loss": 3.8541,
+      "step": 2486272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.284161107154721e-06,
+      "loss": 3.8491,
+      "step": 2486784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.275775159644201e-06,
+      "loss": 3.8472,
+      "step": 2487296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.26738921213368e-06,
+      "loss": 3.8496,
+      "step": 2487808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.259003264623161e-06,
+      "loss": 3.8439,
+      "step": 2488320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.250617317112641e-06,
+      "loss": 3.8503,
+      "step": 2488832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.24223136960212e-06,
+      "loss": 3.8422,
+      "step": 2489344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.2338454220916e-06,
+      "loss": 3.8584,
+      "step": 2489856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.225475853384812e-06,
+      "loss": 3.8507,
+      "step": 2490368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.217089905874292e-06,
+      "loss": 3.8564,
+      "step": 2490880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.208720337167502e-06,
+      "loss": 3.8386,
+      "step": 2491392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.200334389656982e-06,
+      "loss": 3.846,
+      "step": 2491904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.191948442146462e-06,
+      "loss": 3.8502,
+      "step": 2492416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.183562494635942e-06,
+      "loss": 3.8561,
+      "step": 2492928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.175192925929153e-06,
+      "loss": 3.8571,
+      "step": 2493440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.166806978418633e-06,
+      "loss": 3.8388,
+      "step": 2493952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.158421030908115e-06,
+      "loss": 3.8484,
+      "step": 2494464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.150035083397593e-06,
+      "loss": 3.8456,
+      "step": 2494976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.141649135887073e-06,
+      "loss": 3.8538,
+      "step": 2495488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.133263188376553e-06,
+      "loss": 3.8441,
+      "step": 2496000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.124877240866033e-06,
+      "loss": 3.842,
+      "step": 2496512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.116491293355513e-06,
+      "loss": 3.8536,
+      "step": 2497024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.108105345844993e-06,
+      "loss": 3.8442,
+      "step": 2497536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.099735777138205e-06,
+      "loss": 3.8399,
+      "step": 2498048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.091349829627684e-06,
+      "loss": 3.8379,
+      "step": 2498560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.082980260920896e-06,
+      "loss": 3.8405,
+      "step": 2499072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.074594313410376e-06,
+      "loss": 3.856,
+      "step": 2499584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.066208365899854e-06,
+      "loss": 3.8447,
+      "step": 2500096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.057822418389334e-06,
+      "loss": 3.8496,
+      "step": 2500608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.049436470878816e-06,
+      "loss": 3.8532,
+      "step": 2501120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.041050523368296e-06,
+      "loss": 3.8653,
+      "step": 2501632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.032664575857774e-06,
+      "loss": 3.8485,
+      "step": 2502144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.024278628347254e-06,
+      "loss": 3.8412,
+      "step": 2502656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.015909059640466e-06,
+      "loss": 3.85,
+      "step": 2503168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.007539490933677e-06,
+      "loss": 3.854,
+      "step": 2503680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.999153543423157e-06,
+      "loss": 3.8428,
+      "step": 2504192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.990767595912637e-06,
+      "loss": 3.8462,
+      "step": 2504704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.982381648402117e-06,
+      "loss": 3.8456,
+      "step": 2505216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.973995700891597e-06,
+      "loss": 3.8511,
+      "step": 2505728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.965609753381077e-06,
+      "loss": 3.8473,
+      "step": 2506240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.957223805870557e-06,
+      "loss": 3.8501,
+      "step": 2506752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.948837858360035e-06,
+      "loss": 3.8451,
+      "step": 2507264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.940484668456979e-06,
+      "loss": 3.8509,
+      "step": 2507776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.932098720946459e-06,
+      "loss": 3.8435,
+      "step": 2508288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.92371277343594e-06,
+      "loss": 3.8631,
+      "step": 2508800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.915326825925418e-06,
+      "loss": 3.8511,
+      "step": 2509312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.906940878414898e-06,
+      "loss": 3.8463,
+      "step": 2509824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.898554930904378e-06,
+      "loss": 3.8463,
+      "step": 2510336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.89016898339386e-06,
+      "loss": 3.856,
+      "step": 2510848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.881783035883338e-06,
+      "loss": 3.849,
+      "step": 2511360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.873397088372818e-06,
+      "loss": 3.8572,
+      "step": 2511872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.86502751966603e-06,
+      "loss": 3.8468,
+      "step": 2512384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.856657950959241e-06,
+      "loss": 3.8618,
+      "step": 2512896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.848272003448721e-06,
+      "loss": 3.853,
+      "step": 2513408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.839886055938201e-06,
+      "loss": 3.8539,
+      "step": 2513920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.831500108427681e-06,
+      "loss": 3.8493,
+      "step": 2514432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.823114160917161e-06,
+      "loss": 3.8585,
+      "step": 2514944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.814728213406641e-06,
+      "loss": 3.8512,
+      "step": 2515456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.806342265896121e-06,
+      "loss": 3.8398,
+      "step": 2515968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.7979563183856e-06,
+      "loss": 3.8614,
+      "step": 2516480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.789570370875081e-06,
+      "loss": 3.8532,
+      "step": 2516992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.781184423364561e-06,
+      "loss": 3.8521,
+      "step": 2517504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.772814854657772e-06,
+      "loss": 3.8565,
+      "step": 2518016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.764428907147252e-06,
+      "loss": 3.8422,
+      "step": 2518528
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.97613525390625,
+      "eval_runtime": 302.1061,
+      "eval_samples_per_second": 1263.103,
+      "eval_steps_per_second": 39.473,
+      "step": 2518560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.756042959636732e-06,
+      "loss": 3.845,
+      "step": 2519040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.747657012126212e-06,
+      "loss": 3.8503,
+      "step": 2519552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.73927106461569e-06,
+      "loss": 3.8543,
+      "step": 2520064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.730885117105172e-06,
+      "loss": 3.8516,
+      "step": 2520576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.722499169594652e-06,
+      "loss": 3.8618,
+      "step": 2521088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.714113222084132e-06,
+      "loss": 3.8472,
+      "step": 2521600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.705727274573612e-06,
+      "loss": 3.8502,
+      "step": 2522112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.697357705866822e-06,
+      "loss": 3.8458,
+      "step": 2522624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.688971758356302e-06,
+      "loss": 3.8619,
+      "step": 2523136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.680585810845782e-06,
+      "loss": 3.8459,
+      "step": 2523648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.672199863335262e-06,
+      "loss": 3.8475,
+      "step": 2524160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.663813915824742e-06,
+      "loss": 3.8549,
+      "step": 2524672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.655427968314222e-06,
+      "loss": 3.8505,
+      "step": 2525184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.647042020803702e-06,
+      "loss": 3.8433,
+      "step": 2525696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.638656073293182e-06,
+      "loss": 3.8481,
+      "step": 2526208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.630270125782662e-06,
+      "loss": 3.8334,
+      "step": 2526720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.62188417827214e-06,
+      "loss": 3.8459,
+      "step": 2527232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.613498230761622e-06,
+      "loss": 3.8413,
+      "step": 2527744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.605112283251102e-06,
+      "loss": 3.8388,
+      "step": 2528256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.59672633574058e-06,
+      "loss": 3.8674,
+      "step": 2528768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.58834038823006e-06,
+      "loss": 3.8451,
+      "step": 2529280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.579970819523273e-06,
+      "loss": 3.8522,
+      "step": 2529792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.571584872012752e-06,
+      "loss": 3.8494,
+      "step": 2530304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.563198924502232e-06,
+      "loss": 3.8632,
+      "step": 2530816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.554812976991712e-06,
+      "loss": 3.838,
+      "step": 2531328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.546427029481192e-06,
+      "loss": 3.8439,
+      "step": 2531840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.53804108197067e-06,
+      "loss": 3.8481,
+      "step": 2532352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.529655134460152e-06,
+      "loss": 3.8497,
+      "step": 2532864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.521269186949632e-06,
+      "loss": 3.8409,
+      "step": 2533376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.51288323943911e-06,
+      "loss": 3.8414,
+      "step": 2533888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.50449729192859e-06,
+      "loss": 3.8504,
+      "step": 2534400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.496111344418072e-06,
+      "loss": 3.8553,
+      "step": 2534912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.487725396907552e-06,
+      "loss": 3.8579,
+      "step": 2535424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.479355828200763e-06,
+      "loss": 3.8519,
+      "step": 2535936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.470969880690243e-06,
+      "loss": 3.8437,
+      "step": 2536448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.462583933179723e-06,
+      "loss": 3.8555,
+      "step": 2536960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.454214364472934e-06,
+      "loss": 3.8341,
+      "step": 2537472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.445828416962413e-06,
+      "loss": 3.8563,
+      "step": 2537984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.437442469451893e-06,
+      "loss": 3.8372,
+      "step": 2538496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.429056521941373e-06,
+      "loss": 3.8434,
+      "step": 2539008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.420686953234584e-06,
+      "loss": 3.8439,
+      "step": 2539520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.412301005724064e-06,
+      "loss": 3.8571,
+      "step": 2540032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.403915058213544e-06,
+      "loss": 3.8484,
+      "step": 2540544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.395529110703026e-06,
+      "loss": 3.8433,
+      "step": 2541056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.387143163192504e-06,
+      "loss": 3.8515,
+      "step": 2541568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.378773594485717e-06,
+      "loss": 3.8432,
+      "step": 2542080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.370387646975196e-06,
+      "loss": 3.8543,
+      "step": 2542592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.362001699464675e-06,
+      "loss": 3.8304,
+      "step": 2543104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.353615751954155e-06,
+      "loss": 3.8366,
+      "step": 2543616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.345229804443636e-06,
+      "loss": 3.8546,
+      "step": 2544128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.336843856933116e-06,
+      "loss": 3.8565,
+      "step": 2544640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.328457909422595e-06,
+      "loss": 3.8396,
+      "step": 2545152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.320088340715807e-06,
+      "loss": 3.8371,
+      "step": 2545664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.311702393205287e-06,
+      "loss": 3.847,
+      "step": 2546176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.30334920330223e-06,
+      "loss": 3.8271,
+      "step": 2546688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.294963255791709e-06,
+      "loss": 3.859,
+      "step": 2547200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.286577308281189e-06,
+      "loss": 3.8391,
+      "step": 2547712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.278191360770668e-06,
+      "loss": 3.8497,
+      "step": 2548224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.269805413260148e-06,
+      "loss": 3.8456,
+      "step": 2548736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.261419465749628e-06,
+      "loss": 3.8377,
+      "step": 2549248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.253033518239108e-06,
+      "loss": 3.8483,
+      "step": 2549760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.24464757072859e-06,
+      "loss": 3.8399,
+      "step": 2550272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.236261623218068e-06,
+      "loss": 3.837,
+      "step": 2550784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.227892054511279e-06,
+      "loss": 3.8394,
+      "step": 2551296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.21950610700076e-06,
+      "loss": 3.8574,
+      "step": 2551808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.211120159490239e-06,
+      "loss": 3.8436,
+      "step": 2552320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.202734211979719e-06,
+      "loss": 3.8425,
+      "step": 2552832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.194348264469199e-06,
+      "loss": 3.8329,
+      "step": 2553344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.18596231695868e-06,
+      "loss": 3.8423,
+      "step": 2553856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.177576369448159e-06,
+      "loss": 3.8568,
+      "step": 2554368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.169190421937639e-06,
+      "loss": 3.8557,
+      "step": 2554880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.160820853230851e-06,
+      "loss": 3.8455,
+      "step": 2555392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.15243490572033e-06,
+      "loss": 3.8567,
+      "step": 2555904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.14404895820981e-06,
+      "loss": 3.8564,
+      "step": 2556416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.135663010699291e-06,
+      "loss": 3.8474,
+      "step": 2556928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.127293441992502e-06,
+      "loss": 3.8559,
+      "step": 2557440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.11890749448198e-06,
+      "loss": 3.8469,
+      "step": 2557952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.110521546971462e-06,
+      "loss": 3.8369,
+      "step": 2558464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.102135599460942e-06,
+      "loss": 3.8429,
+      "step": 2558976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.093766030754153e-06,
+      "loss": 3.8454,
+      "step": 2559488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.085380083243632e-06,
+      "loss": 3.8483,
+      "step": 2560000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076994135733112e-06,
+      "loss": 3.8663,
+      "step": 2560512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.068608188222592e-06,
+      "loss": 3.8556,
+      "step": 2561024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.060222240712072e-06,
+      "loss": 3.834,
+      "step": 2561536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.051836293201552e-06,
+      "loss": 3.8482,
+      "step": 2562048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.043450345691032e-06,
+      "loss": 3.8511,
+      "step": 2562560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.03506439818051e-06,
+      "loss": 3.848,
+      "step": 2563072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.026711208277454e-06,
+      "loss": 3.8436,
+      "step": 2563584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.018325260766934e-06,
+      "loss": 3.844,
+      "step": 2564096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.009939313256415e-06,
+      "loss": 3.8441,
+      "step": 2564608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.001553365745894e-06,
+      "loss": 3.8461,
+      "step": 2565120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.993167418235374e-06,
+      "loss": 3.8392,
+      "step": 2565632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.984781470724854e-06,
+      "loss": 3.8535,
+      "step": 2566144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.976395523214335e-06,
+      "loss": 3.8477,
+      "step": 2566656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.968009575703814e-06,
+      "loss": 3.8537,
+      "step": 2567168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.959640006997026e-06,
+      "loss": 3.838,
+      "step": 2567680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.951254059486506e-06,
+      "loss": 3.8485,
+      "step": 2568192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.942884490779715e-06,
+      "loss": 3.8467,
+      "step": 2568704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.934498543269197e-06,
+      "loss": 3.8506,
+      "step": 2569216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.926112595758677e-06,
+      "loss": 3.8508,
+      "step": 2569728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.917726648248157e-06,
+      "loss": 3.8417,
+      "step": 2570240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.909340700737637e-06,
+      "loss": 3.844,
+      "step": 2570752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.900954753227117e-06,
+      "loss": 3.8449,
+      "step": 2571264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.892568805716597e-06,
+      "loss": 3.8486,
+      "step": 2571776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.884182858206075e-06,
+      "loss": 3.8375,
+      "step": 2572288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.875813289499287e-06,
+      "loss": 3.84,
+      "step": 2572800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.867427341988767e-06,
+      "loss": 3.8502,
+      "step": 2573312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.859041394478247e-06,
+      "loss": 3.8407,
+      "step": 2573824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.850671825771458e-06,
+      "loss": 3.8412,
+      "step": 2574336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.842285878260938e-06,
+      "loss": 3.8374,
+      "step": 2574848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.833899930750418e-06,
+      "loss": 3.8354,
+      "step": 2575360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.825513983239898e-06,
+      "loss": 3.8555,
+      "step": 2575872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.817128035729378e-06,
+      "loss": 3.8383,
+      "step": 2576384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.808742088218858e-06,
+      "loss": 3.8489,
+      "step": 2576896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.800356140708338e-06,
+      "loss": 3.8467,
+      "step": 2577408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.791970193197818e-06,
+      "loss": 3.8626,
+      "step": 2577920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.783600624491028e-06,
+      "loss": 3.8487,
+      "step": 2578432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.775231055784241e-06,
+      "loss": 3.8364,
+      "step": 2578944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.766845108273721e-06,
+      "loss": 3.848,
+      "step": 2579456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.758459160763199e-06,
+      "loss": 3.8509,
+      "step": 2579968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.75007321325268e-06,
+      "loss": 3.8395,
+      "step": 2580480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.741703644545892e-06,
+      "loss": 3.8442,
+      "step": 2580992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.733317697035372e-06,
+      "loss": 3.8417,
+      "step": 2581504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.724931749524851e-06,
+      "loss": 3.853,
+      "step": 2582016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.716545802014331e-06,
+      "loss": 3.8409,
+      "step": 2582528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.708159854503811e-06,
+      "loss": 3.8482,
+      "step": 2583040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.699773906993291e-06,
+      "loss": 3.8421,
+      "step": 2583552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.691404338286502e-06,
+      "loss": 3.8466,
+      "step": 2584064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.683018390775982e-06,
+      "loss": 3.8382,
+      "step": 2584576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.674632443265462e-06,
+      "loss": 3.8616,
+      "step": 2585088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.666246495754942e-06,
+      "loss": 3.8534,
+      "step": 2585600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.657860548244422e-06,
+      "loss": 3.8381,
+      "step": 2586112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.649474600733902e-06,
+      "loss": 3.8445,
+      "step": 2586624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.641088653223382e-06,
+      "loss": 3.8515,
+      "step": 2587136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.632702705712862e-06,
+      "loss": 3.8497,
+      "step": 2587648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.624316758202342e-06,
+      "loss": 3.8569,
+      "step": 2588160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.615930810691821e-06,
+      "loss": 3.8411,
+      "step": 2588672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6075612419850325e-06,
+      "loss": 3.8549,
+      "step": 2589184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5991752944745125e-06,
+      "loss": 3.8485,
+      "step": 2589696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.590789346963993e-06,
+      "loss": 3.8538,
+      "step": 2590208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.582403399453472e-06,
+      "loss": 3.8459,
+      "step": 2590720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.574017451942952e-06,
+      "loss": 3.8504,
+      "step": 2591232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.565647883236164e-06,
+      "loss": 3.8491,
+      "step": 2591744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.557261935725643e-06,
+      "loss": 3.8385,
+      "step": 2592256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.548875988215123e-06,
+      "loss": 3.858,
+      "step": 2592768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.540490040704604e-06,
+      "loss": 3.851,
+      "step": 2593280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.532104093194084e-06,
+      "loss": 3.8497,
+      "step": 2593792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5237509032910254e-06,
+      "loss": 3.8495,
+      "step": 2594304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.515364955780506e-06,
+      "loss": 3.8393,
+      "step": 2594816
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9753286838531494,
+      "eval_runtime": 307.0852,
+      "eval_samples_per_second": 1242.623,
+      "eval_steps_per_second": 38.833,
+      "step": 2594880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.506979008269986e-06,
+      "loss": 3.8436,
+      "step": 2595328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.498593060759466e-06,
+      "loss": 3.8423,
+      "step": 2595840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.490207113248945e-06,
+      "loss": 3.8534,
+      "step": 2596352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.481821165738426e-06,
+      "loss": 3.8453,
+      "step": 2596864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.473435218227906e-06,
+      "loss": 3.8615,
+      "step": 2597376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.465049270717385e-06,
+      "loss": 3.8439,
+      "step": 2597888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.456663323206865e-06,
+      "loss": 3.8449,
+      "step": 2598400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.448277375696346e-06,
+      "loss": 3.8404,
+      "step": 2598912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.439891428185824e-06,
+      "loss": 3.8595,
+      "step": 2599424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.431505480675305e-06,
+      "loss": 3.8449,
+      "step": 2599936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.423119533164785e-06,
+      "loss": 3.8447,
+      "step": 2600448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.4147499644579965e-06,
+      "loss": 3.8457,
+      "step": 2600960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.406380395751207e-06,
+      "loss": 3.8483,
+      "step": 2601472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.397994448240687e-06,
+      "loss": 3.8419,
+      "step": 2601984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.389608500730168e-06,
+      "loss": 3.8445,
+      "step": 2602496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.381222553219648e-06,
+      "loss": 3.8358,
+      "step": 2603008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.372836605709127e-06,
+      "loss": 3.8408,
+      "step": 2603520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.364450658198607e-06,
+      "loss": 3.8364,
+      "step": 2604032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.356064710688088e-06,
+      "loss": 3.834,
+      "step": 2604544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.347695141981298e-06,
+      "loss": 3.8684,
+      "step": 2605056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.339309194470778e-06,
+      "loss": 3.8441,
+      "step": 2605568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.330956004567721e-06,
+      "loss": 3.8422,
+      "step": 2606080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.322570057057201e-06,
+      "loss": 3.8478,
+      "step": 2606592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.31418410954668e-06,
+      "loss": 3.8572,
+      "step": 2607104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.305798162036161e-06,
+      "loss": 3.8355,
+      "step": 2607616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.297428593329373e-06,
+      "loss": 3.8427,
+      "step": 2608128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.289042645818853e-06,
+      "loss": 3.842,
+      "step": 2608640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.280656698308332e-06,
+      "loss": 3.8498,
+      "step": 2609152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.272270750797812e-06,
+      "loss": 3.836,
+      "step": 2609664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.2638848032872925e-06,
+      "loss": 3.8397,
+      "step": 2610176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.255498855776771e-06,
+      "loss": 3.8432,
+      "step": 2610688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.2471129082662515e-06,
+      "loss": 3.8516,
+      "step": 2611200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.2387269607557315e-06,
+      "loss": 3.859,
+      "step": 2611712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.230341013245211e-06,
+      "loss": 3.8477,
+      "step": 2612224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.221955065734691e-06,
+      "loss": 3.8414,
+      "step": 2612736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.213569118224171e-06,
+      "loss": 3.8506,
+      "step": 2613248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.205183170713651e-06,
+      "loss": 3.8341,
+      "step": 2613760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.196813602006862e-06,
+      "loss": 3.8505,
+      "step": 2614272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.188427654496342e-06,
+      "loss": 3.8363,
+      "step": 2614784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.180041706985823e-06,
+      "loss": 3.8382,
+      "step": 2615296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.171655759475301e-06,
+      "loss": 3.8437,
+      "step": 2615808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.163269811964782e-06,
+      "loss": 3.8542,
+      "step": 2616320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.1549002432579936e-06,
+      "loss": 3.8451,
+      "step": 2616832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.1465142957474735e-06,
+      "loss": 3.8396,
+      "step": 2617344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.138128348236953e-06,
+      "loss": 3.8481,
+      "step": 2617856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.1297424007264326e-06,
+      "loss": 3.8427,
+      "step": 2618368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.121372832019644e-06,
+      "loss": 3.85,
+      "step": 2618880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.112986884509125e-06,
+      "loss": 3.8323,
+      "step": 2619392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.104617315802335e-06,
+      "loss": 3.8325,
+      "step": 2619904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.096231368291816e-06,
+      "loss": 3.8523,
+      "step": 2620416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.087845420781296e-06,
+      "loss": 3.8504,
+      "step": 2620928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.079459473270775e-06,
+      "loss": 3.8376,
+      "step": 2621440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.0710899045639865e-06,
+      "loss": 3.8393,
+      "step": 2621952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.062703957053466e-06,
+      "loss": 3.8425,
+      "step": 2622464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.054318009542947e-06,
+      "loss": 3.8266,
+      "step": 2622976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.045932062032426e-06,
+      "loss": 3.8536,
+      "step": 2623488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.037546114521906e-06,
+      "loss": 3.8388,
+      "step": 2624000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.029176545815118e-06,
+      "loss": 3.8458,
+      "step": 2624512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.020790598304598e-06,
+      "loss": 3.8409,
+      "step": 2625024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.012404650794077e-06,
+      "loss": 3.8414,
+      "step": 2625536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.004018703283558e-06,
+      "loss": 3.8402,
+      "step": 2626048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.995632755773038e-06,
+      "loss": 3.8407,
+      "step": 2626560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.987263187066248e-06,
+      "loss": 3.8319,
+      "step": 2627072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.9788772395557285e-06,
+      "loss": 3.8378,
+      "step": 2627584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.9704912920452084e-06,
+      "loss": 3.8526,
+      "step": 2628096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.962105344534689e-06,
+      "loss": 3.8439,
+      "step": 2628608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.9537193970241675e-06,
+      "loss": 3.8403,
+      "step": 2629120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.945333449513648e-06,
+      "loss": 3.8301,
+      "step": 2629632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.936947502003128e-06,
+      "loss": 3.8388,
+      "step": 2630144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.928561554492607e-06,
+      "loss": 3.8548,
+      "step": 2630656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.920191985785819e-06,
+      "loss": 3.8524,
+      "step": 2631168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.9118060382753e-06,
+      "loss": 3.8446,
+      "step": 2631680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.90342009076478e-06,
+      "loss": 3.8538,
+      "step": 2632192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.895034143254259e-06,
+      "loss": 3.8537,
+      "step": 2632704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.8866645745474705e-06,
+      "loss": 3.8406,
+      "step": 2633216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.8782786270369505e-06,
+      "loss": 3.848,
+      "step": 2633728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.86989267952643e-06,
+      "loss": 3.8463,
+      "step": 2634240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.8615067320159095e-06,
+      "loss": 3.837,
+      "step": 2634752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.853137163309121e-06,
+      "loss": 3.8441,
+      "step": 2635264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.844751215798602e-06,
+      "loss": 3.8385,
+      "step": 2635776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.836365268288081e-06,
+      "loss": 3.8461,
+      "step": 2636288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.827979320777561e-06,
+      "loss": 3.86,
+      "step": 2636800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.819609752070773e-06,
+      "loss": 3.8558,
+      "step": 2637312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.811223804560253e-06,
+      "loss": 3.8316,
+      "step": 2637824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.802837857049732e-06,
+      "loss": 3.8481,
+      "step": 2638336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.7944519095392126e-06,
+      "loss": 3.8455,
+      "step": 2638848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.7860659620286925e-06,
+      "loss": 3.8466,
+      "step": 2639360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.777680014518172e-06,
+      "loss": 3.8435,
+      "step": 2639872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.7692940670076516e-06,
+      "loss": 3.8379,
+      "step": 2640384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.760908119497132e-06,
+      "loss": 3.8448,
+      "step": 2640896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.752538550790344e-06,
+      "loss": 3.8411,
+      "step": 2641408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.744152603279823e-06,
+      "loss": 3.8372,
+      "step": 2641920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.735766655769303e-06,
+      "loss": 3.8495,
+      "step": 2642432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.727397087062515e-06,
+      "loss": 3.8412,
+      "step": 2642944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.719011139551994e-06,
+      "loss": 3.8536,
+      "step": 2643456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.710625192041474e-06,
+      "loss": 3.8395,
+      "step": 2643968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.702239244530955e-06,
+      "loss": 3.8403,
+      "step": 2644480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.693853297020433e-06,
+      "loss": 3.8434,
+      "step": 2644992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.685467349509914e-06,
+      "loss": 3.8487,
+      "step": 2645504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.677081401999394e-06,
+      "loss": 3.8477,
+      "step": 2646016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.668695454488874e-06,
+      "loss": 3.8373,
+      "step": 2646528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.660309506978353e-06,
+      "loss": 3.8427,
+      "step": 2647040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.6519235594678335e-06,
+      "loss": 3.8423,
+      "step": 2647552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.643537611957313e-06,
+      "loss": 3.8488,
+      "step": 2648064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.635168043250524e-06,
+      "loss": 3.8318,
+      "step": 2648576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.626782095740004e-06,
+      "loss": 3.8355,
+      "step": 2649088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.618396148229485e-06,
+      "loss": 3.8503,
+      "step": 2649600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.610010200718965e-06,
+      "loss": 3.8414,
+      "step": 2650112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.601624253208444e-06,
+      "loss": 3.834,
+      "step": 2650624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.593238305697924e-06,
+      "loss": 3.8347,
+      "step": 2651136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.584852358187405e-06,
+      "loss": 3.8327,
+      "step": 2651648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.576466410676883e-06,
+      "loss": 3.8507,
+      "step": 2652160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.568080463166364e-06,
+      "loss": 3.8422,
+      "step": 2652672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.5597108944595755e-06,
+      "loss": 3.8412,
+      "step": 2653184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.5513249469490554e-06,
+      "loss": 3.8445,
+      "step": 2653696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.5429389994385346e-06,
+      "loss": 3.8593,
+      "step": 2654208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.5345530519280145e-06,
+      "loss": 3.8477,
+      "step": 2654720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.526183483221226e-06,
+      "loss": 3.8344,
+      "step": 2655232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.517797535710705e-06,
+      "loss": 3.8447,
+      "step": 2655744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.509411588200186e-06,
+      "loss": 3.8467,
+      "step": 2656256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.501042019493398e-06,
+      "loss": 3.8391,
+      "step": 2656768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.492672450786609e-06,
+      "loss": 3.8423,
+      "step": 2657280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.484286503276088e-06,
+      "loss": 3.8383,
+      "step": 2657792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.475900555765568e-06,
+      "loss": 3.8507,
+      "step": 2658304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.467514608255048e-06,
+      "loss": 3.8372,
+      "step": 2658816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.459128660744529e-06,
+      "loss": 3.85,
+      "step": 2659328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.450742713234008e-06,
+      "loss": 3.8349,
+      "step": 2659840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.442356765723488e-06,
+      "loss": 3.8472,
+      "step": 2660352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.433970818212968e-06,
+      "loss": 3.8346,
+      "step": 2660864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.425584870702447e-06,
+      "loss": 3.8575,
+      "step": 2661376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.417198923191928e-06,
+      "loss": 3.8496,
+      "step": 2661888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.408812975681408e-06,
+      "loss": 3.8418,
+      "step": 2662400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.400427028170887e-06,
+      "loss": 3.8379,
+      "step": 2662912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.392057459464099e-06,
+      "loss": 3.8493,
+      "step": 2663424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.383671511953579e-06,
+      "loss": 3.8441,
+      "step": 2663936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.3752855644430596e-06,
+      "loss": 3.8546,
+      "step": 2664448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.3669159957362695e-06,
+      "loss": 3.8388,
+      "step": 2664960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.3585300482257495e-06,
+      "loss": 3.8528,
+      "step": 2665472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.35014410071523e-06,
+      "loss": 3.8455,
+      "step": 2665984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.34175815320471e-06,
+      "loss": 3.856,
+      "step": 2666496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.333372205694189e-06,
+      "loss": 3.8418,
+      "step": 2667008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.324986258183669e-06,
+      "loss": 3.8482,
+      "step": 2667520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.31660031067315e-06,
+      "loss": 3.8476,
+      "step": 2668032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.308214363162629e-06,
+      "loss": 3.8374,
+      "step": 2668544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.299861173259572e-06,
+      "loss": 3.8538,
+      "step": 2669056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.2914752257490525e-06,
+      "loss": 3.8524,
+      "step": 2669568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.283089278238532e-06,
+      "loss": 3.8391,
+      "step": 2670080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.2747033307280115e-06,
+      "loss": 3.8494,
+      "step": 2670592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.266333762021223e-06,
+      "loss": 3.8389,
+      "step": 2671104
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9745423793792725,
+      "eval_runtime": 295.713,
+      "eval_samples_per_second": 1290.41,
+      "eval_steps_per_second": 40.326,
+      "step": 2671200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.257947814510703e-06,
+      "loss": 3.8407,
+      "step": 2671616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.249561867000183e-06,
+      "loss": 3.8394,
+      "step": 2672128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.241175919489663e-06,
+      "loss": 3.8473,
+      "step": 2672640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.232789971979143e-06,
+      "loss": 3.8476,
+      "step": 2673152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.224404024468623e-06,
+      "loss": 3.8616,
+      "step": 2673664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.216018076958103e-06,
+      "loss": 3.8395,
+      "step": 2674176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.207632129447583e-06,
+      "loss": 3.8445,
+      "step": 2674688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.199246181937062e-06,
+      "loss": 3.8393,
+      "step": 2675200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.190860234426543e-06,
+      "loss": 3.8578,
+      "step": 2675712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.182474286916022e-06,
+      "loss": 3.8414,
+      "step": 2676224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.174088339405503e-06,
+      "loss": 3.8408,
+      "step": 2676736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.165702391894982e-06,
+      "loss": 3.8483,
+      "step": 2677248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.157332823188193e-06,
+      "loss": 3.8425,
+      "step": 2677760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.148946875677673e-06,
+      "loss": 3.8381,
+      "step": 2678272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.140560928167153e-06,
+      "loss": 3.8408,
+      "step": 2678784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.132174980656633e-06,
+      "loss": 3.835,
+      "step": 2679296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.123789033146113e-06,
+      "loss": 3.8335,
+      "step": 2679808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.115403085635593e-06,
+      "loss": 3.8344,
+      "step": 2680320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.107017138125073e-06,
+      "loss": 3.8356,
+      "step": 2680832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.098631190614552e-06,
+      "loss": 3.8627,
+      "step": 2681344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.090245243104033e-06,
+      "loss": 3.8391,
+      "step": 2681856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.081859295593512e-06,
+      "loss": 3.8401,
+      "step": 2682368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.073489726886724e-06,
+      "loss": 3.8451,
+      "step": 2682880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.065103779376204e-06,
+      "loss": 3.8528,
+      "step": 2683392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.056717831865684e-06,
+      "loss": 3.8375,
+      "step": 2683904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.048331884355164e-06,
+      "loss": 3.8411,
+      "step": 2684416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.039945936844643e-06,
+      "loss": 3.8369,
+      "step": 2684928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.031559989334124e-06,
+      "loss": 3.8432,
+      "step": 2685440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.023174041823603e-06,
+      "loss": 3.834,
+      "step": 2685952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.014804473116814e-06,
+      "loss": 3.8391,
+      "step": 2686464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.006418525606294e-06,
+      "loss": 3.842,
+      "step": 2686976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.998032578095775e-06,
+      "loss": 3.8487,
+      "step": 2687488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.989646630585254e-06,
+      "loss": 3.8562,
+      "step": 2688000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.981260683074734e-06,
+      "loss": 3.8472,
+      "step": 2688512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.972874735564214e-06,
+      "loss": 3.8388,
+      "step": 2689024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.964488788053694e-06,
+      "loss": 3.8475,
+      "step": 2689536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.956119219346906e-06,
+      "loss": 3.8357,
+      "step": 2690048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.947733271836385e-06,
+      "loss": 3.845,
+      "step": 2690560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.939347324325866e-06,
+      "loss": 3.8317,
+      "step": 2691072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.930961376815345e-06,
+      "loss": 3.836,
+      "step": 2691584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.922575429304825e-06,
+      "loss": 3.8383,
+      "step": 2692096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.914205860598036e-06,
+      "loss": 3.8554,
+      "step": 2692608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.905819913087516e-06,
+      "loss": 3.8471,
+      "step": 2693120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897433965576996e-06,
+      "loss": 3.8369,
+      "step": 2693632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.889048018066476e-06,
+      "loss": 3.8428,
+      "step": 2694144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.880662070555955e-06,
+      "loss": 3.8385,
+      "step": 2694656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.872292501849167e-06,
+      "loss": 3.8478,
+      "step": 2695168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.863906554338647e-06,
+      "loss": 3.8321,
+      "step": 2695680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.855520606828127e-06,
+      "loss": 3.8275,
+      "step": 2696192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.847134659317607e-06,
+      "loss": 3.8517,
+      "step": 2696704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8387650906108184e-06,
+      "loss": 3.8484,
+      "step": 2697216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.830379143100298e-06,
+      "loss": 3.832,
+      "step": 2697728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.821993195589778e-06,
+      "loss": 3.84,
+      "step": 2698240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.813607248079258e-06,
+      "loss": 3.8425,
+      "step": 2698752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.805221300568737e-06,
+      "loss": 3.8244,
+      "step": 2699264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.796851731861949e-06,
+      "loss": 3.8463,
+      "step": 2699776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.788465784351429e-06,
+      "loss": 3.8338,
+      "step": 2700288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.780079836840909e-06,
+      "loss": 3.8466,
+      "step": 2700800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.771693889330389e-06,
+      "loss": 3.8324,
+      "step": 2701312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.763307941819869e-06,
+      "loss": 3.8445,
+      "step": 2701824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.754921994309349e-06,
+      "loss": 3.8338,
+      "step": 2702336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7465524256025605e-06,
+      "loss": 3.8393,
+      "step": 2702848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7381664780920396e-06,
+      "loss": 3.8268,
+      "step": 2703360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7297805305815195e-06,
+      "loss": 3.838,
+      "step": 2703872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7213945830709995e-06,
+      "loss": 3.8529,
+      "step": 2704384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7130086355604794e-06,
+      "loss": 3.837,
+      "step": 2704896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.704639066853691e-06,
+      "loss": 3.84,
+      "step": 2705408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.696253119343171e-06,
+      "loss": 3.8266,
+      "step": 2705920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.687867171832651e-06,
+      "loss": 3.8374,
+      "step": 2706432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.679481224322131e-06,
+      "loss": 3.8518,
+      "step": 2706944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.67109527681161e-06,
+      "loss": 3.8481,
+      "step": 2707456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.662709329301091e-06,
+      "loss": 3.8435,
+      "step": 2707968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.65432338179057e-06,
+      "loss": 3.8512,
+      "step": 2708480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.645953813083782e-06,
+      "loss": 3.8495,
+      "step": 2708992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6375678655732616e-06,
+      "loss": 3.8417,
+      "step": 2709504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6291819180627415e-06,
+      "loss": 3.8437,
+      "step": 2710016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6207959705522215e-06,
+      "loss": 3.8439,
+      "step": 2710528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.612426401845433e-06,
+      "loss": 3.8378,
+      "step": 2711040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.604040454334913e-06,
+      "loss": 3.8407,
+      "step": 2711552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.595654506824392e-06,
+      "loss": 3.8372,
+      "step": 2712064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.587268559313873e-06,
+      "loss": 3.8409,
+      "step": 2712576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.578882611803352e-06,
+      "loss": 3.8558,
+      "step": 2713088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.570513043096564e-06,
+      "loss": 3.8518,
+      "step": 2713600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.562127095586044e-06,
+      "loss": 3.8283,
+      "step": 2714112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.553741148075524e-06,
+      "loss": 3.851,
+      "step": 2714624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.545355200565004e-06,
+      "loss": 3.8442,
+      "step": 2715136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.536969253054483e-06,
+      "loss": 3.8416,
+      "step": 2715648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5285833055439635e-06,
+      "loss": 3.8405,
+      "step": 2716160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5201973580334435e-06,
+      "loss": 3.833,
+      "step": 2716672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.511827789326655e-06,
+      "loss": 3.8438,
+      "step": 2717184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.503441841816134e-06,
+      "loss": 3.8378,
+      "step": 2717696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.495055894305615e-06,
+      "loss": 3.8377,
+      "step": 2718208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.486669946795094e-06,
+      "loss": 3.8446,
+      "step": 2718720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.478300378088306e-06,
+      "loss": 3.8391,
+      "step": 2719232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4699308093815165e-06,
+      "loss": 3.8511,
+      "step": 2719744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.461544861870997e-06,
+      "loss": 3.8377,
+      "step": 2720256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4531589143604764e-06,
+      "loss": 3.8352,
+      "step": 2720768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.444789345653688e-06,
+      "loss": 3.8387,
+      "step": 2721280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.436403398143168e-06,
+      "loss": 3.8489,
+      "step": 2721792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.428017450632648e-06,
+      "loss": 3.8438,
+      "step": 2722304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.419631503122128e-06,
+      "loss": 3.8367,
+      "step": 2722816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.411245555611608e-06,
+      "loss": 3.837,
+      "step": 2723328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.402859608101088e-06,
+      "loss": 3.8411,
+      "step": 2723840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.394473660590568e-06,
+      "loss": 3.8498,
+      "step": 2724352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.386087713080047e-06,
+      "loss": 3.8334,
+      "step": 2724864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.377701765569528e-06,
+      "loss": 3.8318,
+      "step": 2725376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.369315818059007e-06,
+      "loss": 3.8479,
+      "step": 2725888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.360929870548488e-06,
+      "loss": 3.8357,
+      "step": 2726400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.352543923037967e-06,
+      "loss": 3.8315,
+      "step": 2726912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.344174354331178e-06,
+      "loss": 3.8314,
+      "step": 2727424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.335788406820658e-06,
+      "loss": 3.8343,
+      "step": 2727936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.327402459310138e-06,
+      "loss": 3.8481,
+      "step": 2728448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.319016511799618e-06,
+      "loss": 3.8344,
+      "step": 2728960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.310646943092829e-06,
+      "loss": 3.8388,
+      "step": 2729472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.30226099558231e-06,
+      "loss": 3.8441,
+      "step": 2729984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.293875048071789e-06,
+      "loss": 3.8559,
+      "step": 2730496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.285489100561269e-06,
+      "loss": 3.8506,
+      "step": 2731008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.277103153050749e-06,
+      "loss": 3.8298,
+      "step": 2731520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.26873358434396e-06,
+      "loss": 3.8418,
+      "step": 2732032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2603476368334405e-06,
+      "loss": 3.8407,
+      "step": 2732544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2519616893229196e-06,
+      "loss": 3.8404,
+      "step": 2733056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.243592120616131e-06,
+      "loss": 3.8421,
+      "step": 2733568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.235206173105611e-06,
+      "loss": 3.8342,
+      "step": 2734080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.226820225595092e-06,
+      "loss": 3.8466,
+      "step": 2734592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.218434278084571e-06,
+      "loss": 3.8365,
+      "step": 2735104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.210048330574051e-06,
+      "loss": 3.8444,
+      "step": 2735616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.201662383063531e-06,
+      "loss": 3.835,
+      "step": 2736128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.193276435553011e-06,
+      "loss": 3.8443,
+      "step": 2736640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.184890488042491e-06,
+      "loss": 3.8312,
+      "step": 2737152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.176520919335702e-06,
+      "loss": 3.8592,
+      "step": 2737664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1681349718251825e-06,
+      "loss": 3.8417,
+      "step": 2738176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.159765403118393e-06,
+      "loss": 3.8434,
+      "step": 2738688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.151379455607873e-06,
+      "loss": 3.8356,
+      "step": 2739200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.142993508097353e-06,
+      "loss": 3.8486,
+      "step": 2739712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.134607560586833e-06,
+      "loss": 3.8429,
+      "step": 2740224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.126237991880045e-06,
+      "loss": 3.8498,
+      "step": 2740736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.117852044369524e-06,
+      "loss": 3.8376,
+      "step": 2741248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.109466096859005e-06,
+      "loss": 3.8489,
+      "step": 2741760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.101080149348484e-06,
+      "loss": 3.8445,
+      "step": 2742272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.092694201837965e-06,
+      "loss": 3.8556,
+      "step": 2742784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.084308254327444e-06,
+      "loss": 3.8358,
+      "step": 2743296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.075922306816924e-06,
+      "loss": 3.8457,
+      "step": 2743808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.067536359306404e-06,
+      "loss": 3.8438,
+      "step": 2744320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.059166790599615e-06,
+      "loss": 3.8404,
+      "step": 2744832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.050780843089095e-06,
+      "loss": 3.8484,
+      "step": 2745344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.042394895578575e-06,
+      "loss": 3.8482,
+      "step": 2745856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.034008948068055e-06,
+      "loss": 3.8397,
+      "step": 2746368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.025639379361266e-06,
+      "loss": 3.8442,
+      "step": 2746880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.017253431850747e-06,
+      "loss": 3.841,
+      "step": 2747392
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9738526344299316,
+      "eval_runtime": 300.6975,
+      "eval_samples_per_second": 1269.02,
+      "eval_steps_per_second": 39.658,
+      "step": 2747520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.008867484340226e-06,
+      "loss": 3.8397,
+      "step": 2747904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.000481536829706e-06,
+      "loss": 3.8345,
+      "step": 2748416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.992095589319186e-06,
+      "loss": 3.8435,
+      "step": 2748928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.983709641808666e-06,
+      "loss": 3.8441,
+      "step": 2749440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.975323694298146e-06,
+      "loss": 3.8582,
+      "step": 2749952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.966937746787626e-06,
+      "loss": 3.841,
+      "step": 2750464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.958568178080837e-06,
+      "loss": 3.84,
+      "step": 2750976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.950182230570316e-06,
+      "loss": 3.8385,
+      "step": 2751488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.941796283059796e-06,
+      "loss": 3.8526,
+      "step": 2752000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.933410335549276e-06,
+      "loss": 3.8422,
+      "step": 2752512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.925040766842488e-06,
+      "loss": 3.8376,
+      "step": 2753024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.916654819331968e-06,
+      "loss": 3.8453,
+      "step": 2753536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.908285250625179e-06,
+      "loss": 3.8374,
+      "step": 2754048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8998993031146595e-06,
+      "loss": 3.8386,
+      "step": 2754560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8915133556041386e-06,
+      "loss": 3.8447,
+      "step": 2755072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.883127408093619e-06,
+      "loss": 3.8289,
+      "step": 2755584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8747414605830985e-06,
+      "loss": 3.8342,
+      "step": 2756096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8663555130725784e-06,
+      "loss": 3.832,
+      "step": 2756608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.857969565562058e-06,
+      "loss": 3.8316,
+      "step": 2757120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.849583618051538e-06,
+      "loss": 3.8556,
+      "step": 2757632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.841197670541018e-06,
+      "loss": 3.8425,
+      "step": 2758144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.832811723030498e-06,
+      "loss": 3.837,
+      "step": 2758656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.824425775519978e-06,
+      "loss": 3.8437,
+      "step": 2759168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.81605620681319e-06,
+      "loss": 3.8514,
+      "step": 2759680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.807670259302669e-06,
+      "loss": 3.8345,
+      "step": 2760192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.79928431179215e-06,
+      "loss": 3.8407,
+      "step": 2760704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.790898364281629e-06,
+      "loss": 3.8349,
+      "step": 2761216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.78251241677111e-06,
+      "loss": 3.8379,
+      "step": 2761728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.774126469260589e-06,
+      "loss": 3.8335,
+      "step": 2762240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.765740521750069e-06,
+      "loss": 3.8372,
+      "step": 2762752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.757354574239549e-06,
+      "loss": 3.8391,
+      "step": 2763264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.74898500553276e-06,
+      "loss": 3.845,
+      "step": 2763776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.740615436825972e-06,
+      "loss": 3.8531,
+      "step": 2764288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.732229489315451e-06,
+      "loss": 3.8466,
+      "step": 2764800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.723843541804932e-06,
+      "loss": 3.8336,
+      "step": 2765312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.715457594294411e-06,
+      "loss": 3.8432,
+      "step": 2765824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.707071646783892e-06,
+      "loss": 3.8379,
+      "step": 2766336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.698685699273371e-06,
+      "loss": 3.839,
+      "step": 2766848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.690299751762851e-06,
+      "loss": 3.8297,
+      "step": 2767360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6819301830560625e-06,
+      "loss": 3.8338,
+      "step": 2767872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.673544235545542e-06,
+      "loss": 3.8386,
+      "step": 2768384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.665158288035022e-06,
+      "loss": 3.8522,
+      "step": 2768896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6567723405245015e-06,
+      "loss": 3.8434,
+      "step": 2769408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6483863930139815e-06,
+      "loss": 3.8361,
+      "step": 2769920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.640000445503461e-06,
+      "loss": 3.8406,
+      "step": 2770432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.631614497992941e-06,
+      "loss": 3.8355,
+      "step": 2770944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.623244929286153e-06,
+      "loss": 3.8459,
+      "step": 2771456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.614858981775633e-06,
+      "loss": 3.8333,
+      "step": 2771968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.606473034265113e-06,
+      "loss": 3.8229,
+      "step": 2772480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.598087086754593e-06,
+      "loss": 3.847,
+      "step": 2772992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.589701139244072e-06,
+      "loss": 3.8446,
+      "step": 2773504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.581331570537284e-06,
+      "loss": 3.8293,
+      "step": 2774016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.572945623026764e-06,
+      "loss": 3.8377,
+      "step": 2774528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.564576054319975e-06,
+      "loss": 3.8412,
+      "step": 2775040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.556190106809455e-06,
+      "loss": 3.8214,
+      "step": 2775552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.547804159298935e-06,
+      "loss": 3.8436,
+      "step": 2776064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.539418211788415e-06,
+      "loss": 3.8327,
+      "step": 2776576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.531032264277895e-06,
+      "loss": 3.8424,
+      "step": 2777088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.522646316767375e-06,
+      "loss": 3.8302,
+      "step": 2777600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.514260369256854e-06,
+      "loss": 3.8401,
+      "step": 2778112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.505874421746335e-06,
+      "loss": 3.8323,
+      "step": 2778624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497488474235814e-06,
+      "loss": 3.8402,
+      "step": 2779136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.489118905529026e-06,
+      "loss": 3.8241,
+      "step": 2779648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.480732958018506e-06,
+      "loss": 3.8384,
+      "step": 2780160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4723470105079856e-06,
+      "loss": 3.8459,
+      "step": 2780672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4639610629974655e-06,
+      "loss": 3.8391,
+      "step": 2781184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.455575115486945e-06,
+      "loss": 3.8342,
+      "step": 2781696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4471891679764254e-06,
+      "loss": 3.8246,
+      "step": 2782208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.438803220465905e-06,
+      "loss": 3.8323,
+      "step": 2782720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.430417272955385e-06,
+      "loss": 3.8455,
+      "step": 2783232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.422047704248596e-06,
+      "loss": 3.8491,
+      "step": 2783744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.413661756738077e-06,
+      "loss": 3.8422,
+      "step": 2784256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.405275809227556e-06,
+      "loss": 3.85,
+      "step": 2784768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.396889861717036e-06,
+      "loss": 3.8477,
+      "step": 2785280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.388520293010248e-06,
+      "loss": 3.8385,
+      "step": 2785792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.380134345499727e-06,
+      "loss": 3.8384,
+      "step": 2786304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3717483979892075e-06,
+      "loss": 3.8444,
+      "step": 2786816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.363362450478687e-06,
+      "loss": 3.8338,
+      "step": 2787328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.354992881771898e-06,
+      "loss": 3.8348,
+      "step": 2787840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.346606934261378e-06,
+      "loss": 3.8348,
+      "step": 2788352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.33823736555459e-06,
+      "loss": 3.8375,
+      "step": 2788864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.32985141804407e-06,
+      "loss": 3.8544,
+      "step": 2789376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.32146547053355e-06,
+      "loss": 3.8513,
+      "step": 2789888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.31307952302303e-06,
+      "loss": 3.828,
+      "step": 2790400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.304709954316241e-06,
+      "loss": 3.8433,
+      "step": 2790912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2963240068057205e-06,
+      "loss": 3.844,
+      "step": 2791424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2879380592952005e-06,
+      "loss": 3.8421,
+      "step": 2791936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.27955211178468e-06,
+      "loss": 3.8405,
+      "step": 2792448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.27116616427416e-06,
+      "loss": 3.8298,
+      "step": 2792960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.26278021676364e-06,
+      "loss": 3.8392,
+      "step": 2793472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.254410648056852e-06,
+      "loss": 3.8366,
+      "step": 2793984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.246024700546332e-06,
+      "loss": 3.8358,
+      "step": 2794496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.237638753035812e-06,
+      "loss": 3.8369,
+      "step": 2795008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.229252805525291e-06,
+      "loss": 3.8431,
+      "step": 2795520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.220866858014772e-06,
+      "loss": 3.8439,
+      "step": 2796032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.212480910504251e-06,
+      "loss": 3.8443,
+      "step": 2796544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.204094962993732e-06,
+      "loss": 3.8298,
+      "step": 2797056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.195709015483211e-06,
+      "loss": 3.8391,
+      "step": 2797568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1873394467764224e-06,
+      "loss": 3.8423,
+      "step": 2798080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.178953499265902e-06,
+      "loss": 3.8445,
+      "step": 2798592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1705675517553815e-06,
+      "loss": 3.8331,
+      "step": 2799104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.162181604244862e-06,
+      "loss": 3.8357,
+      "step": 2799616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.153812035538073e-06,
+      "loss": 3.8383,
+      "step": 2800128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.145426088027553e-06,
+      "loss": 3.8463,
+      "step": 2800640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.137040140517033e-06,
+      "loss": 3.8324,
+      "step": 2801152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.128654193006514e-06,
+      "loss": 3.8306,
+      "step": 2801664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.120268245495993e-06,
+      "loss": 3.8397,
+      "step": 2802176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.111882297985473e-06,
+      "loss": 3.8367,
+      "step": 2802688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1035127292786845e-06,
+      "loss": 3.8293,
+      "step": 2803200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.095126781768164e-06,
+      "loss": 3.8312,
+      "step": 2803712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0867408342576444e-06,
+      "loss": 3.8322,
+      "step": 2804224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0783548867471235e-06,
+      "loss": 3.8471,
+      "step": 2804736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.070001696844067e-06,
+      "loss": 3.8353,
+      "step": 2805248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.061615749333547e-06,
+      "loss": 3.8327,
+      "step": 2805760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.053229801823027e-06,
+      "loss": 3.8435,
+      "step": 2806272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.044843854312507e-06,
+      "loss": 3.8543,
+      "step": 2806784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.036457906801986e-06,
+      "loss": 3.8474,
+      "step": 2807296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.028071959291467e-06,
+      "loss": 3.8296,
+      "step": 2807808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.019686011780946e-06,
+      "loss": 3.837,
+      "step": 2808320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0113000642704265e-06,
+      "loss": 3.8402,
+      "step": 2808832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.002930495563637e-06,
+      "loss": 3.8359,
+      "step": 2809344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.994560926856849e-06,
+      "loss": 3.8411,
+      "step": 2809856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.986174979346328e-06,
+      "loss": 3.8302,
+      "step": 2810368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.977789031835809e-06,
+      "loss": 3.8414,
+      "step": 2810880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.969403084325289e-06,
+      "loss": 3.8397,
+      "step": 2811392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.961017136814768e-06,
+      "loss": 3.8417,
+      "step": 2811904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.952631189304249e-06,
+      "loss": 3.8306,
+      "step": 2812416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.944277999401191e-06,
+      "loss": 3.8384,
+      "step": 2812928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.935892051890671e-06,
+      "loss": 3.8364,
+      "step": 2813440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.92750610438015e-06,
+      "loss": 3.8509,
+      "step": 2813952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.919120156869631e-06,
+      "loss": 3.8386,
+      "step": 2814464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.91073420935911e-06,
+      "loss": 3.8444,
+      "step": 2814976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.902348261848591e-06,
+      "loss": 3.8326,
+      "step": 2815488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.89396231433807e-06,
+      "loss": 3.8473,
+      "step": 2816000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.88557636682755e-06,
+      "loss": 3.8425,
+      "step": 2816512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.87719041931703e-06,
+      "loss": 3.8465,
+      "step": 2817024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.868820850610242e-06,
+      "loss": 3.8333,
+      "step": 2817536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.860434903099722e-06,
+      "loss": 3.8511,
+      "step": 2818048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8520489555892016e-06,
+      "loss": 3.8392,
+      "step": 2818560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8436630080786815e-06,
+      "loss": 3.8539,
+      "step": 2819072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8352770605681615e-06,
+      "loss": 3.8348,
+      "step": 2819584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.826891113057641e-06,
+      "loss": 3.8447,
+      "step": 2820096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.818505165547121e-06,
+      "loss": 3.8405,
+      "step": 2820608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.810135596840332e-06,
+      "loss": 3.8345,
+      "step": 2821120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8017496493298126e-06,
+      "loss": 3.8462,
+      "step": 2821632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.793363701819292e-06,
+      "loss": 3.8455,
+      "step": 2822144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7849777543087725e-06,
+      "loss": 3.8369,
+      "step": 2822656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7766081856019833e-06,
+      "loss": 3.8408,
+      "step": 2823168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7682222380914636e-06,
+      "loss": 3.84,
+      "step": 2823680
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9733495712280273,
+      "eval_runtime": 298.9858,
+      "eval_samples_per_second": 1276.285,
+      "eval_steps_per_second": 39.885,
+      "step": 2823840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.759836290580943e-06,
+      "loss": 3.8384,
+      "step": 2824192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7514503430704227e-06,
+      "loss": 3.8337,
+      "step": 2824704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.743064395559903e-06,
+      "loss": 3.8425,
+      "step": 2825216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7346784480493826e-06,
+      "loss": 3.8412,
+      "step": 2825728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.726292500538863e-06,
+      "loss": 3.8536,
+      "step": 2826240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7179065530283425e-06,
+      "loss": 3.8373,
+      "step": 2826752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7095206055178225e-06,
+      "loss": 3.8368,
+      "step": 2827264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.701134658007303e-06,
+      "loss": 3.8397,
+      "step": 2827776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6927487104967824e-06,
+      "loss": 3.8491,
+      "step": 2828288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6843627629862628e-06,
+      "loss": 3.8405,
+      "step": 2828800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6759768154757423e-06,
+      "loss": 3.8386,
+      "step": 2829312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.667590867965222e-06,
+      "loss": 3.8388,
+      "step": 2829824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6592212992584335e-06,
+      "loss": 3.8384,
+      "step": 2830336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.650835351747913e-06,
+      "loss": 3.8355,
+      "step": 2830848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6424494042373934e-06,
+      "loss": 3.8414,
+      "step": 2831360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.634063456726873e-06,
+      "loss": 3.8295,
+      "step": 2831872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6256938880200846e-06,
+      "loss": 3.8309,
+      "step": 2832384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6173079405095645e-06,
+      "loss": 3.8311,
+      "step": 2832896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6089219929990445e-06,
+      "loss": 3.8271,
+      "step": 2833408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6005360454885244e-06,
+      "loss": 3.8523,
+      "step": 2833920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.592150097978004e-06,
+      "loss": 3.8429,
+      "step": 2834432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5837641504674843e-06,
+      "loss": 3.8381,
+      "step": 2834944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.575378202956964e-06,
+      "loss": 3.8368,
+      "step": 2835456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5669922554464442e-06,
+      "loss": 3.8486,
+      "step": 2835968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5586063079359238e-06,
+      "loss": 3.8349,
+      "step": 2836480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5502367392291354e-06,
+      "loss": 3.8394,
+      "step": 2836992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.541850791718615e-06,
+      "loss": 3.8296,
+      "step": 2837504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5334648442080945e-06,
+      "loss": 3.8376,
+      "step": 2838016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.525078896697575e-06,
+      "loss": 3.8345,
+      "step": 2838528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5166929491870544e-06,
+      "loss": 3.8321,
+      "step": 2839040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5083070016765343e-06,
+      "loss": 3.8375,
+      "step": 2839552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.499937432969746e-06,
+      "loss": 3.8422,
+      "step": 2840064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4915514854592255e-06,
+      "loss": 3.8478,
+      "step": 2840576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.483165537948706e-06,
+      "loss": 3.8453,
+      "step": 2841088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4747795904381854e-06,
+      "loss": 3.8321,
+      "step": 2841600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.466393642927666e-06,
+      "loss": 3.8441,
+      "step": 2842112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4580076954171453e-06,
+      "loss": 3.8347,
+      "step": 2842624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.449621747906625e-06,
+      "loss": 3.8295,
+      "step": 2843136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4412358003961053e-06,
+      "loss": 3.8329,
+      "step": 2843648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4328498528855848e-06,
+      "loss": 3.8306,
+      "step": 2844160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4244802841787964e-06,
+      "loss": 3.8348,
+      "step": 2844672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.416094336668276e-06,
+      "loss": 3.8513,
+      "step": 2845184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4077083891577563e-06,
+      "loss": 3.8374,
+      "step": 2845696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.399322441647236e-06,
+      "loss": 3.8336,
+      "step": 2846208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.390952872940448e-06,
+      "loss": 3.8429,
+      "step": 2846720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3825669254299275e-06,
+      "loss": 3.8349,
+      "step": 2847232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.374180977919407e-06,
+      "loss": 3.838,
+      "step": 2847744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3657950304088874e-06,
+      "loss": 3.831,
+      "step": 2848256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.357409082898367e-06,
+      "loss": 3.8188,
+      "step": 2848768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3490395141915785e-06,
+      "loss": 3.8495,
+      "step": 2849280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3406699454847893e-06,
+      "loss": 3.8398,
+      "step": 2849792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3322839979742697e-06,
+      "loss": 3.8326,
+      "step": 2850304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3238980504637492e-06,
+      "loss": 3.8316,
+      "step": 2850816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3155121029532296e-06,
+      "loss": 3.837,
+      "step": 2851328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3071261554427096e-06,
+      "loss": 3.8215,
+      "step": 2851840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2987565867359212e-06,
+      "loss": 3.838,
+      "step": 2852352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2903706392254007e-06,
+      "loss": 3.8342,
+      "step": 2852864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2819846917148803e-06,
+      "loss": 3.8382,
+      "step": 2853376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2735987442043607e-06,
+      "loss": 3.8323,
+      "step": 2853888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.26521279669384e-06,
+      "loss": 3.835,
+      "step": 2854400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2568268491833206e-06,
+      "loss": 3.8294,
+      "step": 2854912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2484409016728e-06,
+      "loss": 3.8399,
+      "step": 2855424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2400549541622796e-06,
+      "loss": 3.823,
+      "step": 2855936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2316853854554913e-06,
+      "loss": 3.8328,
+      "step": 2856448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2232994379449712e-06,
+      "loss": 3.8442,
+      "step": 2856960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.214913490434451e-06,
+      "loss": 3.8376,
+      "step": 2857472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.206527542923931e-06,
+      "loss": 3.8308,
+      "step": 2857984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1981579742171428e-06,
+      "loss": 3.8257,
+      "step": 2858496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1897720267066223e-06,
+      "loss": 3.8298,
+      "step": 2859008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1813860791961027e-06,
+      "loss": 3.8392,
+      "step": 2859520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1730001316855822e-06,
+      "loss": 3.8476,
+      "step": 2860032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.164630562978794e-06,
+      "loss": 3.8403,
+      "step": 2860544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1562446154682734e-06,
+      "loss": 3.8478,
+      "step": 2861056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.147858667957753e-06,
+      "loss": 3.8432,
+      "step": 2861568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1394727204472333e-06,
+      "loss": 3.8467,
+      "step": 2862080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1311031517404445e-06,
+      "loss": 3.8341,
+      "step": 2862592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1227172042299245e-06,
+      "loss": 3.8395,
+      "step": 2863104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1143312567194044e-06,
+      "loss": 3.837,
+      "step": 2863616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1059780668163473e-06,
+      "loss": 3.8298,
+      "step": 2864128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0975921193058273e-06,
+      "loss": 3.8406,
+      "step": 2864640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0892061717953072e-06,
+      "loss": 3.8302,
+      "step": 2865152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0808202242847868e-06,
+      "loss": 3.8477,
+      "step": 2865664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0724342767742667e-06,
+      "loss": 3.8513,
+      "step": 2866176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0640483292637467e-06,
+      "loss": 3.8236,
+      "step": 2866688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0556623817532266e-06,
+      "loss": 3.8443,
+      "step": 2867200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0472764342427066e-06,
+      "loss": 3.8364,
+      "step": 2867712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.038890486732186e-06,
+      "loss": 3.8394,
+      "step": 2868224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.030504539221666e-06,
+      "loss": 3.8372,
+      "step": 2868736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.022118591711146e-06,
+      "loss": 3.8293,
+      "step": 2869248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.013732644200626e-06,
+      "loss": 3.8385,
+      "step": 2869760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.005346696690106e-06,
+      "loss": 3.8334,
+      "step": 2870272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.996993506787049e-06,
+      "loss": 3.8359,
+      "step": 2870784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.988607559276529e-06,
+      "loss": 3.8326,
+      "step": 2871296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9802216117660087e-06,
+      "loss": 3.8408,
+      "step": 2871808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9718356642554883e-06,
+      "loss": 3.8423,
+      "step": 2872320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9634660955487e-06,
+      "loss": 3.8409,
+      "step": 2872832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9550801480381795e-06,
+      "loss": 3.8268,
+      "step": 2873344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9466942005276594e-06,
+      "loss": 3.8384,
+      "step": 2873856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9383082530171394e-06,
+      "loss": 3.8364,
+      "step": 2874368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9299223055066197e-06,
+      "loss": 3.8476,
+      "step": 2874880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9215363579960997e-06,
+      "loss": 3.8296,
+      "step": 2875392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9131504104855792e-06,
+      "loss": 3.8358,
+      "step": 2875904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.904764462975059e-06,
+      "loss": 3.833,
+      "step": 2876416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8963948942682704e-06,
+      "loss": 3.844,
+      "step": 2876928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.888025325561482e-06,
+      "loss": 3.8274,
+      "step": 2877440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8796393780509616e-06,
+      "loss": 3.8334,
+      "step": 2877952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8712534305404415e-06,
+      "loss": 3.8364,
+      "step": 2878464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8628674830299215e-06,
+      "loss": 3.8358,
+      "step": 2878976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8544815355194014e-06,
+      "loss": 3.8256,
+      "step": 2879488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8460955880088814e-06,
+      "loss": 3.8329,
+      "step": 2880000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8377096404983613e-06,
+      "loss": 3.8285,
+      "step": 2880512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8293236929878413e-06,
+      "loss": 3.8388,
+      "step": 2881024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8209377454773213e-06,
+      "loss": 3.8372,
+      "step": 2881536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.812551797966801e-06,
+      "loss": 3.827,
+      "step": 2882048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8041822292600124e-06,
+      "loss": 3.8424,
+      "step": 2882560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7958126605532236e-06,
+      "loss": 3.8534,
+      "step": 2883072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7874267130427036e-06,
+      "loss": 3.8463,
+      "step": 2883584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7790407655321836e-06,
+      "loss": 3.8259,
+      "step": 2884096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7706548180216635e-06,
+      "loss": 3.8352,
+      "step": 2884608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7622852493148747e-06,
+      "loss": 3.8379,
+      "step": 2885120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7538993018043547e-06,
+      "loss": 3.8325,
+      "step": 2885632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7455133542938346e-06,
+      "loss": 3.8393,
+      "step": 2886144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7371274067833146e-06,
+      "loss": 3.8277,
+      "step": 2886656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7287414592727945e-06,
+      "loss": 3.8416,
+      "step": 2887168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7203555117622745e-06,
+      "loss": 3.8368,
+      "step": 2887680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.711969564251754e-06,
+      "loss": 3.8407,
+      "step": 2888192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7035999955449657e-06,
+      "loss": 3.8316,
+      "step": 2888704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.695214048034445e-06,
+      "loss": 3.8331,
+      "step": 2889216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.686828100523925e-06,
+      "loss": 3.8388,
+      "step": 2889728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.678442153013405e-06,
+      "loss": 3.8442,
+      "step": 2890240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.670056205502885e-06,
+      "loss": 3.8365,
+      "step": 2890752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.661670257992365e-06,
+      "loss": 3.8406,
+      "step": 2891264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6533006892855762e-06,
+      "loss": 3.8316,
+      "step": 2891776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.644914741775056e-06,
+      "loss": 3.8454,
+      "step": 2892288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.636528794264536e-06,
+      "loss": 3.8394,
+      "step": 2892800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.628142846754016e-06,
+      "loss": 3.844,
+      "step": 2893312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.619756899243496e-06,
+      "loss": 3.8321,
+      "step": 2893824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.611370951732976e-06,
+      "loss": 3.8455,
+      "step": 2894336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.602985004222456e-06,
+      "loss": 3.8354,
+      "step": 2894848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5945990567119355e-06,
+      "loss": 3.8573,
+      "step": 2895360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.586229488005147e-06,
+      "loss": 3.8309,
+      "step": 2895872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5778599192983584e-06,
+      "loss": 3.8451,
+      "step": 2896384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5694739717878383e-06,
+      "loss": 3.8376,
+      "step": 2896896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.561088024277318e-06,
+      "loss": 3.835,
+      "step": 2897408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.552702076766798e-06,
+      "loss": 3.8416,
+      "step": 2897920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5443161292562778e-06,
+      "loss": 3.8408,
+      "step": 2898432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5359465605494894e-06,
+      "loss": 3.842,
+      "step": 2898944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5275606130389693e-06,
+      "loss": 3.8341,
+      "step": 2899456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5191746655284493e-06,
+      "loss": 3.8427,
+      "step": 2899968
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9728095531463623,
+      "eval_runtime": 292.4462,
+      "eval_samples_per_second": 1304.825,
+      "eval_steps_per_second": 40.777,
+      "step": 2900160
+    }
+  ],
+  "logging_steps": 512,
+  "max_steps": 3052726,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 10,
+  "total_flos": 1.1898925406261944e+18,
+  "trial_name": null,
+  "trial_params": null
+}