diff --git "a/checkpoint-2594880/trainer_state.json" "b/checkpoint-2594880/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2594880/trainer_state.json"
@@ -0,0 +1,30705 @@
+{
+  "best_metric": 3.8496768474578857,
+  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/passive/transformer/1/checkpoints/checkpoint-915840",
+  "epoch": 0.025000606015738065,
+  "eval_steps": 10,
+  "global_step": 2594880,
+  "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": 10.9323,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999161405248948e-05,
+      "loss": 6.829,
+      "step": 512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998322810497896e-05,
+      "loss": 6.1955,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997484215746844e-05,
+      "loss": 5.9876,
+      "step": 1536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996645620995792e-05,
+      "loss": 5.8238,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99580702624474e-05,
+      "loss": 5.7039,
+      "step": 2560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994968431493688e-05,
+      "loss": 5.6145,
+      "step": 3072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994129836742636e-05,
+      "loss": 5.5373,
+      "step": 3584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993291241991584e-05,
+      "loss": 5.4552,
+      "step": 4096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992452647240532e-05,
+      "loss": 5.4069,
+      "step": 4608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99161405248948e-05,
+      "loss": 5.3539,
+      "step": 5120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.990775457738428e-05,
+      "loss": 5.3134,
+      "step": 5632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989936862987376e-05,
+      "loss": 5.2658,
+      "step": 6144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989099906116697e-05,
+      "loss": 5.2176,
+      "step": 6656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.988261311365645e-05,
+      "loss": 5.178,
+      "step": 7168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.987422716614593e-05,
+      "loss": 5.1336,
+      "step": 7680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986584121863541e-05,
+      "loss": 5.115,
+      "step": 8192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.985745527112489e-05,
+      "loss": 5.0825,
+      "step": 8704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984906932361437e-05,
+      "loss": 5.0539,
+      "step": 9216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984068337610385e-05,
+      "loss": 5.0235,
+      "step": 9728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.983229742859333e-05,
+      "loss": 5.015,
+      "step": 10240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9823927859886547e-05,
+      "loss": 4.9821,
+      "step": 10752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9815541912376026e-05,
+      "loss": 4.9593,
+      "step": 11264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9807155964865506e-05,
+      "loss": 4.9296,
+      "step": 11776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9798770017354986e-05,
+      "loss": 4.9278,
+      "step": 12288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9790400448648195e-05,
+      "loss": 4.8883,
+      "step": 12800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9782014501137675e-05,
+      "loss": 4.8781,
+      "step": 13312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9773628553627155e-05,
+      "loss": 4.8482,
+      "step": 13824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9765242606116635e-05,
+      "loss": 4.8377,
+      "step": 14336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9756856658606115e-05,
+      "loss": 4.8239,
+      "step": 14848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9748470711095595e-05,
+      "loss": 4.8065,
+      "step": 15360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9740117521192533e-05,
+      "loss": 4.7973,
+      "step": 15872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9731731573682013e-05,
+      "loss": 4.784,
+      "step": 16384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97233456261715e-05,
+      "loss": 4.7687,
+      "step": 16896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.971495967866098e-05,
+      "loss": 4.7659,
+      "step": 17408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.970657373115046e-05,
+      "loss": 4.746,
+      "step": 17920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.969820416244367e-05,
+      "loss": 4.7399,
+      "step": 18432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.968981821493315e-05,
+      "loss": 4.6986,
+      "step": 18944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.968143226742263e-05,
+      "loss": 4.7107,
+      "step": 19456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.967304631991211e-05,
+      "loss": 4.6746,
+      "step": 19968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.966466037240159e-05,
+      "loss": 4.6858,
+      "step": 20480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965627442489107e-05,
+      "loss": 4.6678,
+      "step": 20992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964788847738054e-05,
+      "loss": 4.6666,
+      "step": 21504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963950252987002e-05,
+      "loss": 4.6418,
+      "step": 22016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963113296116324e-05,
+      "loss": 4.653,
+      "step": 22528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962274701365272e-05,
+      "loss": 4.6397,
+      "step": 23040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96143610661422e-05,
+      "loss": 4.6337,
+      "step": 23552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960597511863168e-05,
+      "loss": 4.6287,
+      "step": 24064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9597605549924894e-05,
+      "loss": 4.5955,
+      "step": 24576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9589219602414374e-05,
+      "loss": 4.5991,
+      "step": 25088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958083365490385e-05,
+      "loss": 4.6011,
+      "step": 25600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.957244770739333e-05,
+      "loss": 4.5891,
+      "step": 26112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.956407813868654e-05,
+      "loss": 4.5802,
+      "step": 26624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9555692191176016e-05,
+      "loss": 4.5608,
+      "step": 27136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9547306243665496e-05,
+      "loss": 4.5591,
+      "step": 27648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9538920296154976e-05,
+      "loss": 4.5413,
+      "step": 28160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9530534348644456e-05,
+      "loss": 4.5632,
+      "step": 28672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9522148401133936e-05,
+      "loss": 4.5192,
+      "step": 29184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951377883242715e-05,
+      "loss": 4.5418,
+      "step": 29696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.950539288491663e-05,
+      "loss": 4.5339,
+      "step": 30208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.949700693740611e-05,
+      "loss": 4.5138,
+      "step": 30720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948862098989559e-05,
+      "loss": 4.5131,
+      "step": 31232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.94802514211888e-05,
+      "loss": 4.504,
+      "step": 31744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947186547367828e-05,
+      "loss": 4.4957,
+      "step": 32256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.946347952616776e-05,
+      "loss": 4.4877,
+      "step": 32768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.945509357865724e-05,
+      "loss": 4.4991,
+      "step": 33280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.944672400995045e-05,
+      "loss": 4.4766,
+      "step": 33792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.943833806243993e-05,
+      "loss": 4.4741,
+      "step": 34304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.942995211492941e-05,
+      "loss": 4.4664,
+      "step": 34816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9421582546222625e-05,
+      "loss": 4.4654,
+      "step": 35328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9413196598712105e-05,
+      "loss": 4.4755,
+      "step": 35840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9404810651201585e-05,
+      "loss": 4.4707,
+      "step": 36352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9396424703691065e-05,
+      "loss": 4.4465,
+      "step": 36864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9388038756180545e-05,
+      "loss": 4.4613,
+      "step": 37376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9379652808670025e-05,
+      "loss": 4.4609,
+      "step": 37888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9371266861159505e-05,
+      "loss": 4.4338,
+      "step": 38400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9362897292452714e-05,
+      "loss": 4.4412,
+      "step": 38912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9354511344942194e-05,
+      "loss": 4.4307,
+      "step": 39424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9346125397431674e-05,
+      "loss": 4.4251,
+      "step": 39936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9337739449921154e-05,
+      "loss": 4.4148,
+      "step": 40448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9329353502410634e-05,
+      "loss": 4.4198,
+      "step": 40960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9320967554900114e-05,
+      "loss": 4.4241,
+      "step": 41472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9312581607389594e-05,
+      "loss": 4.4273,
+      "step": 41984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.930421203868281e-05,
+      "loss": 4.4024,
+      "step": 42496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.929582609117229e-05,
+      "loss": 4.3811,
+      "step": 43008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.928744014366177e-05,
+      "loss": 4.3965,
+      "step": 43520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.927905419615125e-05,
+      "loss": 4.3952,
+      "step": 44032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.927066824864073e-05,
+      "loss": 4.3948,
+      "step": 44544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.926229867993394e-05,
+      "loss": 4.3824,
+      "step": 45056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.925392911122715e-05,
+      "loss": 4.3843,
+      "step": 45568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.924554316371663e-05,
+      "loss": 4.3804,
+      "step": 46080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.923715721620611e-05,
+      "loss": 4.377,
+      "step": 46592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922877126869559e-05,
+      "loss": 4.3679,
+      "step": 47104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922038532118507e-05,
+      "loss": 4.3705,
+      "step": 47616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921201575247828e-05,
+      "loss": 4.3589,
+      "step": 48128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.920362980496776e-05,
+      "loss": 4.3724,
+      "step": 48640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919524385745724e-05,
+      "loss": 4.3578,
+      "step": 49152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918685790994672e-05,
+      "loss": 4.3395,
+      "step": 49664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91784719624362e-05,
+      "loss": 4.3522,
+      "step": 50176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9170086014925676e-05,
+      "loss": 4.3529,
+      "step": 50688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9161700067415156e-05,
+      "loss": 4.356,
+      "step": 51200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9153314119904636e-05,
+      "loss": 4.3399,
+      "step": 51712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914494455119785e-05,
+      "loss": 4.3391,
+      "step": 52224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9136558603687325e-05,
+      "loss": 4.3242,
+      "step": 52736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9128172656176805e-05,
+      "loss": 4.3343,
+      "step": 53248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9119786708666285e-05,
+      "loss": 4.3156,
+      "step": 53760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911143351876323e-05,
+      "loss": 4.3236,
+      "step": 54272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.910304757125272e-05,
+      "loss": 4.321,
+      "step": 54784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90946616237422e-05,
+      "loss": 4.3199,
+      "step": 55296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.908627567623168e-05,
+      "loss": 4.3109,
+      "step": 55808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.907788972872115e-05,
+      "loss": 4.3137,
+      "step": 56320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906950378121063e-05,
+      "loss": 4.3049,
+      "step": 56832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906111783370011e-05,
+      "loss": 4.3124,
+      "step": 57344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9052748264993326e-05,
+      "loss": 4.3029,
+      "step": 57856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90443623174828e-05,
+      "loss": 4.2991,
+      "step": 58368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.903597636997228e-05,
+      "loss": 4.3147,
+      "step": 58880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.902759042246176e-05,
+      "loss": 4.3027,
+      "step": 59392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9019220853754975e-05,
+      "loss": 4.2995,
+      "step": 59904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9010834906244455e-05,
+      "loss": 4.2886,
+      "step": 60416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9002448958733935e-05,
+      "loss": 4.2941,
+      "step": 60928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8994063011223415e-05,
+      "loss": 4.2865,
+      "step": 61440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8985677063712895e-05,
+      "loss": 4.2855,
+      "step": 61952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8977291116202375e-05,
+      "loss": 4.2943,
+      "step": 62464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8968921547495584e-05,
+      "loss": 4.2783,
+      "step": 62976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8960535599985064e-05,
+      "loss": 4.2839,
+      "step": 63488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8952149652474544e-05,
+      "loss": 4.2651,
+      "step": 64000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8943763704964024e-05,
+      "loss": 4.2693,
+      "step": 64512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8935377757453504e-05,
+      "loss": 4.2748,
+      "step": 65024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8926991809942984e-05,
+      "loss": 4.2663,
+      "step": 65536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891862224123619e-05,
+      "loss": 4.2585,
+      "step": 66048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891023629372567e-05,
+      "loss": 4.2845,
+      "step": 66560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.890185034621515e-05,
+      "loss": 4.2653,
+      "step": 67072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.889346439870464e-05,
+      "loss": 4.2586,
+      "step": 67584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.888507845119412e-05,
+      "loss": 4.2576,
+      "step": 68096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.887670888248733e-05,
+      "loss": 4.2578,
+      "step": 68608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.886832293497681e-05,
+      "loss": 4.2585,
+      "step": 69120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.885993698746629e-05,
+      "loss": 4.2567,
+      "step": 69632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.885155103995577e-05,
+      "loss": 4.2509,
+      "step": 70144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.884318147124898e-05,
+      "loss": 4.2584,
+      "step": 70656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.883479552373846e-05,
+      "loss": 4.2557,
+      "step": 71168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.882640957622794e-05,
+      "loss": 4.2495,
+      "step": 71680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.881802362871742e-05,
+      "loss": 4.2407,
+      "step": 72192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88096376812069e-05,
+      "loss": 4.246,
+      "step": 72704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.880125173369638e-05,
+      "loss": 4.2374,
+      "step": 73216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879286578618586e-05,
+      "loss": 4.2389,
+      "step": 73728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.878447983867534e-05,
+      "loss": 4.2352,
+      "step": 74240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.877611026996855e-05,
+      "loss": 4.2332,
+      "step": 74752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.876772432245803e-05,
+      "loss": 4.239,
+      "step": 75264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875933837494751e-05,
+      "loss": 4.2262,
+      "step": 75776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8750952427436986e-05,
+      "loss": 4.2233,
+      "step": 76288
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.200882911682129,
+      "eval_runtime": 302.3694,
+      "eval_samples_per_second": 1262.003,
+      "eval_steps_per_second": 39.439,
+      "step": 76320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.87425828587302e-05,
+      "loss": 4.2084,
+      "step": 76800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.873419691121968e-05,
+      "loss": 4.21,
+      "step": 77312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.872582734251289e-05,
+      "loss": 4.2365,
+      "step": 77824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.871744139500237e-05,
+      "loss": 4.2209,
+      "step": 78336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.870905544749185e-05,
+      "loss": 4.2144,
+      "step": 78848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.870066949998133e-05,
+      "loss": 4.2116,
+      "step": 79360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.869228355247081e-05,
+      "loss": 4.2083,
+      "step": 79872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.868389760496029e-05,
+      "loss": 4.1963,
+      "step": 80384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8675528036253507e-05,
+      "loss": 4.2087,
+      "step": 80896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8667142088742986e-05,
+      "loss": 4.2032,
+      "step": 81408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.865875614123246e-05,
+      "loss": 4.206,
+      "step": 81920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.865037019372194e-05,
+      "loss": 4.2077,
+      "step": 82432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8642000625015155e-05,
+      "loss": 4.1922,
+      "step": 82944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8633614677504635e-05,
+      "loss": 4.1882,
+      "step": 83456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.862522872999411e-05,
+      "loss": 4.1812,
+      "step": 83968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.861684278248359e-05,
+      "loss": 4.1794,
+      "step": 84480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.860845683497307e-05,
+      "loss": 4.1899,
+      "step": 84992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.860007088746255e-05,
+      "loss": 4.1797,
+      "step": 85504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.859168493995203e-05,
+      "loss": 4.1838,
+      "step": 86016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.858329899244151e-05,
+      "loss": 4.2022,
+      "step": 86528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8574913044930995e-05,
+      "loss": 4.1809,
+      "step": 87040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8566543476224204e-05,
+      "loss": 4.1847,
+      "step": 87552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8558157528713684e-05,
+      "loss": 4.172,
+      "step": 88064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8549771581203164e-05,
+      "loss": 4.1883,
+      "step": 88576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8541385633692644e-05,
+      "loss": 4.1681,
+      "step": 89088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.853301606498585e-05,
+      "loss": 4.1691,
+      "step": 89600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.852463011747533e-05,
+      "loss": 4.1632,
+      "step": 90112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.851624416996481e-05,
+      "loss": 4.1672,
+      "step": 90624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.850785822245429e-05,
+      "loss": 4.1594,
+      "step": 91136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.849947227494377e-05,
+      "loss": 4.1576,
+      "step": 91648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.849110270623698e-05,
+      "loss": 4.1613,
+      "step": 92160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.848271675872646e-05,
+      "loss": 4.1675,
+      "step": 92672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.847433081121595e-05,
+      "loss": 4.1589,
+      "step": 93184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.846594486370543e-05,
+      "loss": 4.1619,
+      "step": 93696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.845755891619491e-05,
+      "loss": 4.1548,
+      "step": 94208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.844918934748812e-05,
+      "loss": 4.1643,
+      "step": 94720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.84408033999776e-05,
+      "loss": 4.1372,
+      "step": 95232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.843241745246708e-05,
+      "loss": 4.1529,
+      "step": 95744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.842403150495656e-05,
+      "loss": 4.1308,
+      "step": 96256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.841566193624977e-05,
+      "loss": 4.1516,
+      "step": 96768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.840727598873925e-05,
+      "loss": 4.1364,
+      "step": 97280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.839889004122873e-05,
+      "loss": 4.1497,
+      "step": 97792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.839050409371821e-05,
+      "loss": 4.1317,
+      "step": 98304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8382134525011416e-05,
+      "loss": 4.1458,
+      "step": 98816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.83737485775009e-05,
+      "loss": 4.144,
+      "step": 99328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.836536262999038e-05,
+      "loss": 4.1423,
+      "step": 99840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.835697668247986e-05,
+      "loss": 4.1457,
+      "step": 100352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834860711377307e-05,
+      "loss": 4.1179,
+      "step": 100864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834022116626255e-05,
+      "loss": 4.1238,
+      "step": 101376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.833183521875203e-05,
+      "loss": 4.1366,
+      "step": 101888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832344927124151e-05,
+      "loss": 4.1342,
+      "step": 102400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831507970253472e-05,
+      "loss": 4.1264,
+      "step": 102912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.83066937550242e-05,
+      "loss": 4.1215,
+      "step": 103424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.829830780751368e-05,
+      "loss": 4.1168,
+      "step": 103936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.828992186000316e-05,
+      "loss": 4.1084,
+      "step": 104448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.828155229129637e-05,
+      "loss": 4.1234,
+      "step": 104960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8273166343785856e-05,
+      "loss": 4.105,
+      "step": 105472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8264780396275336e-05,
+      "loss": 4.1149,
+      "step": 105984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8256394448764816e-05,
+      "loss": 4.1243,
+      "step": 106496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8248024880058025e-05,
+      "loss": 4.1045,
+      "step": 107008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8239638932547505e-05,
+      "loss": 4.1043,
+      "step": 107520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8231252985036985e-05,
+      "loss": 4.1042,
+      "step": 108032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8222867037526465e-05,
+      "loss": 4.0964,
+      "step": 108544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8214497468819674e-05,
+      "loss": 4.0984,
+      "step": 109056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8206111521309154e-05,
+      "loss": 4.1156,
+      "step": 109568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8197725573798634e-05,
+      "loss": 4.0981,
+      "step": 110080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8189339626288114e-05,
+      "loss": 4.0959,
+      "step": 110592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.818097005758132e-05,
+      "loss": 4.0935,
+      "step": 111104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.817258411007081e-05,
+      "loss": 4.0933,
+      "step": 111616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.816419816256029e-05,
+      "loss": 4.1088,
+      "step": 112128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.815581221504977e-05,
+      "loss": 4.1027,
+      "step": 112640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.814742626753924e-05,
+      "loss": 4.0923,
+      "step": 113152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.813905669883246e-05,
+      "loss": 4.104,
+      "step": 113664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.813067075132194e-05,
+      "loss": 4.1052,
+      "step": 114176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.812228480381141e-05,
+      "loss": 4.0876,
+      "step": 114688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.811391523510463e-05,
+      "loss": 4.0936,
+      "step": 115200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.810552928759411e-05,
+      "loss": 4.0909,
+      "step": 115712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.809714334008359e-05,
+      "loss": 4.0888,
+      "step": 116224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.808875739257306e-05,
+      "loss": 4.0813,
+      "step": 116736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.808037144506255e-05,
+      "loss": 4.0844,
+      "step": 117248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.807200187635576e-05,
+      "loss": 4.0962,
+      "step": 117760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8063615928845243e-05,
+      "loss": 4.0938,
+      "step": 118272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8055229981334717e-05,
+      "loss": 4.0827,
+      "step": 118784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.804686041262793e-05,
+      "loss": 4.0602,
+      "step": 119296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.803847446511741e-05,
+      "loss": 4.0764,
+      "step": 119808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8030088517606886e-05,
+      "loss": 4.0819,
+      "step": 120320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8021702570096366e-05,
+      "loss": 4.0806,
+      "step": 120832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.801333300138958e-05,
+      "loss": 4.0763,
+      "step": 121344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.800494705387906e-05,
+      "loss": 4.0722,
+      "step": 121856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.799657748517227e-05,
+      "loss": 4.0791,
+      "step": 122368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.798819153766175e-05,
+      "loss": 4.0707,
+      "step": 122880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.797980559015123e-05,
+      "loss": 4.0663,
+      "step": 123392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.797141964264071e-05,
+      "loss": 4.079,
+      "step": 123904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.796303369513019e-05,
+      "loss": 4.0618,
+      "step": 124416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7954664126423406e-05,
+      "loss": 4.0825,
+      "step": 124928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7946278178912886e-05,
+      "loss": 4.0719,
+      "step": 125440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.793789223140236e-05,
+      "loss": 4.0517,
+      "step": 125952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.792950628389184e-05,
+      "loss": 4.0694,
+      "step": 126464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.792112033638132e-05,
+      "loss": 4.0647,
+      "step": 126976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.79127343888708e-05,
+      "loss": 4.075,
+      "step": 127488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.790434844136028e-05,
+      "loss": 4.0654,
+      "step": 128000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.789596249384976e-05,
+      "loss": 4.0621,
+      "step": 128512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.788759292514297e-05,
+      "loss": 4.0499,
+      "step": 129024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.787920697763245e-05,
+      "loss": 4.0635,
+      "step": 129536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7870821030121935e-05,
+      "loss": 4.0452,
+      "step": 130048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7862451461415144e-05,
+      "loss": 4.0532,
+      "step": 130560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7854065513904624e-05,
+      "loss": 4.0588,
+      "step": 131072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7845679566394104e-05,
+      "loss": 4.0558,
+      "step": 131584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7837293618883584e-05,
+      "loss": 4.0446,
+      "step": 132096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7828907671373064e-05,
+      "loss": 4.0554,
+      "step": 132608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7820521723862544e-05,
+      "loss": 4.049,
+      "step": 133120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7812135776352024e-05,
+      "loss": 4.0541,
+      "step": 133632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7803749828841504e-05,
+      "loss": 4.0507,
+      "step": 134144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.779538026013471e-05,
+      "loss": 4.0429,
+      "step": 134656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.778699431262419e-05,
+      "loss": 4.0674,
+      "step": 135168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.777860836511367e-05,
+      "loss": 4.0555,
+      "step": 135680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.777022241760315e-05,
+      "loss": 4.0525,
+      "step": 136192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.776185284889637e-05,
+      "loss": 4.0391,
+      "step": 136704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.775346690138585e-05,
+      "loss": 4.0466,
+      "step": 137216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.774508095387533e-05,
+      "loss": 4.0452,
+      "step": 137728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.773669500636481e-05,
+      "loss": 4.0392,
+      "step": 138240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.772832543765802e-05,
+      "loss": 4.0547,
+      "step": 138752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.77199394901475e-05,
+      "loss": 4.0425,
+      "step": 139264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.771155354263698e-05,
+      "loss": 4.0413,
+      "step": 139776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.770316759512646e-05,
+      "loss": 4.035,
+      "step": 140288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7694798026419666e-05,
+      "loss": 4.0369,
+      "step": 140800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7686412078909146e-05,
+      "loss": 4.038,
+      "step": 141312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7678026131398626e-05,
+      "loss": 4.037,
+      "step": 141824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.766965656269184e-05,
+      "loss": 4.0266,
+      "step": 142336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.766127061518132e-05,
+      "loss": 4.0544,
+      "step": 142848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.76528846676708e-05,
+      "loss": 4.0437,
+      "step": 143360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.764449872016028e-05,
+      "loss": 4.0336,
+      "step": 143872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.763611277264976e-05,
+      "loss": 4.0305,
+      "step": 144384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.762772682513924e-05,
+      "loss": 4.0299,
+      "step": 144896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.761934087762872e-05,
+      "loss": 4.0397,
+      "step": 145408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7610954930118195e-05,
+      "loss": 4.0356,
+      "step": 145920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.760258536141141e-05,
+      "loss": 4.0309,
+      "step": 146432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.759419941390089e-05,
+      "loss": 4.0415,
+      "step": 146944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.758581346639037e-05,
+      "loss": 4.0358,
+      "step": 147456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7577427518879844e-05,
+      "loss": 4.0378,
+      "step": 147968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.756905795017306e-05,
+      "loss": 4.0258,
+      "step": 148480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.756067200266254e-05,
+      "loss": 4.0328,
+      "step": 148992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.755228605515202e-05,
+      "loss": 4.028,
+      "step": 149504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.75439001076415e-05,
+      "loss": 4.0261,
+      "step": 150016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7535530538934716e-05,
+      "loss": 4.0267,
+      "step": 150528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7527144591424196e-05,
+      "loss": 4.0249,
+      "step": 151040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.751875864391367e-05,
+      "loss": 4.0292,
+      "step": 151552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.751037269640315e-05,
+      "loss": 4.0176,
+      "step": 152064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7502003127696365e-05,
+      "loss": 4.0186,
+      "step": 152576
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.032747268676758,
+      "eval_runtime": 302.674,
+      "eval_samples_per_second": 1260.733,
+      "eval_steps_per_second": 39.399,
+      "step": 152640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7493617180185845e-05,
+      "loss": 4.0048,
+      "step": 153088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.748523123267532e-05,
+      "loss": 4.007,
+      "step": 153600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.74768452851648e-05,
+      "loss": 4.0337,
+      "step": 154112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.746845933765428e-05,
+      "loss": 4.0255,
+      "step": 154624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7460073390143764e-05,
+      "loss": 4.0172,
+      "step": 155136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7451687442633244e-05,
+      "loss": 4.0119,
+      "step": 155648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7443301495122724e-05,
+      "loss": 4.015,
+      "step": 156160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7434915547612204e-05,
+      "loss": 4.0007,
+      "step": 156672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7426529600101684e-05,
+      "loss": 4.0185,
+      "step": 157184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7418143652591164e-05,
+      "loss": 4.0119,
+      "step": 157696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7409757705080644e-05,
+      "loss": 4.0115,
+      "step": 158208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.740138813637385e-05,
+      "loss": 4.0153,
+      "step": 158720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.739303494647079e-05,
+      "loss": 4.002,
+      "step": 159232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.738464899896027e-05,
+      "loss": 4.0025,
+      "step": 159744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.737626305144975e-05,
+      "loss": 3.9976,
+      "step": 160256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.736787710393923e-05,
+      "loss": 3.9954,
+      "step": 160768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.735949115642871e-05,
+      "loss": 3.9998,
+      "step": 161280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.73511052089182e-05,
+      "loss": 3.994,
+      "step": 161792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.734271926140768e-05,
+      "loss": 4.0013,
+      "step": 162304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.733433331389716e-05,
+      "loss": 4.0239,
+      "step": 162816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.732594736638664e-05,
+      "loss": 3.9984,
+      "step": 163328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.731756141887612e-05,
+      "loss": 4.0038,
+      "step": 163840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.73091754713656e-05,
+      "loss": 3.9984,
+      "step": 164352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.730080590265881e-05,
+      "loss": 4.0074,
+      "step": 164864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.729241995514829e-05,
+      "loss": 3.9885,
+      "step": 165376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.728403400763777e-05,
+      "loss": 3.9984,
+      "step": 165888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.727564806012725e-05,
+      "loss": 3.9898,
+      "step": 166400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7267278491420456e-05,
+      "loss": 3.9903,
+      "step": 166912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7258892543909936e-05,
+      "loss": 3.9847,
+      "step": 167424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7250506596399416e-05,
+      "loss": 3.985,
+      "step": 167936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7242120648888896e-05,
+      "loss": 3.9899,
+      "step": 168448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.723375108018211e-05,
+      "loss": 4.0005,
+      "step": 168960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.722536513267159e-05,
+      "loss": 3.9893,
+      "step": 169472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.721697918516107e-05,
+      "loss": 3.9968,
+      "step": 169984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.720859323765055e-05,
+      "loss": 3.9822,
+      "step": 170496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.720020729014003e-05,
+      "loss": 4.0003,
+      "step": 171008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7191821342629505e-05,
+      "loss": 3.97,
+      "step": 171520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7183435395118985e-05,
+      "loss": 3.9902,
+      "step": 172032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.71750658264122e-05,
+      "loss": 3.9657,
+      "step": 172544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.716667987890168e-05,
+      "loss": 3.9867,
+      "step": 173056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7158293931391154e-05,
+      "loss": 3.9753,
+      "step": 173568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7149907983880634e-05,
+      "loss": 3.9899,
+      "step": 174080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.714153841517385e-05,
+      "loss": 3.9752,
+      "step": 174592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.713315246766333e-05,
+      "loss": 3.9808,
+      "step": 175104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.712476652015281e-05,
+      "loss": 3.9836,
+      "step": 175616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.711638057264229e-05,
+      "loss": 3.9864,
+      "step": 176128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7108011003935505e-05,
+      "loss": 3.9837,
+      "step": 176640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.709962505642498e-05,
+      "loss": 3.9659,
+      "step": 177152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.709123910891446e-05,
+      "loss": 3.9652,
+      "step": 177664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.708285316140394e-05,
+      "loss": 3.9821,
+      "step": 178176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7074483592697154e-05,
+      "loss": 3.98,
+      "step": 178688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.706611402399036e-05,
+      "loss": 3.9719,
+      "step": 179200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705772807647984e-05,
+      "loss": 3.9691,
+      "step": 179712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.704934212896932e-05,
+      "loss": 3.9675,
+      "step": 180224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.70409561814588e-05,
+      "loss": 3.9536,
+      "step": 180736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.703257023394828e-05,
+      "loss": 3.9688,
+      "step": 181248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.702418428643776e-05,
+      "loss": 3.9559,
+      "step": 181760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.701579833892724e-05,
+      "loss": 3.9672,
+      "step": 182272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.700742877022045e-05,
+      "loss": 3.9763,
+      "step": 182784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.699904282270993e-05,
+      "loss": 3.9564,
+      "step": 183296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.699065687519941e-05,
+      "loss": 3.9572,
+      "step": 183808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.698227092768889e-05,
+      "loss": 3.9613,
+      "step": 184320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.69739013589821e-05,
+      "loss": 3.9492,
+      "step": 184832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.696551541147158e-05,
+      "loss": 3.9492,
+      "step": 185344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.695712946396106e-05,
+      "loss": 3.97,
+      "step": 185856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.694874351645054e-05,
+      "loss": 3.9539,
+      "step": 186368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.694037394774376e-05,
+      "loss": 3.9525,
+      "step": 186880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.693198800023324e-05,
+      "loss": 3.9506,
+      "step": 187392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.692361843152645e-05,
+      "loss": 3.9528,
+      "step": 187904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6915232484015926e-05,
+      "loss": 3.9649,
+      "step": 188416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6906846536505406e-05,
+      "loss": 3.9618,
+      "step": 188928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6898460588994886e-05,
+      "loss": 3.9499,
+      "step": 189440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6890074641484366e-05,
+      "loss": 3.9648,
+      "step": 189952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6881688693973846e-05,
+      "loss": 3.961,
+      "step": 190464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6873302746463326e-05,
+      "loss": 3.9504,
+      "step": 190976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6864916798952806e-05,
+      "loss": 3.9549,
+      "step": 191488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6856547230246015e-05,
+      "loss": 3.9563,
+      "step": 192000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6848161282735495e-05,
+      "loss": 3.9511,
+      "step": 192512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.683979171402871e-05,
+      "loss": 3.948,
+      "step": 193024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.683140576651819e-05,
+      "loss": 3.9472,
+      "step": 193536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.682301981900767e-05,
+      "loss": 3.96,
+      "step": 194048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.681463387149715e-05,
+      "loss": 3.9609,
+      "step": 194560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.680624792398663e-05,
+      "loss": 3.9449,
+      "step": 195072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.679786197647611e-05,
+      "loss": 3.9335,
+      "step": 195584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.678947602896559e-05,
+      "loss": 3.9396,
+      "step": 196096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.678109008145507e-05,
+      "loss": 3.9483,
+      "step": 196608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.677272051274828e-05,
+      "loss": 3.9506,
+      "step": 197120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.676435094404149e-05,
+      "loss": 3.9422,
+      "step": 197632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.675596499653097e-05,
+      "loss": 3.9392,
+      "step": 198144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.674757904902045e-05,
+      "loss": 3.9536,
+      "step": 198656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6739193101509935e-05,
+      "loss": 3.9364,
+      "step": 199168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6730807153999415e-05,
+      "loss": 3.939,
+      "step": 199680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6722421206488895e-05,
+      "loss": 3.9469,
+      "step": 200192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6714051637782104e-05,
+      "loss": 3.9354,
+      "step": 200704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6705665690271584e-05,
+      "loss": 3.9547,
+      "step": 201216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6697279742761064e-05,
+      "loss": 3.9413,
+      "step": 201728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6688893795250544e-05,
+      "loss": 3.9252,
+      "step": 202240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6680507847740024e-05,
+      "loss": 3.942,
+      "step": 202752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6672121900229504e-05,
+      "loss": 3.9358,
+      "step": 203264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6663735952718984e-05,
+      "loss": 3.952,
+      "step": 203776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.665535000520846e-05,
+      "loss": 3.942,
+      "step": 204288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.66469968153054e-05,
+      "loss": 3.9341,
+      "step": 204800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.663861086779488e-05,
+      "loss": 3.922,
+      "step": 205312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.663022492028437e-05,
+      "loss": 3.94,
+      "step": 205824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.662183897277385e-05,
+      "loss": 3.9226,
+      "step": 206336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.661345302526333e-05,
+      "loss": 3.9328,
+      "step": 206848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.660508345655654e-05,
+      "loss": 3.9358,
+      "step": 207360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.659669750904602e-05,
+      "loss": 3.9322,
+      "step": 207872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.65883115615355e-05,
+      "loss": 3.9218,
+      "step": 208384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657992561402498e-05,
+      "loss": 3.9407,
+      "step": 208896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657153966651446e-05,
+      "loss": 3.9249,
+      "step": 209408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.656315371900393e-05,
+      "loss": 3.9292,
+      "step": 209920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6554784150297146e-05,
+      "loss": 3.9343,
+      "step": 210432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6546398202786626e-05,
+      "loss": 3.9238,
+      "step": 210944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6538012255276106e-05,
+      "loss": 3.9455,
+      "step": 211456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6529626307765586e-05,
+      "loss": 3.934,
+      "step": 211968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.65212567390588e-05,
+      "loss": 3.9351,
+      "step": 212480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.651287079154828e-05,
+      "loss": 3.9219,
+      "step": 212992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.650448484403776e-05,
+      "loss": 3.9289,
+      "step": 213504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6496098896527235e-05,
+      "loss": 3.9306,
+      "step": 214016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6487712949016715e-05,
+      "loss": 3.9174,
+      "step": 214528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6479327001506195e-05,
+      "loss": 3.9337,
+      "step": 215040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6470941053995675e-05,
+      "loss": 3.9266,
+      "step": 215552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6462555106485155e-05,
+      "loss": 3.9249,
+      "step": 216064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6454185537778364e-05,
+      "loss": 3.9247,
+      "step": 216576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6445799590267844e-05,
+      "loss": 3.9238,
+      "step": 217088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6437413642757324e-05,
+      "loss": 3.9162,
+      "step": 217600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6429027695246804e-05,
+      "loss": 3.9262,
+      "step": 218112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642065812654002e-05,
+      "loss": 3.9156,
+      "step": 218624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.64122721790295e-05,
+      "loss": 3.9364,
+      "step": 219136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.640388623151898e-05,
+      "loss": 3.9335,
+      "step": 219648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.639550028400846e-05,
+      "loss": 3.9185,
+      "step": 220160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.638711433649794e-05,
+      "loss": 3.9168,
+      "step": 220672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.637872838898742e-05,
+      "loss": 3.918,
+      "step": 221184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.63703424414769e-05,
+      "loss": 3.9255,
+      "step": 221696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.636195649396638e-05,
+      "loss": 3.9251,
+      "step": 222208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.635360330406332e-05,
+      "loss": 3.9197,
+      "step": 222720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.63452173565528e-05,
+      "loss": 3.9264,
+      "step": 223232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.633683140904228e-05,
+      "loss": 3.9247,
+      "step": 223744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.632844546153176e-05,
+      "loss": 3.9284,
+      "step": 224256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6320059514021244e-05,
+      "loss": 3.9127,
+      "step": 224768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6311689945314454e-05,
+      "loss": 3.9265,
+      "step": 225280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6303303997803934e-05,
+      "loss": 3.9211,
+      "step": 225792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6294918050293413e-05,
+      "loss": 3.9114,
+      "step": 226304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6286532102782893e-05,
+      "loss": 3.9212,
+      "step": 226816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6278146155272373e-05,
+      "loss": 3.9196,
+      "step": 227328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626976020776185e-05,
+      "loss": 3.9171,
+      "step": 227840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626137426025133e-05,
+      "loss": 3.9079,
+      "step": 228352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.625300469154454e-05,
+      "loss": 3.9093,
+      "step": 228864
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.957947015762329,
+      "eval_runtime": 310.3378,
+      "eval_samples_per_second": 1229.599,
+      "eval_steps_per_second": 38.426,
+      "step": 228960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.624461874403402e-05,
+      "loss": 3.901,
+      "step": 229376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.62362327965235e-05,
+      "loss": 3.8992,
+      "step": 229888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.622784684901298e-05,
+      "loss": 3.9255,
+      "step": 230400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.621946090150246e-05,
+      "loss": 3.9197,
+      "step": 230912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.621107495399194e-05,
+      "loss": 3.9194,
+      "step": 231424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.620268900648142e-05,
+      "loss": 3.8999,
+      "step": 231936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.61943030589709e-05,
+      "loss": 3.9147,
+      "step": 232448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.618591711146038e-05,
+      "loss": 3.8927,
+      "step": 232960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.617753116394986e-05,
+      "loss": 3.9172,
+      "step": 233472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.616914521643934e-05,
+      "loss": 3.9098,
+      "step": 233984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.616077564773255e-05,
+      "loss": 3.9031,
+      "step": 234496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.615238970022203e-05,
+      "loss": 3.9158,
+      "step": 235008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.614402013151524e-05,
+      "loss": 3.8985,
+      "step": 235520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.613563418400472e-05,
+      "loss": 3.9035,
+      "step": 236032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.61272482364942e-05,
+      "loss": 3.892,
+      "step": 236544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.611886228898368e-05,
+      "loss": 3.8996,
+      "step": 237056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.611047634147316e-05,
+      "loss": 3.8963,
+      "step": 237568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.610209039396265e-05,
+      "loss": 3.8959,
+      "step": 238080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.609370444645213e-05,
+      "loss": 3.8991,
+      "step": 238592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6085318498941607e-05,
+      "loss": 3.9207,
+      "step": 239104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6076932551431087e-05,
+      "loss": 3.8986,
+      "step": 239616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6068546603920566e-05,
+      "loss": 3.9052,
+      "step": 240128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6060160656410046e-05,
+      "loss": 3.9,
+      "step": 240640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6051791087703256e-05,
+      "loss": 3.9076,
+      "step": 241152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6043405140192735e-05,
+      "loss": 3.8913,
+      "step": 241664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6035019192682215e-05,
+      "loss": 3.8992,
+      "step": 242176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6026633245171695e-05,
+      "loss": 3.8914,
+      "step": 242688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6018263676464904e-05,
+      "loss": 3.8905,
+      "step": 243200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6009877728954384e-05,
+      "loss": 3.8858,
+      "step": 243712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6001491781443864e-05,
+      "loss": 3.8938,
+      "step": 244224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5993105833933344e-05,
+      "loss": 3.8913,
+      "step": 244736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.598473626522656e-05,
+      "loss": 3.9036,
+      "step": 245248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.597636669651977e-05,
+      "loss": 3.8932,
+      "step": 245760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.596798074900925e-05,
+      "loss": 3.8972,
+      "step": 246272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.595959480149873e-05,
+      "loss": 3.8882,
+      "step": 246784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.595120885398821e-05,
+      "loss": 3.9021,
+      "step": 247296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.594282290647769e-05,
+      "loss": 3.8821,
+      "step": 247808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.593443695896717e-05,
+      "loss": 3.8923,
+      "step": 248320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.592606739026038e-05,
+      "loss": 3.8715,
+      "step": 248832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.591768144274986e-05,
+      "loss": 3.8905,
+      "step": 249344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.590929549523934e-05,
+      "loss": 3.8848,
+      "step": 249856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.590090954772882e-05,
+      "loss": 3.8975,
+      "step": 250368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.58925236002183e-05,
+      "loss": 3.8854,
+      "step": 250880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5884137652707785e-05,
+      "loss": 3.8876,
+      "step": 251392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.587575170519726e-05,
+      "loss": 3.8905,
+      "step": 251904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5867382136490474e-05,
+      "loss": 3.8962,
+      "step": 252416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5858996188979954e-05,
+      "loss": 3.886,
+      "step": 252928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.585061024146943e-05,
+      "loss": 3.8771,
+      "step": 253440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.584222429395891e-05,
+      "loss": 3.8686,
+      "step": 253952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.583383834644839e-05,
+      "loss": 3.8916,
+      "step": 254464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.58254687777416e-05,
+      "loss": 3.8904,
+      "step": 254976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5817082830231076e-05,
+      "loss": 3.882,
+      "step": 255488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5808696882720556e-05,
+      "loss": 3.8755,
+      "step": 256000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5800310935210036e-05,
+      "loss": 3.8802,
+      "step": 256512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.579192498769952e-05,
+      "loss": 3.8612,
+      "step": 257024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5783539040189e-05,
+      "loss": 3.8842,
+      "step": 257536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.577516947148221e-05,
+      "loss": 3.8649,
+      "step": 258048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.576678352397169e-05,
+      "loss": 3.8776,
+      "step": 258560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.575839757646117e-05,
+      "loss": 3.8841,
+      "step": 259072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.575001162895065e-05,
+      "loss": 3.8748,
+      "step": 259584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.574162568144013e-05,
+      "loss": 3.8652,
+      "step": 260096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.573323973392961e-05,
+      "loss": 3.874,
+      "step": 260608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.572485378641909e-05,
+      "loss": 3.8614,
+      "step": 261120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.571646783890857e-05,
+      "loss": 3.8658,
+      "step": 261632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.570809827020178e-05,
+      "loss": 3.8799,
+      "step": 262144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.569971232269126e-05,
+      "loss": 3.869,
+      "step": 262656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.569132637518074e-05,
+      "loss": 3.8647,
+      "step": 263168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5682956806473956e-05,
+      "loss": 3.8605,
+      "step": 263680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5674570858963436e-05,
+      "loss": 3.8672,
+      "step": 264192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5666184911452916e-05,
+      "loss": 3.8721,
+      "step": 264704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5657798963942396e-05,
+      "loss": 3.8751,
+      "step": 265216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5649413016431876e-05,
+      "loss": 3.8682,
+      "step": 265728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5641027068921356e-05,
+      "loss": 3.8781,
+      "step": 266240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5632641121410836e-05,
+      "loss": 3.8752,
+      "step": 266752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5624271552704045e-05,
+      "loss": 3.8649,
+      "step": 267264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5615885605193525e-05,
+      "loss": 3.8673,
+      "step": 267776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5607499657683005e-05,
+      "loss": 3.8748,
+      "step": 268288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5599113710172485e-05,
+      "loss": 3.87,
+      "step": 268800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5590744141465694e-05,
+      "loss": 3.8634,
+      "step": 269312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5582358193955174e-05,
+      "loss": 3.8641,
+      "step": 269824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5573972246444654e-05,
+      "loss": 3.8711,
+      "step": 270336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.556558629893414e-05,
+      "loss": 3.8742,
+      "step": 270848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5557200351423614e-05,
+      "loss": 3.8654,
+      "step": 271360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5548814403913094e-05,
+      "loss": 3.8515,
+      "step": 271872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5540428456402574e-05,
+      "loss": 3.8513,
+      "step": 272384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5532042508892054e-05,
+      "loss": 3.8625,
+      "step": 272896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.552367294018526e-05,
+      "loss": 3.8699,
+      "step": 273408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.551530337147848e-05,
+      "loss": 3.8647,
+      "step": 273920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.550691742396796e-05,
+      "loss": 3.8515,
+      "step": 274432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.549853147645744e-05,
+      "loss": 3.8701,
+      "step": 274944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.549014552894691e-05,
+      "loss": 3.8593,
+      "step": 275456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.548175958143639e-05,
+      "loss": 3.8571,
+      "step": 275968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.547339001272961e-05,
+      "loss": 3.8603,
+      "step": 276480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.546500406521909e-05,
+      "loss": 3.8629,
+      "step": 276992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.545661811770857e-05,
+      "loss": 3.8741,
+      "step": 277504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.544823217019805e-05,
+      "loss": 3.8549,
+      "step": 278016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.543984622268753e-05,
+      "loss": 3.8441,
+      "step": 278528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5431476653980736e-05,
+      "loss": 3.8629,
+      "step": 279040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5423090706470216e-05,
+      "loss": 3.8548,
+      "step": 279552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5414704758959696e-05,
+      "loss": 3.8692,
+      "step": 280064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5406318811449176e-05,
+      "loss": 3.8642,
+      "step": 280576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5397949242742385e-05,
+      "loss": 3.8521,
+      "step": 281088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5389563295231865e-05,
+      "loss": 3.8445,
+      "step": 281600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5381177347721345e-05,
+      "loss": 3.8613,
+      "step": 282112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.537279140021083e-05,
+      "loss": 3.8438,
+      "step": 282624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.536440545270031e-05,
+      "loss": 3.8539,
+      "step": 283136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.535603588399352e-05,
+      "loss": 3.8522,
+      "step": 283648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5347649936483e-05,
+      "loss": 3.8579,
+      "step": 284160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.533926398897248e-05,
+      "loss": 3.8423,
+      "step": 284672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.533087804146196e-05,
+      "loss": 3.8602,
+      "step": 285184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.532249209395144e-05,
+      "loss": 3.8458,
+      "step": 285696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.531410614644092e-05,
+      "loss": 3.851,
+      "step": 286208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.530573657773413e-05,
+      "loss": 3.8592,
+      "step": 286720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.529735063022361e-05,
+      "loss": 3.8461,
+      "step": 287232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.528896468271309e-05,
+      "loss": 3.8673,
+      "step": 287744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.528057873520257e-05,
+      "loss": 3.858,
+      "step": 288256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5272209166495786e-05,
+      "loss": 3.8567,
+      "step": 288768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5263823218985266e-05,
+      "loss": 3.8453,
+      "step": 289280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5255437271474746e-05,
+      "loss": 3.8511,
+      "step": 289792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5247051323964226e-05,
+      "loss": 3.8491,
+      "step": 290304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5238665376453706e-05,
+      "loss": 3.8445,
+      "step": 290816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5230295807746915e-05,
+      "loss": 3.8537,
+      "step": 291328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5221909860236395e-05,
+      "loss": 3.8503,
+      "step": 291840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5213523912725875e-05,
+      "loss": 3.8497,
+      "step": 292352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5205137965215355e-05,
+      "loss": 3.8462,
+      "step": 292864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5196752017704834e-05,
+      "loss": 3.8505,
+      "step": 293376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5188366070194314e-05,
+      "loss": 3.8419,
+      "step": 293888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5179996501487524e-05,
+      "loss": 3.8526,
+      "step": 294400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5171610553977003e-05,
+      "loss": 3.8398,
+      "step": 294912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5163224606466483e-05,
+      "loss": 3.8594,
+      "step": 295424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.515483865895597e-05,
+      "loss": 3.8528,
+      "step": 295936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.514645271144545e-05,
+      "loss": 3.8506,
+      "step": 296448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.513808314273866e-05,
+      "loss": 3.8393,
+      "step": 296960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.512969719522814e-05,
+      "loss": 3.8447,
+      "step": 297472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.512131124771762e-05,
+      "loss": 3.8507,
+      "step": 297984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.51129253002071e-05,
+      "loss": 3.8519,
+      "step": 298496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.510453935269657e-05,
+      "loss": 3.8416,
+      "step": 299008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.509616978398979e-05,
+      "loss": 3.8505,
+      "step": 299520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.508778383647927e-05,
+      "loss": 3.8478,
+      "step": 300032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.507939788896874e-05,
+      "loss": 3.8572,
+      "step": 300544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.507101194145822e-05,
+      "loss": 3.8426,
+      "step": 301056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.506264237275144e-05,
+      "loss": 3.8533,
+      "step": 301568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5054256425240924e-05,
+      "loss": 3.8509,
+      "step": 302080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.50458704777304e-05,
+      "loss": 3.8401,
+      "step": 302592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.503748453021988e-05,
+      "loss": 3.8448,
+      "step": 303104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.502909858270936e-05,
+      "loss": 3.8525,
+      "step": 303616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.502072901400257e-05,
+      "loss": 3.8438,
+      "step": 304128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5012343066492046e-05,
+      "loss": 3.8307,
+      "step": 304640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5003957118981526e-05,
+      "loss": 3.8464,
+      "step": 305152
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.917541742324829,
+      "eval_runtime": 304.6162,
+      "eval_samples_per_second": 1252.694,
+      "eval_steps_per_second": 39.148,
+      "step": 305280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4995571171471006e-05,
+      "loss": 3.833,
+      "step": 305664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4987185223960486e-05,
+      "loss": 3.8275,
+      "step": 306176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4978799276449966e-05,
+      "loss": 3.8502,
+      "step": 306688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4970429707743175e-05,
+      "loss": 3.8502,
+      "step": 307200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.496204376023266e-05,
+      "loss": 3.8486,
+      "step": 307712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.495365781272214e-05,
+      "loss": 3.8358,
+      "step": 308224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.494527186521162e-05,
+      "loss": 3.8419,
+      "step": 308736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.49368859177011e-05,
+      "loss": 3.8207,
+      "step": 309248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.492849997019058e-05,
+      "loss": 3.8464,
+      "step": 309760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.492011402268006e-05,
+      "loss": 3.8413,
+      "step": 310272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.491174445397327e-05,
+      "loss": 3.833,
+      "step": 310784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.490335850646275e-05,
+      "loss": 3.8456,
+      "step": 311296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.489498893775596e-05,
+      "loss": 3.8287,
+      "step": 311808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.488660299024544e-05,
+      "loss": 3.8342,
+      "step": 312320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.487821704273492e-05,
+      "loss": 3.8255,
+      "step": 312832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.48698310952244e-05,
+      "loss": 3.8262,
+      "step": 313344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.486144514771388e-05,
+      "loss": 3.8294,
+      "step": 313856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.485305920020336e-05,
+      "loss": 3.8263,
+      "step": 314368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.484467325269284e-05,
+      "loss": 3.8303,
+      "step": 314880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4836287305182326e-05,
+      "loss": 3.8474,
+      "step": 315392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4827901357671806e-05,
+      "loss": 3.8327,
+      "step": 315904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4819515410161286e-05,
+      "loss": 3.8386,
+      "step": 316416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4811145841454495e-05,
+      "loss": 3.8306,
+      "step": 316928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4802759893943975e-05,
+      "loss": 3.8368,
+      "step": 317440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4794373946433455e-05,
+      "loss": 3.8243,
+      "step": 317952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4785987998922935e-05,
+      "loss": 3.831,
+      "step": 318464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.477760205141241e-05,
+      "loss": 3.8286,
+      "step": 318976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4769232482705624e-05,
+      "loss": 3.8227,
+      "step": 319488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4760846535195104e-05,
+      "loss": 3.8174,
+      "step": 320000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.475246058768458e-05,
+      "loss": 3.8269,
+      "step": 320512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4744074640174064e-05,
+      "loss": 3.8273,
+      "step": 321024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.473570507146728e-05,
+      "loss": 3.8372,
+      "step": 321536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.472731912395676e-05,
+      "loss": 3.827,
+      "step": 322048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.471893317644623e-05,
+      "loss": 3.8326,
+      "step": 322560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.471054722893571e-05,
+      "loss": 3.8199,
+      "step": 323072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.470217766022893e-05,
+      "loss": 3.8361,
+      "step": 323584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.469379171271841e-05,
+      "loss": 3.8169,
+      "step": 324096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.468540576520788e-05,
+      "loss": 3.8244,
+      "step": 324608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.467701981769736e-05,
+      "loss": 3.8077,
+      "step": 325120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.466865024899058e-05,
+      "loss": 3.8231,
+      "step": 325632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.466026430148005e-05,
+      "loss": 3.8201,
+      "step": 326144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.465187835396953e-05,
+      "loss": 3.8304,
+      "step": 326656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.464349240645902e-05,
+      "loss": 3.8182,
+      "step": 327168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.46351064589485e-05,
+      "loss": 3.8249,
+      "step": 327680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.462672051143798e-05,
+      "loss": 3.8244,
+      "step": 328192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4618350942731186e-05,
+      "loss": 3.8287,
+      "step": 328704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4609964995220666e-05,
+      "loss": 3.825,
+      "step": 329216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4601579047710146e-05,
+      "loss": 3.8127,
+      "step": 329728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4593193100199626e-05,
+      "loss": 3.8019,
+      "step": 330240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4584807152689106e-05,
+      "loss": 3.8294,
+      "step": 330752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4576437583982315e-05,
+      "loss": 3.8268,
+      "step": 331264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4568051636471795e-05,
+      "loss": 3.8171,
+      "step": 331776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4559665688961275e-05,
+      "loss": 3.8152,
+      "step": 332288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4551279741450755e-05,
+      "loss": 3.8167,
+      "step": 332800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4542893793940235e-05,
+      "loss": 3.7976,
+      "step": 333312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.453452422523345e-05,
+      "loss": 3.8138,
+      "step": 333824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.452613827772293e-05,
+      "loss": 3.8058,
+      "step": 334336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.451775233021241e-05,
+      "loss": 3.8124,
+      "step": 334848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.450936638270189e-05,
+      "loss": 3.8233,
+      "step": 335360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.450098043519137e-05,
+      "loss": 3.8124,
+      "step": 335872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.449259448768085e-05,
+      "loss": 3.8006,
+      "step": 336384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.448420854017033e-05,
+      "loss": 3.8159,
+      "step": 336896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.447583897146354e-05,
+      "loss": 3.7971,
+      "step": 337408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.446745302395302e-05,
+      "loss": 3.8057,
+      "step": 337920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.44590670764425e-05,
+      "loss": 3.816,
+      "step": 338432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.445068112893198e-05,
+      "loss": 3.8094,
+      "step": 338944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.444231156022519e-05,
+      "loss": 3.8021,
+      "step": 339456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.443392561271467e-05,
+      "loss": 3.8025,
+      "step": 339968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4425539665204156e-05,
+      "loss": 3.8018,
+      "step": 340480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4417153717693636e-05,
+      "loss": 3.8113,
+      "step": 340992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4408784148986845e-05,
+      "loss": 3.8115,
+      "step": 341504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4400398201476325e-05,
+      "loss": 3.8075,
+      "step": 342016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4392012253965805e-05,
+      "loss": 3.8163,
+      "step": 342528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4383626306455285e-05,
+      "loss": 3.8156,
+      "step": 343040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4375240358944764e-05,
+      "loss": 3.8045,
+      "step": 343552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4366870790237974e-05,
+      "loss": 3.8004,
+      "step": 344064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4358484842727454e-05,
+      "loss": 3.8193,
+      "step": 344576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4350098895216933e-05,
+      "loss": 3.8091,
+      "step": 345088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4341712947706413e-05,
+      "loss": 3.7989,
+      "step": 345600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.433332700019589e-05,
+      "loss": 3.8044,
+      "step": 346112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.43249574314891e-05,
+      "loss": 3.8113,
+      "step": 346624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.431657148397859e-05,
+      "loss": 3.818,
+      "step": 347136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.430818553646807e-05,
+      "loss": 3.8004,
+      "step": 347648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.429979958895754e-05,
+      "loss": 3.7904,
+      "step": 348160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.429141364144702e-05,
+      "loss": 3.798,
+      "step": 348672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.428304407274024e-05,
+      "loss": 3.7974,
+      "step": 349184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.427465812522971e-05,
+      "loss": 3.8079,
+      "step": 349696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.426627217771919e-05,
+      "loss": 3.8049,
+      "step": 350208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.425788623020867e-05,
+      "loss": 3.7922,
+      "step": 350720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.424951666150189e-05,
+      "loss": 3.8093,
+      "step": 351232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.424113071399136e-05,
+      "loss": 3.7992,
+      "step": 351744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.423274476648084e-05,
+      "loss": 3.8011,
+      "step": 352256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.422435881897033e-05,
+      "loss": 3.7979,
+      "step": 352768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.421598925026354e-05,
+      "loss": 3.8016,
+      "step": 353280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4207603302753016e-05,
+      "loss": 3.8152,
+      "step": 353792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4199217355242496e-05,
+      "loss": 3.8024,
+      "step": 354304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4190831407731976e-05,
+      "loss": 3.7825,
+      "step": 354816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4182461839025185e-05,
+      "loss": 3.803,
+      "step": 355328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4174075891514665e-05,
+      "loss": 3.7975,
+      "step": 355840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4165689944004145e-05,
+      "loss": 3.8054,
+      "step": 356352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4157303996493625e-05,
+      "loss": 3.8087,
+      "step": 356864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4148934427786834e-05,
+      "loss": 3.7959,
+      "step": 357376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4140548480276314e-05,
+      "loss": 3.7873,
+      "step": 357888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4132162532765794e-05,
+      "loss": 3.8008,
+      "step": 358400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.412377658525528e-05,
+      "loss": 3.7826,
+      "step": 358912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.411539063774476e-05,
+      "loss": 3.7983,
+      "step": 359424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.410702106903797e-05,
+      "loss": 3.7948,
+      "step": 359936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.409863512152745e-05,
+      "loss": 3.8019,
+      "step": 360448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.409024917401693e-05,
+      "loss": 3.7846,
+      "step": 360960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.408186322650641e-05,
+      "loss": 3.8022,
+      "step": 361472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.407347727899589e-05,
+      "loss": 3.7879,
+      "step": 361984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.40651077102891e-05,
+      "loss": 3.7923,
+      "step": 362496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.405672176277858e-05,
+      "loss": 3.8028,
+      "step": 363008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.404833581526806e-05,
+      "loss": 3.7939,
+      "step": 363520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.403994986775754e-05,
+      "loss": 3.8087,
+      "step": 364032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.403158029905075e-05,
+      "loss": 3.8007,
+      "step": 364544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4023194351540234e-05,
+      "loss": 3.7956,
+      "step": 365056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4014808404029714e-05,
+      "loss": 3.7969,
+      "step": 365568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4006422456519194e-05,
+      "loss": 3.7899,
+      "step": 366080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.39980528878124e-05,
+      "loss": 3.7976,
+      "step": 366592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.398966694030188e-05,
+      "loss": 3.788,
+      "step": 367104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.398128099279136e-05,
+      "loss": 3.801,
+      "step": 367616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.397289504528084e-05,
+      "loss": 3.7901,
+      "step": 368128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.396452547657405e-05,
+      "loss": 3.7902,
+      "step": 368640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.395613952906353e-05,
+      "loss": 3.7899,
+      "step": 369152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.394775358155301e-05,
+      "loss": 3.7974,
+      "step": 369664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.393936763404249e-05,
+      "loss": 3.7865,
+      "step": 370176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.39309980653357e-05,
+      "loss": 3.7963,
+      "step": 370688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.392261211782519e-05,
+      "loss": 3.7867,
+      "step": 371200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.391422617031467e-05,
+      "loss": 3.8017,
+      "step": 371712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.390584022280415e-05,
+      "loss": 3.7964,
+      "step": 372224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.389745427529363e-05,
+      "loss": 3.796,
+      "step": 372736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.388908470658684e-05,
+      "loss": 3.7862,
+      "step": 373248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.388069875907632e-05,
+      "loss": 3.7872,
+      "step": 373760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.38723128115658e-05,
+      "loss": 3.801,
+      "step": 374272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.386392686405528e-05,
+      "loss": 3.793,
+      "step": 374784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3855557295348486e-05,
+      "loss": 3.7857,
+      "step": 375296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3847171347837966e-05,
+      "loss": 3.7995,
+      "step": 375808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3838785400327446e-05,
+      "loss": 3.7934,
+      "step": 376320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3830399452816926e-05,
+      "loss": 3.8055,
+      "step": 376832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.382202988411014e-05,
+      "loss": 3.7895,
+      "step": 377344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.381364393659962e-05,
+      "loss": 3.7964,
+      "step": 377856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.38052579890891e-05,
+      "loss": 3.7949,
+      "step": 378368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.379687204157858e-05,
+      "loss": 3.7854,
+      "step": 378880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.378848609406806e-05,
+      "loss": 3.7925,
+      "step": 379392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.378011652536127e-05,
+      "loss": 3.7935,
+      "step": 379904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.377173057785075e-05,
+      "loss": 3.7934,
+      "step": 380416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.376334463034023e-05,
+      "loss": 3.7779,
+      "step": 380928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.375495868282971e-05,
+      "loss": 3.7922,
+      "step": 381440
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.892491102218628,
+      "eval_runtime": 304.2151,
+      "eval_samples_per_second": 1254.346,
+      "eval_steps_per_second": 39.199,
+      "step": 381600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.374657273531919e-05,
+      "loss": 3.7787,
+      "step": 381952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.373818678780867e-05,
+      "loss": 3.773,
+      "step": 382464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3729800840298144e-05,
+      "loss": 3.7954,
+      "step": 382976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3721414892787624e-05,
+      "loss": 3.7981,
+      "step": 383488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.371302894527711e-05,
+      "loss": 3.7928,
+      "step": 384000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.370464299776659e-05,
+      "loss": 3.7835,
+      "step": 384512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.369625705025607e-05,
+      "loss": 3.7863,
+      "step": 385024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.368787110274555e-05,
+      "loss": 3.77,
+      "step": 385536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.367950153403876e-05,
+      "loss": 3.7909,
+      "step": 386048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.367111558652824e-05,
+      "loss": 3.7904,
+      "step": 386560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.366272963901772e-05,
+      "loss": 3.7787,
+      "step": 387072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.36543436915072e-05,
+      "loss": 3.7903,
+      "step": 387584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3645990501604144e-05,
+      "loss": 3.7779,
+      "step": 388096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.363760455409362e-05,
+      "loss": 3.7824,
+      "step": 388608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.36292186065831e-05,
+      "loss": 3.7731,
+      "step": 389120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.362083265907258e-05,
+      "loss": 3.7749,
+      "step": 389632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3612446711562064e-05,
+      "loss": 3.7761,
+      "step": 390144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3604060764051544e-05,
+      "loss": 3.7767,
+      "step": 390656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3595674816541024e-05,
+      "loss": 3.7792,
+      "step": 391168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3587288869030504e-05,
+      "loss": 3.7929,
+      "step": 391680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.357891930032371e-05,
+      "loss": 3.7863,
+      "step": 392192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.357053335281319e-05,
+      "loss": 3.7862,
+      "step": 392704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.356214740530267e-05,
+      "loss": 3.7789,
+      "step": 393216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.355376145779215e-05,
+      "loss": 3.7856,
+      "step": 393728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.354537551028163e-05,
+      "loss": 3.7709,
+      "step": 394240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.353700594157484e-05,
+      "loss": 3.7846,
+      "step": 394752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.352861999406432e-05,
+      "loss": 3.7748,
+      "step": 395264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.35202340465538e-05,
+      "loss": 3.7704,
+      "step": 395776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.351184809904328e-05,
+      "loss": 3.7699,
+      "step": 396288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.35034785303365e-05,
+      "loss": 3.7726,
+      "step": 396800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.349509258282598e-05,
+      "loss": 3.779,
+      "step": 397312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.348670663531546e-05,
+      "loss": 3.7822,
+      "step": 397824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347832068780494e-05,
+      "loss": 3.7811,
+      "step": 398336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3469951119098146e-05,
+      "loss": 3.7822,
+      "step": 398848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3461565171587626e-05,
+      "loss": 3.7681,
+      "step": 399360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3453179224077106e-05,
+      "loss": 3.7833,
+      "step": 399872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3444793276566586e-05,
+      "loss": 3.7701,
+      "step": 400384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3436407329056066e-05,
+      "loss": 3.7682,
+      "step": 400896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3428037760349275e-05,
+      "loss": 3.7643,
+      "step": 401408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3419651812838755e-05,
+      "loss": 3.7716,
+      "step": 401920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3411265865328235e-05,
+      "loss": 3.7738,
+      "step": 402432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3402879917817715e-05,
+      "loss": 3.7796,
+      "step": 402944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.339451034911093e-05,
+      "loss": 3.7699,
+      "step": 403456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.338612440160041e-05,
+      "loss": 3.7708,
+      "step": 403968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.337773845408989e-05,
+      "loss": 3.7782,
+      "step": 404480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.336935250657937e-05,
+      "loss": 3.7768,
+      "step": 404992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.336098293787258e-05,
+      "loss": 3.7768,
+      "step": 405504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.335259699036206e-05,
+      "loss": 3.7637,
+      "step": 406016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.334421104285154e-05,
+      "loss": 3.7483,
+      "step": 406528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.333582509534102e-05,
+      "loss": 3.7816,
+      "step": 407040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.332745552663423e-05,
+      "loss": 3.7776,
+      "step": 407552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.331906957912371e-05,
+      "loss": 3.77,
+      "step": 408064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.331068363161319e-05,
+      "loss": 3.7685,
+      "step": 408576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.330229768410267e-05,
+      "loss": 3.7628,
+      "step": 409088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3293928115395885e-05,
+      "loss": 3.7518,
+      "step": 409600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3285542167885365e-05,
+      "loss": 3.7633,
+      "step": 410112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3277156220374845e-05,
+      "loss": 3.7558,
+      "step": 410624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3268770272864325e-05,
+      "loss": 3.7639,
+      "step": 411136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3260384325353805e-05,
+      "loss": 3.7764,
+      "step": 411648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3252014756647014e-05,
+      "loss": 3.7616,
+      "step": 412160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3243628809136494e-05,
+      "loss": 3.7526,
+      "step": 412672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3235242861625974e-05,
+      "loss": 3.77,
+      "step": 413184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.322685691411545e-05,
+      "loss": 3.7496,
+      "step": 413696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.321848734540866e-05,
+      "loss": 3.7547,
+      "step": 414208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.321010139789814e-05,
+      "loss": 3.7706,
+      "step": 414720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.320171545038762e-05,
+      "loss": 3.7632,
+      "step": 415232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.31933295028771e-05,
+      "loss": 3.7545,
+      "step": 415744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.318495993417032e-05,
+      "loss": 3.7535,
+      "step": 416256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.31765739866598e-05,
+      "loss": 3.754,
+      "step": 416768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.316818803914928e-05,
+      "loss": 3.7613,
+      "step": 417280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.315980209163875e-05,
+      "loss": 3.7679,
+      "step": 417792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.315143252293197e-05,
+      "loss": 3.7563,
+      "step": 418304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.314304657542145e-05,
+      "loss": 3.7683,
+      "step": 418816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.313466062791092e-05,
+      "loss": 3.7673,
+      "step": 419328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3126291059204136e-05,
+      "loss": 3.7663,
+      "step": 419840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3117905111693616e-05,
+      "loss": 3.7492,
+      "step": 420352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3109519164183096e-05,
+      "loss": 3.77,
+      "step": 420864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3101133216672576e-05,
+      "loss": 3.7647,
+      "step": 421376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.309276364796579e-05,
+      "loss": 3.7491,
+      "step": 421888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.308437770045527e-05,
+      "loss": 3.7628,
+      "step": 422400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.307599175294475e-05,
+      "loss": 3.7602,
+      "step": 422912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3067605805434225e-05,
+      "loss": 3.7705,
+      "step": 423424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3059219857923705e-05,
+      "loss": 3.7548,
+      "step": 423936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.305085028921692e-05,
+      "loss": 3.7415,
+      "step": 424448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3042464341706394e-05,
+      "loss": 3.7525,
+      "step": 424960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3034078394195874e-05,
+      "loss": 3.7534,
+      "step": 425472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3025692446685354e-05,
+      "loss": 3.7597,
+      "step": 425984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.301732287797857e-05,
+      "loss": 3.757,
+      "step": 426496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.300893693046805e-05,
+      "loss": 3.7489,
+      "step": 427008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.300055098295753e-05,
+      "loss": 3.7575,
+      "step": 427520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.299216503544701e-05,
+      "loss": 3.7562,
+      "step": 428032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2983795466740226e-05,
+      "loss": 3.7492,
+      "step": 428544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.29754095192297e-05,
+      "loss": 3.7562,
+      "step": 429056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.296702357171918e-05,
+      "loss": 3.7603,
+      "step": 429568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.295863762420866e-05,
+      "loss": 3.7688,
+      "step": 430080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.295026805550187e-05,
+      "loss": 3.7538,
+      "step": 430592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.294188210799135e-05,
+      "loss": 3.7397,
+      "step": 431104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.293349616048083e-05,
+      "loss": 3.756,
+      "step": 431616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.292511021297031e-05,
+      "loss": 3.751,
+      "step": 432128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.291674064426352e-05,
+      "loss": 3.7618,
+      "step": 432640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2908354696753004e-05,
+      "loss": 3.7645,
+      "step": 433152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2899968749242484e-05,
+      "loss": 3.7479,
+      "step": 433664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2891582801731964e-05,
+      "loss": 3.7443,
+      "step": 434176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.288321323302517e-05,
+      "loss": 3.7572,
+      "step": 434688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.287482728551465e-05,
+      "loss": 3.7369,
+      "step": 435200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.286644133800413e-05,
+      "loss": 3.756,
+      "step": 435712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.285805539049361e-05,
+      "loss": 3.7432,
+      "step": 436224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.284968582178682e-05,
+      "loss": 3.759,
+      "step": 436736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.28412998742763e-05,
+      "loss": 3.7405,
+      "step": 437248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.283291392676578e-05,
+      "loss": 3.7547,
+      "step": 437760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.282452797925526e-05,
+      "loss": 3.7485,
+      "step": 438272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.281615841054847e-05,
+      "loss": 3.7443,
+      "step": 438784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.280777246303796e-05,
+      "loss": 3.7587,
+      "step": 439296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.279938651552744e-05,
+      "loss": 3.7471,
+      "step": 439808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.279100056801692e-05,
+      "loss": 3.7644,
+      "step": 440320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2782630999310126e-05,
+      "loss": 3.7603,
+      "step": 440832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2774245051799606e-05,
+      "loss": 3.7499,
+      "step": 441344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2765859104289086e-05,
+      "loss": 3.7555,
+      "step": 441856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2757473156778566e-05,
+      "loss": 3.7458,
+      "step": 442368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2749103588071775e-05,
+      "loss": 3.7495,
+      "step": 442880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2740717640561255e-05,
+      "loss": 3.7476,
+      "step": 443392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2732331693050735e-05,
+      "loss": 3.754,
+      "step": 443904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2723945745540215e-05,
+      "loss": 3.7474,
+      "step": 444416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2715576176833424e-05,
+      "loss": 3.746,
+      "step": 444928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2707190229322904e-05,
+      "loss": 3.7486,
+      "step": 445440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.269880428181239e-05,
+      "loss": 3.7509,
+      "step": 445952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.269041833430187e-05,
+      "loss": 3.7421,
+      "step": 446464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.268204876559508e-05,
+      "loss": 3.7503,
+      "step": 446976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.267366281808456e-05,
+      "loss": 3.7401,
+      "step": 447488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.266527687057404e-05,
+      "loss": 3.7645,
+      "step": 448000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.265689092306352e-05,
+      "loss": 3.7452,
+      "step": 448512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.264852135435673e-05,
+      "loss": 3.7566,
+      "step": 449024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.264013540684621e-05,
+      "loss": 3.742,
+      "step": 449536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.263174945933569e-05,
+      "loss": 3.7465,
+      "step": 450048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.262336351182517e-05,
+      "loss": 3.7558,
+      "step": 450560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.261499394311838e-05,
+      "loss": 3.7489,
+      "step": 451072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.260660799560786e-05,
+      "loss": 3.7439,
+      "step": 451584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2598222048097345e-05,
+      "loss": 3.7576,
+      "step": 452096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2589836100586825e-05,
+      "loss": 3.7497,
+      "step": 452608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2581466531880034e-05,
+      "loss": 3.7607,
+      "step": 453120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2573080584369514e-05,
+      "loss": 3.7438,
+      "step": 453632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2564694636858994e-05,
+      "loss": 3.7528,
+      "step": 454144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2556308689348474e-05,
+      "loss": 3.7508,
+      "step": 454656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.254793912064168e-05,
+      "loss": 3.7434,
+      "step": 455168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.253955317313116e-05,
+      "loss": 3.7516,
+      "step": 455680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.253116722562064e-05,
+      "loss": 3.7512,
+      "step": 456192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.252278127811012e-05,
+      "loss": 3.7548,
+      "step": 456704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.251441170940333e-05,
+      "loss": 3.7268,
+      "step": 457216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.250602576189281e-05,
+      "loss": 3.7538,
+      "step": 457728
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.87673020362854,
+      "eval_runtime": 304.8371,
+      "eval_samples_per_second": 1251.787,
+      "eval_steps_per_second": 39.119,
+      "step": 457920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.24976398143823e-05,
+      "loss": 3.74,
+      "step": 458240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.248925386687178e-05,
+      "loss": 3.7313,
+      "step": 458752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.248088429816499e-05,
+      "loss": 3.7502,
+      "step": 459264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.247249835065447e-05,
+      "loss": 3.7589,
+      "step": 459776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.246411240314395e-05,
+      "loss": 3.7499,
+      "step": 460288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.245572645563343e-05,
+      "loss": 3.739,
+      "step": 460800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2447356886926636e-05,
+      "loss": 3.7448,
+      "step": 461312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2438970939416116e-05,
+      "loss": 3.7287,
+      "step": 461824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2430584991905596e-05,
+      "loss": 3.7462,
+      "step": 462336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2422199044395076e-05,
+      "loss": 3.7457,
+      "step": 462848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2413829475688285e-05,
+      "loss": 3.7449,
+      "step": 463360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2405443528177765e-05,
+      "loss": 3.7456,
+      "step": 463872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.239707395947098e-05,
+      "loss": 3.7403,
+      "step": 464384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.238868801196046e-05,
+      "loss": 3.7378,
+      "step": 464896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.238030206444994e-05,
+      "loss": 3.7329,
+      "step": 465408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.237191611693942e-05,
+      "loss": 3.7279,
+      "step": 465920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.23635301694289e-05,
+      "loss": 3.7354,
+      "step": 466432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.235514422191838e-05,
+      "loss": 3.7348,
+      "step": 466944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.234675827440786e-05,
+      "loss": 3.7369,
+      "step": 467456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.233837232689734e-05,
+      "loss": 3.7497,
+      "step": 467968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.233000275819055e-05,
+      "loss": 3.7475,
+      "step": 468480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.232161681068003e-05,
+      "loss": 3.7421,
+      "step": 468992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.231323086316951e-05,
+      "loss": 3.739,
+      "step": 469504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.230484491565899e-05,
+      "loss": 3.7433,
+      "step": 470016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2296475346952206e-05,
+      "loss": 3.7283,
+      "step": 470528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2288089399441686e-05,
+      "loss": 3.7453,
+      "step": 471040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2279703451931166e-05,
+      "loss": 3.7339,
+      "step": 471552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2271317504420645e-05,
+      "loss": 3.7244,
+      "step": 472064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2262947935713855e-05,
+      "loss": 3.7328,
+      "step": 472576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2254561988203335e-05,
+      "loss": 3.7321,
+      "step": 473088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2246176040692814e-05,
+      "loss": 3.7351,
+      "step": 473600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2237806471986024e-05,
+      "loss": 3.7414,
+      "step": 474112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2229420524475504e-05,
+      "loss": 3.7391,
+      "step": 474624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2221034576964983e-05,
+      "loss": 3.7404,
+      "step": 475136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2212648629454463e-05,
+      "loss": 3.7275,
+      "step": 475648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2204262681943943e-05,
+      "loss": 3.7476,
+      "step": 476160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.219587673443342e-05,
+      "loss": 3.7294,
+      "step": 476672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.21874907869229e-05,
+      "loss": 3.7227,
+      "step": 477184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.217910483941238e-05,
+      "loss": 3.7294,
+      "step": 477696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.21707352707056e-05,
+      "loss": 3.7257,
+      "step": 478208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.216234932319507e-05,
+      "loss": 3.7326,
+      "step": 478720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.215396337568455e-05,
+      "loss": 3.7425,
+      "step": 479232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.214557742817403e-05,
+      "loss": 3.7288,
+      "step": 479744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.213720785946725e-05,
+      "loss": 3.7307,
+      "step": 480256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.212882191195672e-05,
+      "loss": 3.7383,
+      "step": 480768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.21204359644462e-05,
+      "loss": 3.7402,
+      "step": 481280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.211206639573942e-05,
+      "loss": 3.7385,
+      "step": 481792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.21036804482289e-05,
+      "loss": 3.725,
+      "step": 482304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.209529450071838e-05,
+      "loss": 3.701,
+      "step": 482816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.208690855320786e-05,
+      "loss": 3.7439,
+      "step": 483328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.207852260569734e-05,
+      "loss": 3.7373,
+      "step": 483840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2070153036990546e-05,
+      "loss": 3.7332,
+      "step": 484352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2061767089480026e-05,
+      "loss": 3.7261,
+      "step": 484864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2053381141969506e-05,
+      "loss": 3.724,
+      "step": 485376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.204501157326272e-05,
+      "loss": 3.713,
+      "step": 485888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2036625625752195e-05,
+      "loss": 3.7226,
+      "step": 486400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2028239678241675e-05,
+      "loss": 3.7196,
+      "step": 486912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2019853730731155e-05,
+      "loss": 3.7235,
+      "step": 487424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2011484162024364e-05,
+      "loss": 3.7391,
+      "step": 487936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.200309821451385e-05,
+      "loss": 3.7183,
+      "step": 488448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.199471226700333e-05,
+      "loss": 3.7136,
+      "step": 488960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.198632631949281e-05,
+      "loss": 3.731,
+      "step": 489472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.197795675078602e-05,
+      "loss": 3.7151,
+      "step": 489984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.19695708032755e-05,
+      "loss": 3.7155,
+      "step": 490496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.196118485576498e-05,
+      "loss": 3.7263,
+      "step": 491008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.195279890825446e-05,
+      "loss": 3.7232,
+      "step": 491520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.194441296074394e-05,
+      "loss": 3.7135,
+      "step": 492032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.193604339203715e-05,
+      "loss": 3.7147,
+      "step": 492544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.192765744452663e-05,
+      "loss": 3.7149,
+      "step": 493056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.191927149701611e-05,
+      "loss": 3.7262,
+      "step": 493568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.191090192830932e-05,
+      "loss": 3.7257,
+      "step": 494080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1902515980798804e-05,
+      "loss": 3.7215,
+      "step": 494592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1894130033288284e-05,
+      "loss": 3.7248,
+      "step": 495104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1885744085777764e-05,
+      "loss": 3.7291,
+      "step": 495616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1877374517070973e-05,
+      "loss": 3.7285,
+      "step": 496128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.186898856956045e-05,
+      "loss": 3.7139,
+      "step": 496640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.186060262204993e-05,
+      "loss": 3.7283,
+      "step": 497152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.185221667453941e-05,
+      "loss": 3.7293,
+      "step": 497664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.184384710583262e-05,
+      "loss": 3.708,
+      "step": 498176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.18354611583221e-05,
+      "loss": 3.7263,
+      "step": 498688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.182707521081158e-05,
+      "loss": 3.7232,
+      "step": 499200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.181868926330106e-05,
+      "loss": 3.7283,
+      "step": 499712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.181031969459427e-05,
+      "loss": 3.7187,
+      "step": 500224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.180193374708375e-05,
+      "loss": 3.704,
+      "step": 500736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.179354779957324e-05,
+      "loss": 3.7122,
+      "step": 501248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.178516185206272e-05,
+      "loss": 3.7155,
+      "step": 501760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.177679228335593e-05,
+      "loss": 3.7243,
+      "step": 502272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.176840633584541e-05,
+      "loss": 3.7142,
+      "step": 502784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.176002038833489e-05,
+      "loss": 3.7107,
+      "step": 503296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.175163444082437e-05,
+      "loss": 3.7196,
+      "step": 503808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1743264872117576e-05,
+      "loss": 3.7232,
+      "step": 504320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1734878924607056e-05,
+      "loss": 3.7088,
+      "step": 504832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1726492977096536e-05,
+      "loss": 3.7154,
+      "step": 505344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1718107029586016e-05,
+      "loss": 3.7246,
+      "step": 505856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1709737460879225e-05,
+      "loss": 3.7308,
+      "step": 506368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1701351513368705e-05,
+      "loss": 3.718,
+      "step": 506880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.169296556585819e-05,
+      "loss": 3.7043,
+      "step": 507392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.168457961834767e-05,
+      "loss": 3.7183,
+      "step": 507904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.167621004964088e-05,
+      "loss": 3.7107,
+      "step": 508416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.166782410213036e-05,
+      "loss": 3.7245,
+      "step": 508928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.165943815461984e-05,
+      "loss": 3.7279,
+      "step": 509440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.165105220710932e-05,
+      "loss": 3.7129,
+      "step": 509952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.164268263840253e-05,
+      "loss": 3.7066,
+      "step": 510464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.163429669089201e-05,
+      "loss": 3.7157,
+      "step": 510976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.162591074338149e-05,
+      "loss": 3.7008,
+      "step": 511488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.161752479587097e-05,
+      "loss": 3.7196,
+      "step": 512000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.160915522716418e-05,
+      "loss": 3.7003,
+      "step": 512512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.160076927965366e-05,
+      "loss": 3.7217,
+      "step": 513024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1592383332143145e-05,
+      "loss": 3.7051,
+      "step": 513536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1583997384632625e-05,
+      "loss": 3.7151,
+      "step": 514048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1575627815925834e-05,
+      "loss": 3.711,
+      "step": 514560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1567241868415314e-05,
+      "loss": 3.7119,
+      "step": 515072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1558855920904794e-05,
+      "loss": 3.7201,
+      "step": 515584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1550469973394274e-05,
+      "loss": 3.71,
+      "step": 516096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.154210040468748e-05,
+      "loss": 3.7281,
+      "step": 516608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.153371445717696e-05,
+      "loss": 3.7222,
+      "step": 517120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.152532850966644e-05,
+      "loss": 3.715,
+      "step": 517632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.151694256215592e-05,
+      "loss": 3.7182,
+      "step": 518144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.150857299344913e-05,
+      "loss": 3.7081,
+      "step": 518656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.150018704593861e-05,
+      "loss": 3.7136,
+      "step": 519168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.14918010984281e-05,
+      "loss": 3.7121,
+      "step": 519680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.148341515091758e-05,
+      "loss": 3.7101,
+      "step": 520192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.147504558221079e-05,
+      "loss": 3.7125,
+      "step": 520704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.146665963470027e-05,
+      "loss": 3.7111,
+      "step": 521216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.145827368718975e-05,
+      "loss": 3.7123,
+      "step": 521728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.144988773967923e-05,
+      "loss": 3.7132,
+      "step": 522240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.144151817097244e-05,
+      "loss": 3.7084,
+      "step": 522752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.143313222346192e-05,
+      "loss": 3.714,
+      "step": 523264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.14247462759514e-05,
+      "loss": 3.7011,
+      "step": 523776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.141636032844088e-05,
+      "loss": 3.7257,
+      "step": 524288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1407990759734086e-05,
+      "loss": 3.7142,
+      "step": 524800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1399604812223566e-05,
+      "loss": 3.721,
+      "step": 525312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.139121886471305e-05,
+      "loss": 3.701,
+      "step": 525824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.138283291720253e-05,
+      "loss": 3.7145,
+      "step": 526336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.137446334849574e-05,
+      "loss": 3.7178,
+      "step": 526848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.136607740098522e-05,
+      "loss": 3.7154,
+      "step": 527360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.13576914534747e-05,
+      "loss": 3.7042,
+      "step": 527872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.134930550596418e-05,
+      "loss": 3.7222,
+      "step": 528384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.134093593725739e-05,
+      "loss": 3.712,
+      "step": 528896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.133254998974687e-05,
+      "loss": 3.7274,
+      "step": 529408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.132416404223635e-05,
+      "loss": 3.7059,
+      "step": 529920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.131577809472583e-05,
+      "loss": 3.721,
+      "step": 530432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.130740852601904e-05,
+      "loss": 3.7135,
+      "step": 530944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.129902257850852e-05,
+      "loss": 3.7092,
+      "step": 531456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1290636630998006e-05,
+      "loss": 3.7134,
+      "step": 531968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1282250683487486e-05,
+      "loss": 3.7161,
+      "step": 532480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1273881114780695e-05,
+      "loss": 3.7175,
+      "step": 532992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1265495167270175e-05,
+      "loss": 3.6947,
+      "step": 533504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1257109219759655e-05,
+      "loss": 3.7207,
+      "step": 534016
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8663017749786377,
+      "eval_runtime": 302.6553,
+      "eval_samples_per_second": 1260.811,
+      "eval_steps_per_second": 39.401,
+      "step": 534240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1248723272249135e-05,
+      "loss": 3.7056,
+      "step": 534528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1240337324738615e-05,
+      "loss": 3.6955,
+      "step": 535040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1231951377228095e-05,
+      "loss": 3.7138,
+      "step": 535552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.122356542971757e-05,
+      "loss": 3.7209,
+      "step": 536064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.121517948220705e-05,
+      "loss": 3.7139,
+      "step": 536576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.120679353469653e-05,
+      "loss": 3.7069,
+      "step": 537088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.119840758718601e-05,
+      "loss": 3.7112,
+      "step": 537600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.119002163967549e-05,
+      "loss": 3.6964,
+      "step": 538112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1181652070968704e-05,
+      "loss": 3.7106,
+      "step": 538624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1173266123458184e-05,
+      "loss": 3.7105,
+      "step": 539136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1164880175947664e-05,
+      "loss": 3.7078,
+      "step": 539648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1156494228437144e-05,
+      "loss": 3.7101,
+      "step": 540160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.114814103853409e-05,
+      "loss": 3.7002,
+      "step": 540672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.113975509102357e-05,
+      "loss": 3.702,
+      "step": 541184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.113136914351304e-05,
+      "loss": 3.7024,
+      "step": 541696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.112298319600252e-05,
+      "loss": 3.6919,
+      "step": 542208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1114597248492e-05,
+      "loss": 3.7016,
+      "step": 542720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.110621130098148e-05,
+      "loss": 3.698,
+      "step": 543232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.109782535347096e-05,
+      "loss": 3.7015,
+      "step": 543744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.108943940596044e-05,
+      "loss": 3.7119,
+      "step": 544256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.108106983725366e-05,
+      "loss": 3.7154,
+      "step": 544768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.107268388974314e-05,
+      "loss": 3.7111,
+      "step": 545280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.106429794223262e-05,
+      "loss": 3.7008,
+      "step": 545792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.10559119947221e-05,
+      "loss": 3.7067,
+      "step": 546304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.104754242601531e-05,
+      "loss": 3.697,
+      "step": 546816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.103915647850479e-05,
+      "loss": 3.7093,
+      "step": 547328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.103077053099427e-05,
+      "loss": 3.7013,
+      "step": 547840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1022384583483747e-05,
+      "loss": 3.6919,
+      "step": 548352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1014015014776956e-05,
+      "loss": 3.7022,
+      "step": 548864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1005629067266436e-05,
+      "loss": 3.6928,
+      "step": 549376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0997243119755916e-05,
+      "loss": 3.7,
+      "step": 549888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0988857172245396e-05,
+      "loss": 3.7119,
+      "step": 550400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.098048760353861e-05,
+      "loss": 3.7028,
+      "step": 550912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.097210165602809e-05,
+      "loss": 3.7053,
+      "step": 551424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.096371570851757e-05,
+      "loss": 3.6921,
+      "step": 551936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.095532976100705e-05,
+      "loss": 3.713,
+      "step": 552448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.094696019230026e-05,
+      "loss": 3.6989,
+      "step": 552960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.093857424478974e-05,
+      "loss": 3.6879,
+      "step": 553472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.093018829727922e-05,
+      "loss": 3.697,
+      "step": 553984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.092181872857243e-05,
+      "loss": 3.6917,
+      "step": 554496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.091343278106191e-05,
+      "loss": 3.6978,
+      "step": 555008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.090504683355139e-05,
+      "loss": 3.7028,
+      "step": 555520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.089666088604087e-05,
+      "loss": 3.6992,
+      "step": 556032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0888291317334085e-05,
+      "loss": 3.6957,
+      "step": 556544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0879905369823565e-05,
+      "loss": 3.7048,
+      "step": 557056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0871519422313045e-05,
+      "loss": 3.7049,
+      "step": 557568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0863133474802525e-05,
+      "loss": 3.7055,
+      "step": 558080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0854763906095734e-05,
+      "loss": 3.691,
+      "step": 558592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0846377958585214e-05,
+      "loss": 3.6674,
+      "step": 559104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0837992011074694e-05,
+      "loss": 3.7148,
+      "step": 559616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0829606063564174e-05,
+      "loss": 3.6991,
+      "step": 560128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.082123649485738e-05,
+      "loss": 3.7024,
+      "step": 560640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.081285054734686e-05,
+      "loss": 3.6946,
+      "step": 561152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.080446459983634e-05,
+      "loss": 3.6867,
+      "step": 561664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.079607865232582e-05,
+      "loss": 3.6829,
+      "step": 562176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.078770908361904e-05,
+      "loss": 3.6913,
+      "step": 562688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.077932313610852e-05,
+      "loss": 3.6852,
+      "step": 563200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0770937188598e-05,
+      "loss": 3.6863,
+      "step": 563712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.076255124108748e-05,
+      "loss": 3.7084,
+      "step": 564224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.075418167238069e-05,
+      "loss": 3.6845,
+      "step": 564736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.074579572487017e-05,
+      "loss": 3.6811,
+      "step": 565248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.073740977735965e-05,
+      "loss": 3.7026,
+      "step": 565760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.072902382984913e-05,
+      "loss": 3.6769,
+      "step": 566272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.072065426114234e-05,
+      "loss": 3.6819,
+      "step": 566784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.071226831363182e-05,
+      "loss": 3.6942,
+      "step": 567296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.07038823661213e-05,
+      "loss": 3.693,
+      "step": 567808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0695496418610777e-05,
+      "loss": 3.6805,
+      "step": 568320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.068712684990399e-05,
+      "loss": 3.6894,
+      "step": 568832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.067874090239347e-05,
+      "loss": 3.6767,
+      "step": 569344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.067037133368668e-05,
+      "loss": 3.6933,
+      "step": 569856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.066198538617616e-05,
+      "loss": 3.6917,
+      "step": 570368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.065359943866564e-05,
+      "loss": 3.6889,
+      "step": 570880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.064521349115512e-05,
+      "loss": 3.6945,
+      "step": 571392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.06368275436446e-05,
+      "loss": 3.6928,
+      "step": 571904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.062844159613408e-05,
+      "loss": 3.7019,
+      "step": 572416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.062005564862356e-05,
+      "loss": 3.682,
+      "step": 572928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.061168607991677e-05,
+      "loss": 3.694,
+      "step": 573440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.060330013240625e-05,
+      "loss": 3.6972,
+      "step": 573952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.059491418489573e-05,
+      "loss": 3.6719,
+      "step": 574464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.058652823738521e-05,
+      "loss": 3.6966,
+      "step": 574976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0578158668678426e-05,
+      "loss": 3.6889,
+      "step": 575488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0569772721167906e-05,
+      "loss": 3.697,
+      "step": 576000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0561386773657386e-05,
+      "loss": 3.6888,
+      "step": 576512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0553000826146866e-05,
+      "loss": 3.6753,
+      "step": 577024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0544631257440075e-05,
+      "loss": 3.6726,
+      "step": 577536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0536245309929555e-05,
+      "loss": 3.6843,
+      "step": 578048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0527859362419035e-05,
+      "loss": 3.6955,
+      "step": 578560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0519473414908515e-05,
+      "loss": 3.6805,
+      "step": 579072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0511103846201724e-05,
+      "loss": 3.6811,
+      "step": 579584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0502717898691204e-05,
+      "loss": 3.6867,
+      "step": 580096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0494331951180684e-05,
+      "loss": 3.6906,
+      "step": 580608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0485946003670164e-05,
+      "loss": 3.68,
+      "step": 581120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.047757643496338e-05,
+      "loss": 3.6832,
+      "step": 581632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.046919048745286e-05,
+      "loss": 3.6922,
+      "step": 582144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.046080453994234e-05,
+      "loss": 3.6967,
+      "step": 582656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.045241859243182e-05,
+      "loss": 3.6886,
+      "step": 583168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.044404902372503e-05,
+      "loss": 3.6737,
+      "step": 583680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.043566307621451e-05,
+      "loss": 3.6871,
+      "step": 584192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.042727712870399e-05,
+      "loss": 3.6786,
+      "step": 584704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.041889118119347e-05,
+      "loss": 3.6939,
+      "step": 585216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.041052161248668e-05,
+      "loss": 3.6884,
+      "step": 585728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.040213566497616e-05,
+      "loss": 3.6861,
+      "step": 586240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.039374971746564e-05,
+      "loss": 3.679,
+      "step": 586752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.038536376995512e-05,
+      "loss": 3.6782,
+      "step": 587264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0376994201248333e-05,
+      "loss": 3.6732,
+      "step": 587776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.036860825373781e-05,
+      "loss": 3.6861,
+      "step": 588288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.036022230622729e-05,
+      "loss": 3.673,
+      "step": 588800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0351836358716766e-05,
+      "loss": 3.6911,
+      "step": 589312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.034348316881371e-05,
+      "loss": 3.6764,
+      "step": 589824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.033509722130319e-05,
+      "loss": 3.6813,
+      "step": 590336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.032671127379267e-05,
+      "loss": 3.6787,
+      "step": 590848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.031832532628215e-05,
+      "loss": 3.6776,
+      "step": 591360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.030993937877163e-05,
+      "loss": 3.6919,
+      "step": 591872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.030155343126111e-05,
+      "loss": 3.6811,
+      "step": 592384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.029316748375059e-05,
+      "loss": 3.6949,
+      "step": 592896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.028478153624007e-05,
+      "loss": 3.6938,
+      "step": 593408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.027641196753329e-05,
+      "loss": 3.682,
+      "step": 593920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.026802602002277e-05,
+      "loss": 3.6832,
+      "step": 594432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.025964007251224e-05,
+      "loss": 3.6793,
+      "step": 594944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.025125412500172e-05,
+      "loss": 3.6812,
+      "step": 595456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0242884556294936e-05,
+      "loss": 3.685,
+      "step": 595968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0234498608784416e-05,
+      "loss": 3.6787,
+      "step": 596480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.022611266127389e-05,
+      "loss": 3.6812,
+      "step": 596992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.021772671376337e-05,
+      "loss": 3.6826,
+      "step": 597504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0209357145056585e-05,
+      "loss": 3.6804,
+      "step": 598016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0200971197546065e-05,
+      "loss": 3.6825,
+      "step": 598528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0192601628839274e-05,
+      "loss": 3.6738,
+      "step": 599040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0184215681328754e-05,
+      "loss": 3.6813,
+      "step": 599552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.017582973381824e-05,
+      "loss": 3.6752,
+      "step": 600064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0167443786307714e-05,
+      "loss": 3.6936,
+      "step": 600576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0159057838797194e-05,
+      "loss": 3.683,
+      "step": 601088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0150671891286674e-05,
+      "loss": 3.6888,
+      "step": 601600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0142285943776154e-05,
+      "loss": 3.6735,
+      "step": 602112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0133899996265634e-05,
+      "loss": 3.6844,
+      "step": 602624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.012553042755884e-05,
+      "loss": 3.6838,
+      "step": 603136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.011714448004832e-05,
+      "loss": 3.6861,
+      "step": 603648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.01087585325378e-05,
+      "loss": 3.6766,
+      "step": 604160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.010037258502728e-05,
+      "loss": 3.6885,
+      "step": 604672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.009200301632049e-05,
+      "loss": 3.6803,
+      "step": 605184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.008361706880998e-05,
+      "loss": 3.7012,
+      "step": 605696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.007523112129946e-05,
+      "loss": 3.6742,
+      "step": 606208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.006684517378894e-05,
+      "loss": 3.6895,
+      "step": 606720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.005847560508215e-05,
+      "loss": 3.684,
+      "step": 607232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.005008965757163e-05,
+      "loss": 3.6782,
+      "step": 607744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.004170371006111e-05,
+      "loss": 3.6819,
+      "step": 608256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0033334141354317e-05,
+      "loss": 3.6866,
+      "step": 608768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0024948193843797e-05,
+      "loss": 3.686,
+      "step": 609280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0016562246333276e-05,
+      "loss": 3.6664,
+      "step": 609792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0008176298822756e-05,
+      "loss": 3.6892,
+      "step": 610304
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8588056564331055,
+      "eval_runtime": 307.1281,
+      "eval_samples_per_second": 1242.449,
+      "eval_steps_per_second": 38.827,
+      "step": 610560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9999790351312236e-05,
+      "loss": 3.6704,
+      "step": 610816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9991404403801716e-05,
+      "loss": 3.6663,
+      "step": 611328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9983018456291196e-05,
+      "loss": 3.68,
+      "step": 611840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9974632508780676e-05,
+      "loss": 3.6917,
+      "step": 612352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.996624656127016e-05,
+      "loss": 3.6858,
+      "step": 612864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.995786061375964e-05,
+      "loss": 3.6737,
+      "step": 613376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.994947466624912e-05,
+      "loss": 3.6847,
+      "step": 613888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.99410887187386e-05,
+      "loss": 3.6674,
+      "step": 614400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.993271915003181e-05,
+      "loss": 3.6782,
+      "step": 614912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.992433320252129e-05,
+      "loss": 3.6829,
+      "step": 615424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.991594725501077e-05,
+      "loss": 3.6773,
+      "step": 615936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.990757768630398e-05,
+      "loss": 3.6781,
+      "step": 616448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.989919173879346e-05,
+      "loss": 3.6779,
+      "step": 616960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.989080579128294e-05,
+      "loss": 3.6676,
+      "step": 617472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.988241984377242e-05,
+      "loss": 3.6677,
+      "step": 617984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.98740338962619e-05,
+      "loss": 3.664,
+      "step": 618496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.986564794875138e-05,
+      "loss": 3.6706,
+      "step": 619008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.985726200124086e-05,
+      "loss": 3.6726,
+      "step": 619520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9848892432534077e-05,
+      "loss": 3.6676,
+      "step": 620032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.984050648502355e-05,
+      "loss": 3.6828,
+      "step": 620544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.983212053751303e-05,
+      "loss": 3.6872,
+      "step": 621056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.982373459000251e-05,
+      "loss": 3.6836,
+      "step": 621568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9815365021295726e-05,
+      "loss": 3.6754,
+      "step": 622080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.98069790737852e-05,
+      "loss": 3.6745,
+      "step": 622592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.979859312627468e-05,
+      "loss": 3.6738,
+      "step": 623104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.979020717876416e-05,
+      "loss": 3.6734,
+      "step": 623616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.978183761005737e-05,
+      "loss": 3.6729,
+      "step": 624128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9773451662546854e-05,
+      "loss": 3.6676,
+      "step": 624640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9765065715036334e-05,
+      "loss": 3.6694,
+      "step": 625152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9756679767525814e-05,
+      "loss": 3.6646,
+      "step": 625664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9748310198819023e-05,
+      "loss": 3.6679,
+      "step": 626176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9739924251308503e-05,
+      "loss": 3.6809,
+      "step": 626688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.973153830379798e-05,
+      "loss": 3.6766,
+      "step": 627200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.972315235628746e-05,
+      "loss": 3.6753,
+      "step": 627712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.971478278758067e-05,
+      "loss": 3.6675,
+      "step": 628224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.970639684007015e-05,
+      "loss": 3.6767,
+      "step": 628736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.969802727136337e-05,
+      "loss": 3.6698,
+      "step": 629248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.968964132385284e-05,
+      "loss": 3.6629,
+      "step": 629760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.968125537634232e-05,
+      "loss": 3.6672,
+      "step": 630272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.967286942883181e-05,
+      "loss": 3.6608,
+      "step": 630784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.966448348132129e-05,
+      "loss": 3.6665,
+      "step": 631296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.965609753381077e-05,
+      "loss": 3.6742,
+      "step": 631808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.964771158630025e-05,
+      "loss": 3.6693,
+      "step": 632320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.963932563878973e-05,
+      "loss": 3.6696,
+      "step": 632832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.963095607008294e-05,
+      "loss": 3.6771,
+      "step": 633344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.962257012257242e-05,
+      "loss": 3.6726,
+      "step": 633856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.96141841750619e-05,
+      "loss": 3.6742,
+      "step": 634368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.960579822755138e-05,
+      "loss": 3.6668,
+      "step": 634880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9597428658844586e-05,
+      "loss": 3.6384,
+      "step": 635392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9589059090137795e-05,
+      "loss": 3.6824,
+      "step": 635904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9580673142627275e-05,
+      "loss": 3.6697,
+      "step": 636416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.957228719511676e-05,
+      "loss": 3.6736,
+      "step": 636928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.956390124760624e-05,
+      "loss": 3.669,
+      "step": 637440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.955551530009572e-05,
+      "loss": 3.6583,
+      "step": 637952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.95471293525852e-05,
+      "loss": 3.6529,
+      "step": 638464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.953874340507468e-05,
+      "loss": 3.6636,
+      "step": 638976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.953037383636789e-05,
+      "loss": 3.6569,
+      "step": 639488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.952198788885737e-05,
+      "loss": 3.6578,
+      "step": 640000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.951360194134685e-05,
+      "loss": 3.6785,
+      "step": 640512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.950521599383633e-05,
+      "loss": 3.6589,
+      "step": 641024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.949684642512954e-05,
+      "loss": 3.6513,
+      "step": 641536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.948846047761902e-05,
+      "loss": 3.6751,
+      "step": 642048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.94800745301085e-05,
+      "loss": 3.6472,
+      "step": 642560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.947168858259798e-05,
+      "loss": 3.6531,
+      "step": 643072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9463319013891195e-05,
+      "loss": 3.6689,
+      "step": 643584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9454933066380675e-05,
+      "loss": 3.6658,
+      "step": 644096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9446547118870155e-05,
+      "loss": 3.6537,
+      "step": 644608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9438161171359635e-05,
+      "loss": 3.6577,
+      "step": 645120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9429791602652844e-05,
+      "loss": 3.6522,
+      "step": 645632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9421405655142324e-05,
+      "loss": 3.6661,
+      "step": 646144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9413019707631804e-05,
+      "loss": 3.6605,
+      "step": 646656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9404633760121284e-05,
+      "loss": 3.6606,
+      "step": 647168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.939626419141449e-05,
+      "loss": 3.6653,
+      "step": 647680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.938787824390397e-05,
+      "loss": 3.6653,
+      "step": 648192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.937949229639345e-05,
+      "loss": 3.675,
+      "step": 648704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.937110634888293e-05,
+      "loss": 3.6525,
+      "step": 649216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.936273678017615e-05,
+      "loss": 3.6691,
+      "step": 649728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.935435083266563e-05,
+      "loss": 3.6674,
+      "step": 650240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.934596488515511e-05,
+      "loss": 3.6423,
+      "step": 650752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.933757893764459e-05,
+      "loss": 3.6671,
+      "step": 651264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.93292093689378e-05,
+      "loss": 3.6643,
+      "step": 651776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.932082342142728e-05,
+      "loss": 3.6697,
+      "step": 652288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.931243747391676e-05,
+      "loss": 3.6603,
+      "step": 652800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.930405152640624e-05,
+      "loss": 3.6502,
+      "step": 653312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.929568195769945e-05,
+      "loss": 3.643,
+      "step": 653824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.928729601018893e-05,
+      "loss": 3.6573,
+      "step": 654336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.927891006267841e-05,
+      "loss": 3.6671,
+      "step": 654848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.927052411516789e-05,
+      "loss": 3.6545,
+      "step": 655360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.92621545464611e-05,
+      "loss": 3.6523,
+      "step": 655872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.925376859895058e-05,
+      "loss": 3.6548,
+      "step": 656384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.924538265144006e-05,
+      "loss": 3.667,
+      "step": 656896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.923699670392954e-05,
+      "loss": 3.6516,
+      "step": 657408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.922862713522275e-05,
+      "loss": 3.6535,
+      "step": 657920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.922024118771223e-05,
+      "loss": 3.6627,
+      "step": 658432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.921185524020171e-05,
+      "loss": 3.6684,
+      "step": 658944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.920346929269119e-05,
+      "loss": 3.6604,
+      "step": 659456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.91950997239844e-05,
+      "loss": 3.6488,
+      "step": 659968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.918671377647388e-05,
+      "loss": 3.6536,
+      "step": 660480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.917832782896336e-05,
+      "loss": 3.6533,
+      "step": 660992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.916994188145284e-05,
+      "loss": 3.6659,
+      "step": 661504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9161572312746056e-05,
+      "loss": 3.6628,
+      "step": 662016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9153186365235536e-05,
+      "loss": 3.6595,
+      "step": 662528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9144816796528745e-05,
+      "loss": 3.6521,
+      "step": 663040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9136430849018225e-05,
+      "loss": 3.648,
+      "step": 663552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9128044901507705e-05,
+      "loss": 3.6509,
+      "step": 664064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9119658953997185e-05,
+      "loss": 3.6566,
+      "step": 664576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9111273006486665e-05,
+      "loss": 3.6448,
+      "step": 665088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9102887058976145e-05,
+      "loss": 3.6666,
+      "step": 665600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9094517490269354e-05,
+      "loss": 3.6496,
+      "step": 666112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9086131542758834e-05,
+      "loss": 3.6522,
+      "step": 666624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9077745595248314e-05,
+      "loss": 3.6499,
+      "step": 667136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9069359647737794e-05,
+      "loss": 3.6477,
+      "step": 667648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9060973700227274e-05,
+      "loss": 3.6675,
+      "step": 668160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9052587752716754e-05,
+      "loss": 3.6479,
+      "step": 668672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9044201805206234e-05,
+      "loss": 3.6713,
+      "step": 669184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9035815857695714e-05,
+      "loss": 3.6654,
+      "step": 669696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.902744628898892e-05,
+      "loss": 3.6577,
+      "step": 670208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.90190603414784e-05,
+      "loss": 3.6551,
+      "step": 670720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.901067439396788e-05,
+      "loss": 3.6573,
+      "step": 671232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900228844645736e-05,
+      "loss": 3.6537,
+      "step": 671744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.899391887775057e-05,
+      "loss": 3.6603,
+      "step": 672256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.898553293024005e-05,
+      "loss": 3.648,
+      "step": 672768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.897714698272953e-05,
+      "loss": 3.6571,
+      "step": 673280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.896876103521901e-05,
+      "loss": 3.6523,
+      "step": 673792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.896039146651223e-05,
+      "loss": 3.654,
+      "step": 674304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.895200551900171e-05,
+      "loss": 3.6558,
+      "step": 674816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8943635950294924e-05,
+      "loss": 3.6471,
+      "step": 675328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.89352500027844e-05,
+      "loss": 3.6583,
+      "step": 675840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.892686405527388e-05,
+      "loss": 3.6463,
+      "step": 676352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.891847810776336e-05,
+      "loss": 3.6657,
+      "step": 676864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.891009216025284e-05,
+      "loss": 3.6565,
+      "step": 677376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8901722591546046e-05,
+      "loss": 3.6618,
+      "step": 677888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8893336644035526e-05,
+      "loss": 3.6456,
+      "step": 678400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8884950696525006e-05,
+      "loss": 3.6621,
+      "step": 678912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8876564749014486e-05,
+      "loss": 3.6531,
+      "step": 679424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8868178801503966e-05,
+      "loss": 3.6581,
+      "step": 679936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8859792853993446e-05,
+      "loss": 3.6503,
+      "step": 680448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.885140690648293e-05,
+      "loss": 3.6609,
+      "step": 680960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.884302095897241e-05,
+      "loss": 3.6544,
+      "step": 681472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.883465139026562e-05,
+      "loss": 3.6757,
+      "step": 681984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88262654427551e-05,
+      "loss": 3.648,
+      "step": 682496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.881787949524458e-05,
+      "loss": 3.6612,
+      "step": 683008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.880949354773406e-05,
+      "loss": 3.6575,
+      "step": 683520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.880112397902727e-05,
+      "loss": 3.6552,
+      "step": 684032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.879273803151675e-05,
+      "loss": 3.651,
+      "step": 684544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.878435208400623e-05,
+      "loss": 3.6613,
+      "step": 685056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.877596613649571e-05,
+      "loss": 3.6585,
+      "step": 685568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.876759656778892e-05,
+      "loss": 3.6473,
+      "step": 686080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.87592106202784e-05,
+      "loss": 3.6577,
+      "step": 686592
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8553671836853027,
+      "eval_runtime": 307.9957,
+      "eval_samples_per_second": 1238.949,
+      "eval_steps_per_second": 38.718,
+      "step": 686880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8750841051571615e-05,
+      "loss": 3.6493,
+      "step": 687104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8742455104061095e-05,
+      "loss": 3.6377,
+      "step": 687616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8734069156550575e-05,
+      "loss": 3.6524,
+      "step": 688128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8725683209040055e-05,
+      "loss": 3.6679,
+      "step": 688640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8717297261529535e-05,
+      "loss": 3.662,
+      "step": 689152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8708911314019015e-05,
+      "loss": 3.6439,
+      "step": 689664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8700525366508495e-05,
+      "loss": 3.6605,
+      "step": 690176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8692139418997975e-05,
+      "loss": 3.6414,
+      "step": 690688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8683769850291184e-05,
+      "loss": 3.6508,
+      "step": 691200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8675383902780664e-05,
+      "loss": 3.6606,
+      "step": 691712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8666997955270144e-05,
+      "loss": 3.6483,
+      "step": 692224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.865862838656335e-05,
+      "loss": 3.6513,
+      "step": 692736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.865024243905283e-05,
+      "loss": 3.6515,
+      "step": 693248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.864185649154232e-05,
+      "loss": 3.6435,
+      "step": 693760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.86334705440318e-05,
+      "loss": 3.6426,
+      "step": 694272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.862508459652128e-05,
+      "loss": 3.6377,
+      "step": 694784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.861671502781449e-05,
+      "loss": 3.6436,
+      "step": 695296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.860832908030397e-05,
+      "loss": 3.647,
+      "step": 695808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.859994313279345e-05,
+      "loss": 3.6424,
+      "step": 696320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.859155718528293e-05,
+      "loss": 3.6572,
+      "step": 696832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.858317123777241e-05,
+      "loss": 3.6605,
+      "step": 697344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.857478529026188e-05,
+      "loss": 3.6571,
+      "step": 697856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.856639934275136e-05,
+      "loss": 3.6507,
+      "step": 698368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.855801339524084e-05,
+      "loss": 3.6435,
+      "step": 698880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.854964382653406e-05,
+      "loss": 3.65,
+      "step": 699392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.854125787902354e-05,
+      "loss": 3.6468,
+      "step": 699904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.853287193151302e-05,
+      "loss": 3.6439,
+      "step": 700416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.85244859840025e-05,
+      "loss": 3.6439,
+      "step": 700928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.851610003649198e-05,
+      "loss": 3.6429,
+      "step": 701440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.850771408898146e-05,
+      "loss": 3.6433,
+      "step": 701952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.849932814147094e-05,
+      "loss": 3.6423,
+      "step": 702464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.849094219396042e-05,
+      "loss": 3.6531,
+      "step": 702976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8482572625253626e-05,
+      "loss": 3.6555,
+      "step": 703488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8474186677743106e-05,
+      "loss": 3.6475,
+      "step": 704000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8465800730232586e-05,
+      "loss": 3.6442,
+      "step": 704512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8457431161525795e-05,
+      "loss": 3.6438,
+      "step": 705024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8449045214015275e-05,
+      "loss": 3.651,
+      "step": 705536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8440659266504755e-05,
+      "loss": 3.6382,
+      "step": 706048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.843227331899424e-05,
+      "loss": 3.642,
+      "step": 706560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.842390375028745e-05,
+      "loss": 3.6334,
+      "step": 707072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.841553418158066e-05,
+      "loss": 3.6449,
+      "step": 707584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.840714823407014e-05,
+      "loss": 3.6447,
+      "step": 708096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.839876228655962e-05,
+      "loss": 3.6453,
+      "step": 708608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.83903763390491e-05,
+      "loss": 3.6466,
+      "step": 709120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.838199039153858e-05,
+      "loss": 3.6495,
+      "step": 709632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.837360444402806e-05,
+      "loss": 3.6487,
+      "step": 710144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.836521849651754e-05,
+      "loss": 3.6499,
+      "step": 710656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.835683254900702e-05,
+      "loss": 3.6441,
+      "step": 711168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.834846298030023e-05,
+      "loss": 3.6128,
+      "step": 711680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8340093411593445e-05,
+      "loss": 3.6544,
+      "step": 712192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8331707464082925e-05,
+      "loss": 3.6453,
+      "step": 712704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8323321516572405e-05,
+      "loss": 3.649,
+      "step": 713216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8314935569061885e-05,
+      "loss": 3.6435,
+      "step": 713728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8306549621551364e-05,
+      "loss": 3.6354,
+      "step": 714240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8298180052844574e-05,
+      "loss": 3.6264,
+      "step": 714752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8289794105334054e-05,
+      "loss": 3.6359,
+      "step": 715264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8281408157823533e-05,
+      "loss": 3.6355,
+      "step": 715776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8273022210313013e-05,
+      "loss": 3.6299,
+      "step": 716288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8264636262802493e-05,
+      "loss": 3.6539,
+      "step": 716800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.825625031529197e-05,
+      "loss": 3.6372,
+      "step": 717312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.824786436778145e-05,
+      "loss": 3.6217,
+      "step": 717824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.823947842027093e-05,
+      "loss": 3.6519,
+      "step": 718336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.823110885156415e-05,
+      "loss": 3.624,
+      "step": 718848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.822272290405363e-05,
+      "loss": 3.6273,
+      "step": 719360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.821435333534684e-05,
+      "loss": 3.6436,
+      "step": 719872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.820596738783632e-05,
+      "loss": 3.638,
+      "step": 720384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.81975814403258e-05,
+      "loss": 3.6306,
+      "step": 720896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.818919549281528e-05,
+      "loss": 3.6307,
+      "step": 721408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.818080954530476e-05,
+      "loss": 3.629,
+      "step": 721920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.817242359779424e-05,
+      "loss": 3.6385,
+      "step": 722432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.816403765028371e-05,
+      "loss": 3.6362,
+      "step": 722944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.815565170277319e-05,
+      "loss": 3.6385,
+      "step": 723456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.814728213406641e-05,
+      "loss": 3.639,
+      "step": 723968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.813889618655589e-05,
+      "loss": 3.6379,
+      "step": 724480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.813051023904537e-05,
+      "loss": 3.653,
+      "step": 724992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.812212429153485e-05,
+      "loss": 3.6265,
+      "step": 725504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.811375472282806e-05,
+      "loss": 3.6444,
+      "step": 726016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.810536877531754e-05,
+      "loss": 3.6457,
+      "step": 726528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.809699920661075e-05,
+      "loss": 3.6202,
+      "step": 727040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.808861325910023e-05,
+      "loss": 3.6368,
+      "step": 727552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.808022731158971e-05,
+      "loss": 3.6429,
+      "step": 728064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8071841364079185e-05,
+      "loss": 3.6431,
+      "step": 728576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8063455416568665e-05,
+      "loss": 3.6413,
+      "step": 729088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8055069469058145e-05,
+      "loss": 3.621,
+      "step": 729600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8046683521547625e-05,
+      "loss": 3.6172,
+      "step": 730112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8038297574037105e-05,
+      "loss": 3.6347,
+      "step": 730624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.802992800533032e-05,
+      "loss": 3.6409,
+      "step": 731136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.80215420578198e-05,
+      "loss": 3.6251,
+      "step": 731648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.801315611030928e-05,
+      "loss": 3.6329,
+      "step": 732160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.800477016279876e-05,
+      "loss": 3.6294,
+      "step": 732672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.799640059409197e-05,
+      "loss": 3.6411,
+      "step": 733184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.798801464658145e-05,
+      "loss": 3.6318,
+      "step": 733696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.797962869907093e-05,
+      "loss": 3.6308,
+      "step": 734208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.797124275156041e-05,
+      "loss": 3.634,
+      "step": 734720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.796287318285362e-05,
+      "loss": 3.6439,
+      "step": 735232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.79544872353431e-05,
+      "loss": 3.6388,
+      "step": 735744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.794611766663631e-05,
+      "loss": 3.6253,
+      "step": 736256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.793773171912579e-05,
+      "loss": 3.6291,
+      "step": 736768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7929345771615274e-05,
+      "loss": 3.6293,
+      "step": 737280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7920959824104754e-05,
+      "loss": 3.6438,
+      "step": 737792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7912573876594234e-05,
+      "loss": 3.6363,
+      "step": 738304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7904187929083714e-05,
+      "loss": 3.6345,
+      "step": 738816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.789581836037692e-05,
+      "loss": 3.629,
+      "step": 739328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.78874324128664e-05,
+      "loss": 3.6225,
+      "step": 739840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.787904646535588e-05,
+      "loss": 3.6308,
+      "step": 740352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.787066051784536e-05,
+      "loss": 3.6266,
+      "step": 740864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.786229094913857e-05,
+      "loss": 3.6246,
+      "step": 741376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.785390500162805e-05,
+      "loss": 3.6422,
+      "step": 741888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.784551905411753e-05,
+      "loss": 3.6261,
+      "step": 742400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.783713310660701e-05,
+      "loss": 3.628,
+      "step": 742912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.782874715909649e-05,
+      "loss": 3.6249,
+      "step": 743424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.782037759038971e-05,
+      "loss": 3.6223,
+      "step": 743936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.781199164287919e-05,
+      "loss": 3.6413,
+      "step": 744448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.780360569536867e-05,
+      "loss": 3.628,
+      "step": 744960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.779521974785815e-05,
+      "loss": 3.6449,
+      "step": 745472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.778683380034763e-05,
+      "loss": 3.641,
+      "step": 745984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.777844785283711e-05,
+      "loss": 3.6353,
+      "step": 746496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.777006190532659e-05,
+      "loss": 3.6309,
+      "step": 747008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.776167595781607e-05,
+      "loss": 3.6353,
+      "step": 747520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7753322767913006e-05,
+      "loss": 3.6274,
+      "step": 748032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7744936820402486e-05,
+      "loss": 3.6383,
+      "step": 748544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7736550872891966e-05,
+      "loss": 3.6302,
+      "step": 749056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7728164925381446e-05,
+      "loss": 3.6311,
+      "step": 749568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7719778977870926e-05,
+      "loss": 3.6258,
+      "step": 750080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.771139303036041e-05,
+      "loss": 3.6301,
+      "step": 750592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.770300708284989e-05,
+      "loss": 3.6339,
+      "step": 751104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.76946375141431e-05,
+      "loss": 3.6251,
+      "step": 751616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.768625156663258e-05,
+      "loss": 3.6338,
+      "step": 752128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.767786561912206e-05,
+      "loss": 3.6174,
+      "step": 752640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.766947967161154e-05,
+      "loss": 3.6501,
+      "step": 753152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.766111010290475e-05,
+      "loss": 3.6306,
+      "step": 753664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.765272415539423e-05,
+      "loss": 3.6386,
+      "step": 754176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.764433820788371e-05,
+      "loss": 3.6227,
+      "step": 754688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.763595226037319e-05,
+      "loss": 3.6381,
+      "step": 755200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.762756631286267e-05,
+      "loss": 3.6303,
+      "step": 755712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.761918036535215e-05,
+      "loss": 3.6341,
+      "step": 756224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.761079441784163e-05,
+      "loss": 3.6275,
+      "step": 756736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.760240847033111e-05,
+      "loss": 3.63,
+      "step": 757248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7594038901624326e-05,
+      "loss": 3.6323,
+      "step": 757760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7585669332917535e-05,
+      "loss": 3.6531,
+      "step": 758272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7577283385407015e-05,
+      "loss": 3.6234,
+      "step": 758784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7568897437896495e-05,
+      "loss": 3.638,
+      "step": 759296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.756051149038597e-05,
+      "loss": 3.6347,
+      "step": 759808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.755212554287545e-05,
+      "loss": 3.6313,
+      "step": 760320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.754373959536493e-05,
+      "loss": 3.6266,
+      "step": 760832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.753535364785441e-05,
+      "loss": 3.6366,
+      "step": 761344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.752698407914762e-05,
+      "loss": 3.633,
+      "step": 761856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7518598131637104e-05,
+      "loss": 3.6252,
+      "step": 762368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7510212184126584e-05,
+      "loss": 3.6329,
+      "step": 762880
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.85237717628479,
+      "eval_runtime": 308.986,
+      "eval_samples_per_second": 1234.978,
+      "eval_steps_per_second": 38.594,
+      "step": 763200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.750184261541979e-05,
+      "loss": 3.6237,
+      "step": 763392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.749345666790927e-05,
+      "loss": 3.6138,
+      "step": 763904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.748507072039875e-05,
+      "loss": 3.6286,
+      "step": 764416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.747668477288823e-05,
+      "loss": 3.6428,
+      "step": 764928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.746829882537771e-05,
+      "loss": 3.6407,
+      "step": 765440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.745991287786719e-05,
+      "loss": 3.6222,
+      "step": 765952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.745152693035667e-05,
+      "loss": 3.633,
+      "step": 766464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.744314098284615e-05,
+      "loss": 3.623,
+      "step": 766976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.743477141413936e-05,
+      "loss": 3.624,
+      "step": 767488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.742638546662884e-05,
+      "loss": 3.6355,
+      "step": 768000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.741799951911832e-05,
+      "loss": 3.6253,
+      "step": 768512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.740962995041154e-05,
+      "loss": 3.6318,
+      "step": 769024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.740124400290102e-05,
+      "loss": 3.6287,
+      "step": 769536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.73928580553905e-05,
+      "loss": 3.6194,
+      "step": 770048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.738447210787998e-05,
+      "loss": 3.6206,
+      "step": 770560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7376102539173186e-05,
+      "loss": 3.6131,
+      "step": 771072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7367716591662666e-05,
+      "loss": 3.6245,
+      "step": 771584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7359330644152146e-05,
+      "loss": 3.6269,
+      "step": 772096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7350944696641626e-05,
+      "loss": 3.6153,
+      "step": 772608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7342558749131106e-05,
+      "loss": 3.6327,
+      "step": 773120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7334172801620586e-05,
+      "loss": 3.6374,
+      "step": 773632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7325803232913795e-05,
+      "loss": 3.6302,
+      "step": 774144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7317417285403275e-05,
+      "loss": 3.6306,
+      "step": 774656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7309031337892755e-05,
+      "loss": 3.6225,
+      "step": 775168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7300645390382235e-05,
+      "loss": 3.6292,
+      "step": 775680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.729225944287172e-05,
+      "loss": 3.6246,
+      "step": 776192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.728388987416493e-05,
+      "loss": 3.6185,
+      "step": 776704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.727550392665441e-05,
+      "loss": 3.6194,
+      "step": 777216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.726711797914389e-05,
+      "loss": 3.625,
+      "step": 777728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.725873203163337e-05,
+      "loss": 3.6193,
+      "step": 778240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.725034608412285e-05,
+      "loss": 3.6208,
+      "step": 778752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.724196013661233e-05,
+      "loss": 3.629,
+      "step": 779264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.723359056790554e-05,
+      "loss": 3.6294,
+      "step": 779776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.722520462039502e-05,
+      "loss": 3.6311,
+      "step": 780288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.72168186728845e-05,
+      "loss": 3.615,
+      "step": 780800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.720843272537398e-05,
+      "loss": 3.6252,
+      "step": 781312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.720006315666719e-05,
+      "loss": 3.6296,
+      "step": 781824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7191677209156676e-05,
+      "loss": 3.6098,
+      "step": 782336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7183291261646155e-05,
+      "loss": 3.623,
+      "step": 782848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.717490531413563e-05,
+      "loss": 3.6124,
+      "step": 783360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.716651936662511e-05,
+      "loss": 3.6233,
+      "step": 783872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.715813341911459e-05,
+      "loss": 3.6198,
+      "step": 784384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.714974747160407e-05,
+      "loss": 3.6228,
+      "step": 784896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.714136152409355e-05,
+      "loss": 3.6263,
+      "step": 785408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.713299195538676e-05,
+      "loss": 3.6275,
+      "step": 785920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.712460600787624e-05,
+      "loss": 3.6277,
+      "step": 786432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.711622006036572e-05,
+      "loss": 3.6297,
+      "step": 786944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.71078341128552e-05,
+      "loss": 3.6186,
+      "step": 787456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.709946454414841e-05,
+      "loss": 3.5921,
+      "step": 787968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.709109497544163e-05,
+      "loss": 3.6313,
+      "step": 788480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.70827090279311e-05,
+      "loss": 3.6233,
+      "step": 788992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.707432308042058e-05,
+      "loss": 3.6259,
+      "step": 789504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.706593713291006e-05,
+      "loss": 3.6187,
+      "step": 790016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.705755118539954e-05,
+      "loss": 3.6146,
+      "step": 790528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.704916523788902e-05,
+      "loss": 3.604,
+      "step": 791040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.70407792903785e-05,
+      "loss": 3.6132,
+      "step": 791552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.703239334286798e-05,
+      "loss": 3.6171,
+      "step": 792064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.702404015296493e-05,
+      "loss": 3.6019,
+      "step": 792576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.70156542054544e-05,
+      "loss": 3.6303,
+      "step": 793088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.700726825794388e-05,
+      "loss": 3.6163,
+      "step": 793600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.699888231043337e-05,
+      "loss": 3.6002,
+      "step": 794112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.699049636292285e-05,
+      "loss": 3.6254,
+      "step": 794624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6982126794216056e-05,
+      "loss": 3.6042,
+      "step": 795136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6973740846705536e-05,
+      "loss": 3.6067,
+      "step": 795648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6965354899195016e-05,
+      "loss": 3.6194,
+      "step": 796160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6956968951684496e-05,
+      "loss": 3.6157,
+      "step": 796672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6948599382977705e-05,
+      "loss": 3.6075,
+      "step": 797184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6940213435467185e-05,
+      "loss": 3.6083,
+      "step": 797696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6931827487956665e-05,
+      "loss": 3.6077,
+      "step": 798208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6923441540446145e-05,
+      "loss": 3.6158,
+      "step": 798720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6915055592935625e-05,
+      "loss": 3.6207,
+      "step": 799232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6906669645425105e-05,
+      "loss": 3.6124,
+      "step": 799744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6898283697914585e-05,
+      "loss": 3.6167,
+      "step": 800256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6889897750404065e-05,
+      "loss": 3.6127,
+      "step": 800768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.688152818169728e-05,
+      "loss": 3.6329,
+      "step": 801280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.687314223418676e-05,
+      "loss": 3.6072,
+      "step": 801792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.686475628667624e-05,
+      "loss": 3.6216,
+      "step": 802304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.685637033916572e-05,
+      "loss": 3.6181,
+      "step": 802816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.684800077045893e-05,
+      "loss": 3.6026,
+      "step": 803328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.683963120175214e-05,
+      "loss": 3.6161,
+      "step": 803840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.683124525424162e-05,
+      "loss": 3.6204,
+      "step": 804352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.68228593067311e-05,
+      "loss": 3.62,
+      "step": 804864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.681447335922058e-05,
+      "loss": 3.622,
+      "step": 805376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.680608741171006e-05,
+      "loss": 3.5945,
+      "step": 805888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.679770146419954e-05,
+      "loss": 3.598,
+      "step": 806400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.678931551668902e-05,
+      "loss": 3.6108,
+      "step": 806912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6780945947982234e-05,
+      "loss": 3.6206,
+      "step": 807424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6772560000471714e-05,
+      "loss": 3.6042,
+      "step": 807936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6764174052961194e-05,
+      "loss": 3.6106,
+      "step": 808448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6755788105450674e-05,
+      "loss": 3.6071,
+      "step": 808960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.674743491554761e-05,
+      "loss": 3.6193,
+      "step": 809472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.673904896803709e-05,
+      "loss": 3.6096,
+      "step": 809984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.673066302052657e-05,
+      "loss": 3.6098,
+      "step": 810496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.672227707301605e-05,
+      "loss": 3.6144,
+      "step": 811008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.671389112550553e-05,
+      "loss": 3.622,
+      "step": 811520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.670552155679874e-05,
+      "loss": 3.6195,
+      "step": 812032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.669713560928822e-05,
+      "loss": 3.6012,
+      "step": 812544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.668874966177771e-05,
+      "loss": 3.6058,
+      "step": 813056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.668036371426719e-05,
+      "loss": 3.6099,
+      "step": 813568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.667197776675667e-05,
+      "loss": 3.6201,
+      "step": 814080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.666359181924615e-05,
+      "loss": 3.6149,
+      "step": 814592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.665522225053936e-05,
+      "loss": 3.6141,
+      "step": 815104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.664683630302884e-05,
+      "loss": 3.6071,
+      "step": 815616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.663845035551832e-05,
+      "loss": 3.6009,
+      "step": 816128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66300644080078e-05,
+      "loss": 3.6102,
+      "step": 816640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.662167846049728e-05,
+      "loss": 3.6065,
+      "step": 817152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6613308891790486e-05,
+      "loss": 3.603,
+      "step": 817664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6604922944279966e-05,
+      "loss": 3.6192,
+      "step": 818176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6596536996769446e-05,
+      "loss": 3.6044,
+      "step": 818688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6588151049258926e-05,
+      "loss": 3.6056,
+      "step": 819200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6579765101748406e-05,
+      "loss": 3.6027,
+      "step": 819712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.657139553304162e-05,
+      "loss": 3.6013,
+      "step": 820224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.65630095855311e-05,
+      "loss": 3.619,
+      "step": 820736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.655462363802058e-05,
+      "loss": 3.6113,
+      "step": 821248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.654623769051006e-05,
+      "loss": 3.6177,
+      "step": 821760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6537851742999535e-05,
+      "loss": 3.6201,
+      "step": 822272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6529465795489015e-05,
+      "loss": 3.6175,
+      "step": 822784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6521079847978494e-05,
+      "loss": 3.6089,
+      "step": 823296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6512710279271704e-05,
+      "loss": 3.6133,
+      "step": 823808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6504324331761184e-05,
+      "loss": 3.6087,
+      "step": 824320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6495938384250663e-05,
+      "loss": 3.6151,
+      "step": 824832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6487552436740143e-05,
+      "loss": 3.6061,
+      "step": 825344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.647916648922963e-05,
+      "loss": 3.613,
+      "step": 825856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.647078054171911e-05,
+      "loss": 3.6017,
+      "step": 826368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.646239459420859e-05,
+      "loss": 3.612,
+      "step": 826880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.64540250255018e-05,
+      "loss": 3.6095,
+      "step": 827392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.644563907799128e-05,
+      "loss": 3.6044,
+      "step": 827904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.643725313048076e-05,
+      "loss": 3.6102,
+      "step": 828416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.642886718297024e-05,
+      "loss": 3.5955,
+      "step": 828928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.642048123545972e-05,
+      "loss": 3.6307,
+      "step": 829440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.641211166675293e-05,
+      "loss": 3.6077,
+      "step": 829952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.640372571924241e-05,
+      "loss": 3.6212,
+      "step": 830464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.639533977173189e-05,
+      "loss": 3.6022,
+      "step": 830976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.638695382422137e-05,
+      "loss": 3.6131,
+      "step": 831488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.637856787671085e-05,
+      "loss": 3.6074,
+      "step": 832000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.637018192920033e-05,
+      "loss": 3.6159,
+      "step": 832512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6361812360493544e-05,
+      "loss": 3.606,
+      "step": 833024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6353426412983024e-05,
+      "loss": 3.6085,
+      "step": 833536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6345040465472504e-05,
+      "loss": 3.613,
+      "step": 834048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6336654517961984e-05,
+      "loss": 3.6348,
+      "step": 834560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.632828494925519e-05,
+      "loss": 3.5988,
+      "step": 835072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.631989900174467e-05,
+      "loss": 3.6207,
+      "step": 835584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.631151305423415e-05,
+      "loss": 3.6157,
+      "step": 836096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.630312710672363e-05,
+      "loss": 3.6064,
+      "step": 836608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.629474115921311e-05,
+      "loss": 3.6065,
+      "step": 837120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.628635521170259e-05,
+      "loss": 3.6153,
+      "step": 837632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.62779856429958e-05,
+      "loss": 3.6123,
+      "step": 838144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.626959969548528e-05,
+      "loss": 3.6085,
+      "step": 838656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.626121374797477e-05,
+      "loss": 3.6085,
+      "step": 839168
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.851473808288574,
+      "eval_runtime": 306.4172,
+      "eval_samples_per_second": 1245.332,
+      "eval_steps_per_second": 38.918,
+      "step": 839520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.625282780046425e-05,
+      "loss": 3.605,
+      "step": 839680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.624444185295372e-05,
+      "loss": 3.5949,
+      "step": 840192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.623607228424694e-05,
+      "loss": 3.6072,
+      "step": 840704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.622768633673642e-05,
+      "loss": 3.6195,
+      "step": 841216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.621930038922589e-05,
+      "loss": 3.6212,
+      "step": 841728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.621091444171537e-05,
+      "loss": 3.6015,
+      "step": 842240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.620252849420485e-05,
+      "loss": 3.6126,
+      "step": 842752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.619414254669433e-05,
+      "loss": 3.6023,
+      "step": 843264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.618575659918381e-05,
+      "loss": 3.6025,
+      "step": 843776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.617738703047702e-05,
+      "loss": 3.6104,
+      "step": 844288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6169001082966506e-05,
+      "loss": 3.6111,
+      "step": 844800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6160615135455986e-05,
+      "loss": 3.6113,
+      "step": 845312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6152229187945466e-05,
+      "loss": 3.6056,
+      "step": 845824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6143843240434946e-05,
+      "loss": 3.6004,
+      "step": 846336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6135457292924426e-05,
+      "loss": 3.5999,
+      "step": 846848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6127087724217635e-05,
+      "loss": 3.5865,
+      "step": 847360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6118701776707115e-05,
+      "loss": 3.606,
+      "step": 847872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6110315829196595e-05,
+      "loss": 3.6036,
+      "step": 848384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6101929881686075e-05,
+      "loss": 3.596,
+      "step": 848896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6093543934175555e-05,
+      "loss": 3.6095,
+      "step": 849408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6085174365468764e-05,
+      "loss": 3.6189,
+      "step": 849920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6076788417958244e-05,
+      "loss": 3.6083,
+      "step": 850432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6068402470447724e-05,
+      "loss": 3.6102,
+      "step": 850944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6060016522937204e-05,
+      "loss": 3.6015,
+      "step": 851456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6051630575426684e-05,
+      "loss": 3.6092,
+      "step": 851968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.604324462791617e-05,
+      "loss": 3.6046,
+      "step": 852480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.603487505920938e-05,
+      "loss": 3.5981,
+      "step": 852992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.602648911169886e-05,
+      "loss": 3.5987,
+      "step": 853504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.601810316418834e-05,
+      "loss": 3.6031,
+      "step": 854016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.600971721667782e-05,
+      "loss": 3.5985,
+      "step": 854528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.600134764797103e-05,
+      "loss": 3.6037,
+      "step": 855040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.599296170046051e-05,
+      "loss": 3.6064,
+      "step": 855552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.598457575294999e-05,
+      "loss": 3.6053,
+      "step": 856064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.597618980543947e-05,
+      "loss": 3.6127,
+      "step": 856576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.596782023673268e-05,
+      "loss": 3.5955,
+      "step": 857088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.595943428922216e-05,
+      "loss": 3.605,
+      "step": 857600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.595104834171164e-05,
+      "loss": 3.6088,
+      "step": 858112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5942662394201124e-05,
+      "loss": 3.5896,
+      "step": 858624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.593429282549433e-05,
+      "loss": 3.604,
+      "step": 859136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.592590687798381e-05,
+      "loss": 3.5942,
+      "step": 859648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.591752093047329e-05,
+      "loss": 3.5971,
+      "step": 860160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.590913498296277e-05,
+      "loss": 3.6003,
+      "step": 860672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.590074903545225e-05,
+      "loss": 3.6033,
+      "step": 861184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5892363087941726e-05,
+      "loss": 3.6085,
+      "step": 861696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5883977140431206e-05,
+      "loss": 3.6054,
+      "step": 862208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5875591192920686e-05,
+      "loss": 3.6081,
+      "step": 862720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.58672216242139e-05,
+      "loss": 3.608,
+      "step": 863232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.585885205550711e-05,
+      "loss": 3.599,
+      "step": 863744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.585046610799659e-05,
+      "loss": 3.5745,
+      "step": 864256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.584208016048608e-05,
+      "loss": 3.612,
+      "step": 864768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.583369421297556e-05,
+      "loss": 3.5999,
+      "step": 865280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.582532464426877e-05,
+      "loss": 3.6097,
+      "step": 865792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.581693869675825e-05,
+      "loss": 3.597,
+      "step": 866304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.580855274924773e-05,
+      "loss": 3.5906,
+      "step": 866816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.58001668017372e-05,
+      "loss": 3.5912,
+      "step": 867328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.579178085422668e-05,
+      "loss": 3.5861,
+      "step": 867840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5783411285519896e-05,
+      "loss": 3.5991,
+      "step": 868352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5775025338009376e-05,
+      "loss": 3.5798,
+      "step": 868864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.576663939049885e-05,
+      "loss": 3.6158,
+      "step": 869376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.575825344298833e-05,
+      "loss": 3.5973,
+      "step": 869888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5749883874281545e-05,
+      "loss": 3.5791,
+      "step": 870400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.574149792677103e-05,
+      "loss": 3.6056,
+      "step": 870912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5733111979260505e-05,
+      "loss": 3.5855,
+      "step": 871424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5724726031749985e-05,
+      "loss": 3.5888,
+      "step": 871936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.57163564630432e-05,
+      "loss": 3.5955,
+      "step": 872448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5707970515532674e-05,
+      "loss": 3.5941,
+      "step": 872960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5699584568022154e-05,
+      "loss": 3.5866,
+      "step": 873472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5691198620511634e-05,
+      "loss": 3.5921,
+      "step": 873984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.568282905180485e-05,
+      "loss": 3.5837,
+      "step": 874496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.567444310429432e-05,
+      "loss": 3.5968,
+      "step": 875008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.56660571567838e-05,
+      "loss": 3.602,
+      "step": 875520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.565767120927328e-05,
+      "loss": 3.5916,
+      "step": 876032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.564928526176277e-05,
+      "loss": 3.5999,
+      "step": 876544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.564091569305598e-05,
+      "loss": 3.5885,
+      "step": 877056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.563252974554546e-05,
+      "loss": 3.6165,
+      "step": 877568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.562414379803494e-05,
+      "loss": 3.5837,
+      "step": 878080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.561575785052442e-05,
+      "loss": 3.6051,
+      "step": 878592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.56073719030139e-05,
+      "loss": 3.5937,
+      "step": 879104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.559900233430711e-05,
+      "loss": 3.5887,
+      "step": 879616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.559061638679659e-05,
+      "loss": 3.5942,
+      "step": 880128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.558223043928607e-05,
+      "loss": 3.6006,
+      "step": 880640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.557384449177555e-05,
+      "loss": 3.6007,
+      "step": 881152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5565474923068756e-05,
+      "loss": 3.5992,
+      "step": 881664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5557088975558236e-05,
+      "loss": 3.5772,
+      "step": 882176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.554870302804772e-05,
+      "loss": 3.5792,
+      "step": 882688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.55403170805372e-05,
+      "loss": 3.5911,
+      "step": 883200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.553194751183041e-05,
+      "loss": 3.6028,
+      "step": 883712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.552356156431989e-05,
+      "loss": 3.5842,
+      "step": 884224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.551517561680937e-05,
+      "loss": 3.5886,
+      "step": 884736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.550678966929885e-05,
+      "loss": 3.591,
+      "step": 885248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.549840372178833e-05,
+      "loss": 3.5928,
+      "step": 885760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.549001777427781e-05,
+      "loss": 3.5935,
+      "step": 886272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.548164820557102e-05,
+      "loss": 3.5903,
+      "step": 886784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.54732622580605e-05,
+      "loss": 3.5956,
+      "step": 887296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.546487631054998e-05,
+      "loss": 3.6002,
+      "step": 887808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.545649036303946e-05,
+      "loss": 3.6016,
+      "step": 888320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5448120794332677e-05,
+      "loss": 3.5805,
+      "step": 888832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5439734846822157e-05,
+      "loss": 3.5852,
+      "step": 889344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5431348899311636e-05,
+      "loss": 3.5903,
+      "step": 889856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5422962951801116e-05,
+      "loss": 3.597,
+      "step": 890368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5414593383094326e-05,
+      "loss": 3.5985,
+      "step": 890880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5406207435583805e-05,
+      "loss": 3.5919,
+      "step": 891392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5397821488073285e-05,
+      "loss": 3.5886,
+      "step": 891904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5389435540562765e-05,
+      "loss": 3.5824,
+      "step": 892416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5381065971855975e-05,
+      "loss": 3.589,
+      "step": 892928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5372680024345454e-05,
+      "loss": 3.5888,
+      "step": 893440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5364294076834934e-05,
+      "loss": 3.5853,
+      "step": 893952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5355908129324414e-05,
+      "loss": 3.5956,
+      "step": 894464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5347522181813894e-05,
+      "loss": 3.588,
+      "step": 894976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.533915261310711e-05,
+      "loss": 3.5849,
+      "step": 895488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.533076666559659e-05,
+      "loss": 3.5854,
+      "step": 896000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.532238071808607e-05,
+      "loss": 3.5805,
+      "step": 896512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.531399477057555e-05,
+      "loss": 3.603,
+      "step": 897024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.530562520186876e-05,
+      "loss": 3.5925,
+      "step": 897536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.529723925435824e-05,
+      "loss": 3.5972,
+      "step": 898048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.528885330684772e-05,
+      "loss": 3.5946,
+      "step": 898560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.52804673593372e-05,
+      "loss": 3.5995,
+      "step": 899072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.527208141182668e-05,
+      "loss": 3.5925,
+      "step": 899584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.526371184311989e-05,
+      "loss": 3.5864,
+      "step": 900096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.525532589560937e-05,
+      "loss": 3.5965,
+      "step": 900608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.524693994809885e-05,
+      "loss": 3.5894,
+      "step": 901120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.523855400058833e-05,
+      "loss": 3.5846,
+      "step": 901632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5230184431881544e-05,
+      "loss": 3.5962,
+      "step": 902144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5221798484371024e-05,
+      "loss": 3.5863,
+      "step": 902656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5213412536860504e-05,
+      "loss": 3.5897,
+      "step": 903168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5205026589349984e-05,
+      "loss": 3.5866,
+      "step": 903680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.519665702064319e-05,
+      "loss": 3.5878,
+      "step": 904192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.518827107313267e-05,
+      "loss": 3.5889,
+      "step": 904704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.517988512562215e-05,
+      "loss": 3.5816,
+      "step": 905216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.517149917811163e-05,
+      "loss": 3.6083,
+      "step": 905728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.516312960940484e-05,
+      "loss": 3.591,
+      "step": 906240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.515474366189432e-05,
+      "loss": 3.5994,
+      "step": 906752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.51463577143838e-05,
+      "loss": 3.5828,
+      "step": 907264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.513797176687328e-05,
+      "loss": 3.594,
+      "step": 907776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.512958581936276e-05,
+      "loss": 3.5905,
+      "step": 908288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.512119987185224e-05,
+      "loss": 3.5943,
+      "step": 908800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.511283030314546e-05,
+      "loss": 3.5896,
+      "step": 909312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.510444435563493e-05,
+      "loss": 3.586,
+      "step": 909824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.509605840812441e-05,
+      "loss": 3.5901,
+      "step": 910336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.508767246061389e-05,
+      "loss": 3.615,
+      "step": 910848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.50793028919071e-05,
+      "loss": 3.5845,
+      "step": 911360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.507091694439658e-05,
+      "loss": 3.5995,
+      "step": 911872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.506253099688606e-05,
+      "loss": 3.5976,
+      "step": 912384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.505414504937554e-05,
+      "loss": 3.5856,
+      "step": 912896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.504575910186502e-05,
+      "loss": 3.5929,
+      "step": 913408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5037389533158235e-05,
+      "loss": 3.5925,
+      "step": 913920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5029003585647715e-05,
+      "loss": 3.5911,
+      "step": 914432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5020617638137195e-05,
+      "loss": 3.5919,
+      "step": 914944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5012231690626675e-05,
+      "loss": 3.5871,
+      "step": 915456
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8496768474578857,
+      "eval_runtime": 306.2158,
+      "eval_samples_per_second": 1246.151,
+      "eval_steps_per_second": 38.943,
+      "step": 915840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5003862121919884e-05,
+      "loss": 3.6001,
+      "step": 915968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4995476174409364e-05,
+      "loss": 3.5706,
+      "step": 916480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4987090226898844e-05,
+      "loss": 3.5863,
+      "step": 916992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4978704279388324e-05,
+      "loss": 3.6028,
+      "step": 917504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4970318331877804e-05,
+      "loss": 3.5958,
+      "step": 918016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.496194876317101e-05,
+      "loss": 3.5873,
+      "step": 918528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.495356281566049e-05,
+      "loss": 3.5992,
+      "step": 919040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.494517686814997e-05,
+      "loss": 3.5797,
+      "step": 919552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.493679092063945e-05,
+      "loss": 3.5799,
+      "step": 920064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.492842135193267e-05,
+      "loss": 3.5956,
+      "step": 920576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.492003540442215e-05,
+      "loss": 3.5914,
+      "step": 921088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.491164945691163e-05,
+      "loss": 3.5891,
+      "step": 921600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.490326350940111e-05,
+      "loss": 3.5912,
+      "step": 922112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.489487756189059e-05,
+      "loss": 3.5779,
+      "step": 922624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.48865079931838e-05,
+      "loss": 3.5797,
+      "step": 923136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.487812204567328e-05,
+      "loss": 3.5707,
+      "step": 923648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.486973609816276e-05,
+      "loss": 3.5871,
+      "step": 924160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.486135015065224e-05,
+      "loss": 3.5858,
+      "step": 924672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.485296420314172e-05,
+      "loss": 3.5795,
+      "step": 925184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.48445782556312e-05,
+      "loss": 3.5851,
+      "step": 925696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.483620868692441e-05,
+      "loss": 3.6062,
+      "step": 926208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4827822739413893e-05,
+      "loss": 3.5845,
+      "step": 926720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.481943679190337e-05,
+      "loss": 3.5922,
+      "step": 927232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.481105084439285e-05,
+      "loss": 3.5829,
+      "step": 927744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.480268127568606e-05,
+      "loss": 3.5926,
+      "step": 928256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.479429532817554e-05,
+      "loss": 3.5826,
+      "step": 928768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.478590938066502e-05,
+      "loss": 3.583,
+      "step": 929280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.47775234331545e-05,
+      "loss": 3.5797,
+      "step": 929792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.476913748564398e-05,
+      "loss": 3.5848,
+      "step": 930304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.476076791693719e-05,
+      "loss": 3.577,
+      "step": 930816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.475238196942667e-05,
+      "loss": 3.5848,
+      "step": 931328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.474399602191615e-05,
+      "loss": 3.584,
+      "step": 931840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.473561007440563e-05,
+      "loss": 3.5915,
+      "step": 932352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.472724050569884e-05,
+      "loss": 3.5915,
+      "step": 932864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.471885455818833e-05,
+      "loss": 3.5804,
+      "step": 933376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.471046861067781e-05,
+      "loss": 3.5847,
+      "step": 933888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.470208266316729e-05,
+      "loss": 3.5905,
+      "step": 934400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4693713094460496e-05,
+      "loss": 3.5683,
+      "step": 934912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4685327146949976e-05,
+      "loss": 3.5852,
+      "step": 935424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4676941199439456e-05,
+      "loss": 3.575,
+      "step": 935936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4668555251928936e-05,
+      "loss": 3.5781,
+      "step": 936448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.466016930441841e-05,
+      "loss": 3.5837,
+      "step": 936960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4651799735711625e-05,
+      "loss": 3.5849,
+      "step": 937472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4643413788201105e-05,
+      "loss": 3.5878,
+      "step": 937984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4635027840690585e-05,
+      "loss": 3.5833,
+      "step": 938496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4626641893180065e-05,
+      "loss": 3.5899,
+      "step": 939008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.461827232447328e-05,
+      "loss": 3.5892,
+      "step": 939520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.460988637696276e-05,
+      "loss": 3.5816,
+      "step": 940032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.460150042945224e-05,
+      "loss": 3.5568,
+      "step": 940544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4593114481941714e-05,
+      "loss": 3.5885,
+      "step": 941056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.458474491323493e-05,
+      "loss": 3.5845,
+      "step": 941568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.457635896572441e-05,
+      "loss": 3.589,
+      "step": 942080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.456797301821388e-05,
+      "loss": 3.5813,
+      "step": 942592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.455958707070336e-05,
+      "loss": 3.5748,
+      "step": 943104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.455121750199658e-05,
+      "loss": 3.5675,
+      "step": 943616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.454283155448606e-05,
+      "loss": 3.572,
+      "step": 944128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.453444560697553e-05,
+      "loss": 3.5779,
+      "step": 944640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.452605965946502e-05,
+      "loss": 3.561,
+      "step": 945152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4517690090758234e-05,
+      "loss": 3.5983,
+      "step": 945664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4509304143247714e-05,
+      "loss": 3.5799,
+      "step": 946176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.450091819573719e-05,
+      "loss": 3.5618,
+      "step": 946688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.449253224822667e-05,
+      "loss": 3.5862,
+      "step": 947200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.448416267951988e-05,
+      "loss": 3.5688,
+      "step": 947712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4475776732009357e-05,
+      "loss": 3.5695,
+      "step": 948224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4467390784498836e-05,
+      "loss": 3.5766,
+      "step": 948736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4459004836988316e-05,
+      "loss": 3.581,
+      "step": 949248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.445063526828153e-05,
+      "loss": 3.5648,
+      "step": 949760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4442249320771005e-05,
+      "loss": 3.5797,
+      "step": 950272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4433863373260485e-05,
+      "loss": 3.5638,
+      "step": 950784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.442547742574997e-05,
+      "loss": 3.5769,
+      "step": 951296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.441710785704318e-05,
+      "loss": 3.5819,
+      "step": 951808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.440872190953266e-05,
+      "loss": 3.5761,
+      "step": 952320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.440033596202214e-05,
+      "loss": 3.5777,
+      "step": 952832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.439195001451162e-05,
+      "loss": 3.5742,
+      "step": 953344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.43835640670011e-05,
+      "loss": 3.5975,
+      "step": 953856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.437519449829431e-05,
+      "loss": 3.5644,
+      "step": 954368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.436680855078379e-05,
+      "loss": 3.587,
+      "step": 954880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.435842260327327e-05,
+      "loss": 3.574,
+      "step": 955392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.435003665576275e-05,
+      "loss": 3.5696,
+      "step": 955904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.434165070825223e-05,
+      "loss": 3.5774,
+      "step": 956416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.433328113954544e-05,
+      "loss": 3.5757,
+      "step": 956928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4324895192034926e-05,
+      "loss": 3.5883,
+      "step": 957440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4316509244524406e-05,
+      "loss": 3.5832,
+      "step": 957952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4308123297013886e-05,
+      "loss": 3.5583,
+      "step": 958464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4299753728307095e-05,
+      "loss": 3.5603,
+      "step": 958976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4291367780796575e-05,
+      "loss": 3.5665,
+      "step": 959488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4282981833286055e-05,
+      "loss": 3.5837,
+      "step": 960000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4274595885775535e-05,
+      "loss": 3.5704,
+      "step": 960512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4266226317068744e-05,
+      "loss": 3.5691,
+      "step": 961024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4257840369558224e-05,
+      "loss": 3.5777,
+      "step": 961536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4249454422047704e-05,
+      "loss": 3.5729,
+      "step": 962048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4241068474537184e-05,
+      "loss": 3.5749,
+      "step": 962560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.423269890583039e-05,
+      "loss": 3.5696,
+      "step": 963072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.422431295831988e-05,
+      "loss": 3.5759,
+      "step": 963584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.421592701080936e-05,
+      "loss": 3.5825,
+      "step": 964096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.420754106329884e-05,
+      "loss": 3.5855,
+      "step": 964608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.419917149459205e-05,
+      "loss": 3.5622,
+      "step": 965120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.419078554708153e-05,
+      "loss": 3.568,
+      "step": 965632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.418239959957101e-05,
+      "loss": 3.5743,
+      "step": 966144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.417401365206049e-05,
+      "loss": 3.5759,
+      "step": 966656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.416562770454997e-05,
+      "loss": 3.5845,
+      "step": 967168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.415725813584318e-05,
+      "loss": 3.5736,
+      "step": 967680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.414887218833266e-05,
+      "loss": 3.5695,
+      "step": 968192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.414048624082214e-05,
+      "loss": 3.5675,
+      "step": 968704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.413210029331162e-05,
+      "loss": 3.5664,
+      "step": 969216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.412373072460483e-05,
+      "loss": 3.5724,
+      "step": 969728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.411534477709431e-05,
+      "loss": 3.5709,
+      "step": 970240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.410695882958379e-05,
+      "loss": 3.5787,
+      "step": 970752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.409857288207327e-05,
+      "loss": 3.5692,
+      "step": 971264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.409020331336648e-05,
+      "loss": 3.5661,
+      "step": 971776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.408181736585596e-05,
+      "loss": 3.571,
+      "step": 972288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.407343141834544e-05,
+      "loss": 3.5601,
+      "step": 972800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.406504547083492e-05,
+      "loss": 3.5811,
+      "step": 973312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.405667590212813e-05,
+      "loss": 3.5784,
+      "step": 973824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.404828995461761e-05,
+      "loss": 3.5758,
+      "step": 974336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.403990400710709e-05,
+      "loss": 3.5824,
+      "step": 974848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.403151805959657e-05,
+      "loss": 3.58,
+      "step": 975360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.402314849088979e-05,
+      "loss": 3.5734,
+      "step": 975872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.401476254337927e-05,
+      "loss": 3.5689,
+      "step": 976384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.400637659586875e-05,
+      "loss": 3.5818,
+      "step": 976896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.399799064835823e-05,
+      "loss": 3.5716,
+      "step": 977408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3989621079651436e-05,
+      "loss": 3.5699,
+      "step": 977920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3981235132140916e-05,
+      "loss": 3.5763,
+      "step": 978432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3972849184630396e-05,
+      "loss": 3.5665,
+      "step": 978944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3964463237119876e-05,
+      "loss": 3.5714,
+      "step": 979456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3956093668413085e-05,
+      "loss": 3.5677,
+      "step": 979968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3947707720902565e-05,
+      "loss": 3.5699,
+      "step": 980480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3939321773392045e-05,
+      "loss": 3.571,
+      "step": 980992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3930935825881525e-05,
+      "loss": 3.5652,
+      "step": 981504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.392256625717474e-05,
+      "loss": 3.586,
+      "step": 982016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.391418030966422e-05,
+      "loss": 3.577,
+      "step": 982528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.39057943621537e-05,
+      "loss": 3.5801,
+      "step": 983040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.389740841464318e-05,
+      "loss": 3.5678,
+      "step": 983552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.388903884593639e-05,
+      "loss": 3.5713,
+      "step": 984064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.388065289842587e-05,
+      "loss": 3.5761,
+      "step": 984576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.387226695091535e-05,
+      "loss": 3.5766,
+      "step": 985088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.386388100340483e-05,
+      "loss": 3.5737,
+      "step": 985600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.385551143469804e-05,
+      "loss": 3.5662,
+      "step": 986112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.384712548718752e-05,
+      "loss": 3.5765,
+      "step": 986624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3838739539677e-05,
+      "loss": 3.5936,
+      "step": 987136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.383035359216648e-05,
+      "loss": 3.5727,
+      "step": 987648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.382198402345969e-05,
+      "loss": 3.5776,
+      "step": 988160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3813598075949174e-05,
+      "loss": 3.5797,
+      "step": 988672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3805212128438654e-05,
+      "loss": 3.5703,
+      "step": 989184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3796826180928134e-05,
+      "loss": 3.5759,
+      "step": 989696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3788440233417614e-05,
+      "loss": 3.5725,
+      "step": 990208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.378007066471082e-05,
+      "loss": 3.5743,
+      "step": 990720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.37716847172003e-05,
+      "loss": 3.5759,
+      "step": 991232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.376329876968978e-05,
+      "loss": 3.5675,
+      "step": 991744
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8501436710357666,
+      "eval_runtime": 304.2052,
+      "eval_samples_per_second": 1254.387,
+      "eval_steps_per_second": 39.201,
+      "step": 992160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3754912822179256e-05,
+      "loss": 3.5824,
+      "step": 992256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3746526874668736e-05,
+      "loss": 3.5542,
+      "step": 992768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3738140927158216e-05,
+      "loss": 3.5681,
+      "step": 993280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3729754979647696e-05,
+      "loss": 3.5888,
+      "step": 993792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.372138541094091e-05,
+      "loss": 3.5762,
+      "step": 994304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.371299946343039e-05,
+      "loss": 3.5704,
+      "step": 994816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.370461351591987e-05,
+      "loss": 3.5759,
+      "step": 995328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.369622756840935e-05,
+      "loss": 3.5666,
+      "step": 995840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.368785799970256e-05,
+      "loss": 3.5634,
+      "step": 996352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.367947205219204e-05,
+      "loss": 3.5794,
+      "step": 996864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.367108610468152e-05,
+      "loss": 3.5706,
+      "step": 997376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3662700157171e-05,
+      "loss": 3.5727,
+      "step": 997888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3654346967267946e-05,
+      "loss": 3.5745,
+      "step": 998400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3645961019757426e-05,
+      "loss": 3.5609,
+      "step": 998912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3637575072246906e-05,
+      "loss": 3.5617,
+      "step": 999424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.362918912473638e-05,
+      "loss": 3.5535,
+      "step": 999936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3620803177225866e-05,
+      "loss": 3.5701,
+      "step": 1000448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3612417229715346e-05,
+      "loss": 3.5666,
+      "step": 1000960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3604031282204826e-05,
+      "loss": 3.5645,
+      "step": 1001472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3595645334694305e-05,
+      "loss": 3.5663,
+      "step": 1001984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3587275765987515e-05,
+      "loss": 3.588,
+      "step": 1002496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3578889818476995e-05,
+      "loss": 3.5684,
+      "step": 1003008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3570503870966474e-05,
+      "loss": 3.5742,
+      "step": 1003520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3562117923455954e-05,
+      "loss": 3.5697,
+      "step": 1004032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3553748354749164e-05,
+      "loss": 3.579,
+      "step": 1004544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3545362407238643e-05,
+      "loss": 3.5614,
+      "step": 1005056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3536976459728123e-05,
+      "loss": 3.5641,
+      "step": 1005568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3528590512217603e-05,
+      "loss": 3.5686,
+      "step": 1006080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.352020456470708e-05,
+      "loss": 3.5631,
+      "step": 1006592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.35118349960003e-05,
+      "loss": 3.5623,
+      "step": 1007104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.350344904848978e-05,
+      "loss": 3.5667,
+      "step": 1007616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.349506310097926e-05,
+      "loss": 3.563,
+      "step": 1008128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.348667715346874e-05,
+      "loss": 3.5772,
+      "step": 1008640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.347830758476195e-05,
+      "loss": 3.5698,
+      "step": 1009152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.346992163725143e-05,
+      "loss": 3.5652,
+      "step": 1009664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.346153568974091e-05,
+      "loss": 3.5719,
+      "step": 1010176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.345314974223039e-05,
+      "loss": 3.5705,
+      "step": 1010688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.34447801735236e-05,
+      "loss": 3.5521,
+      "step": 1011200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.343639422601308e-05,
+      "loss": 3.567,
+      "step": 1011712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.342800827850256e-05,
+      "loss": 3.5586,
+      "step": 1012224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.341962233099204e-05,
+      "loss": 3.5593,
+      "step": 1012736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.341125276228525e-05,
+      "loss": 3.5668,
+      "step": 1013248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.340286681477473e-05,
+      "loss": 3.5725,
+      "step": 1013760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.339448086726421e-05,
+      "loss": 3.5698,
+      "step": 1014272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.338609491975369e-05,
+      "loss": 3.5678,
+      "step": 1014784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.33777253510469e-05,
+      "loss": 3.5709,
+      "step": 1015296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.336933940353638e-05,
+      "loss": 3.5706,
+      "step": 1015808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.336095345602586e-05,
+      "loss": 3.5644,
+      "step": 1016320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.335256750851534e-05,
+      "loss": 3.5427,
+      "step": 1016832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.334418156100482e-05,
+      "loss": 3.5676,
+      "step": 1017344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.333581199229803e-05,
+      "loss": 3.5713,
+      "step": 1017856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.332742604478751e-05,
+      "loss": 3.5688,
+      "step": 1018368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.331904009727699e-05,
+      "loss": 3.5625,
+      "step": 1018880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.331065414976647e-05,
+      "loss": 3.5588,
+      "step": 1019392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3302284581059687e-05,
+      "loss": 3.5503,
+      "step": 1019904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3293898633549166e-05,
+      "loss": 3.5507,
+      "step": 1020416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3285512686038646e-05,
+      "loss": 3.5644,
+      "step": 1020928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3277126738528126e-05,
+      "loss": 3.5479,
+      "step": 1021440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3268757169821335e-05,
+      "loss": 3.5729,
+      "step": 1021952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3260371222310815e-05,
+      "loss": 3.5672,
+      "step": 1022464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3251985274800295e-05,
+      "loss": 3.5495,
+      "step": 1022976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3243599327289775e-05,
+      "loss": 3.5648,
+      "step": 1023488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3235229758582984e-05,
+      "loss": 3.5485,
+      "step": 1024000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3226843811072464e-05,
+      "loss": 3.5551,
+      "step": 1024512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3218457863561944e-05,
+      "loss": 3.5551,
+      "step": 1025024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3210071916051424e-05,
+      "loss": 3.5659,
+      "step": 1025536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.320170234734464e-05,
+      "loss": 3.5493,
+      "step": 1026048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.319331639983412e-05,
+      "loss": 3.5609,
+      "step": 1026560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.31849304523236e-05,
+      "loss": 3.546,
+      "step": 1027072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.317654450481308e-05,
+      "loss": 3.5606,
+      "step": 1027584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.316817493610629e-05,
+      "loss": 3.565,
+      "step": 1028096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.315978898859577e-05,
+      "loss": 3.5616,
+      "step": 1028608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.315140304108525e-05,
+      "loss": 3.5604,
+      "step": 1029120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.314301709357473e-05,
+      "loss": 3.5601,
+      "step": 1029632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.313463114606421e-05,
+      "loss": 3.5781,
+      "step": 1030144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.312626157735742e-05,
+      "loss": 3.5479,
+      "step": 1030656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.31178756298469e-05,
+      "loss": 3.5716,
+      "step": 1031168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.310948968233638e-05,
+      "loss": 3.5594,
+      "step": 1031680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.310110373482586e-05,
+      "loss": 3.5506,
+      "step": 1032192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3092734166119074e-05,
+      "loss": 3.5583,
+      "step": 1032704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3084348218608554e-05,
+      "loss": 3.5612,
+      "step": 1033216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3075962271098034e-05,
+      "loss": 3.5689,
+      "step": 1033728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3067576323587514e-05,
+      "loss": 3.5681,
+      "step": 1034240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.305920675488072e-05,
+      "loss": 3.544,
+      "step": 1034752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.30508208073702e-05,
+      "loss": 3.5452,
+      "step": 1035264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.304243485985968e-05,
+      "loss": 3.5512,
+      "step": 1035776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.303404891234916e-05,
+      "loss": 3.5648,
+      "step": 1036288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.302567934364237e-05,
+      "loss": 3.5549,
+      "step": 1036800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.301729339613185e-05,
+      "loss": 3.5518,
+      "step": 1037312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.300890744862133e-05,
+      "loss": 3.5601,
+      "step": 1037824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.300052150111081e-05,
+      "loss": 3.555,
+      "step": 1038336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.299213555360029e-05,
+      "loss": 3.5619,
+      "step": 1038848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.298376598489351e-05,
+      "loss": 3.5507,
+      "step": 1039360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.297538003738299e-05,
+      "loss": 3.5629,
+      "step": 1039872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.296699408987246e-05,
+      "loss": 3.5605,
+      "step": 1040384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.295860814236194e-05,
+      "loss": 3.5704,
+      "step": 1040896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2950238573655156e-05,
+      "loss": 3.5437,
+      "step": 1041408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2941852626144636e-05,
+      "loss": 3.5507,
+      "step": 1041920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.293346667863411e-05,
+      "loss": 3.5632,
+      "step": 1042432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.292508073112359e-05,
+      "loss": 3.5577,
+      "step": 1042944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2916711162416805e-05,
+      "loss": 3.5665,
+      "step": 1043456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2908325214906285e-05,
+      "loss": 3.5569,
+      "step": 1043968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2899939267395765e-05,
+      "loss": 3.5508,
+      "step": 1044480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2891553319885245e-05,
+      "loss": 3.5511,
+      "step": 1044992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.288318375117846e-05,
+      "loss": 3.5527,
+      "step": 1045504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2874797803667934e-05,
+      "loss": 3.5534,
+      "step": 1046016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2866411856157414e-05,
+      "loss": 3.5522,
+      "step": 1046528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2858025908646894e-05,
+      "loss": 3.5597,
+      "step": 1047040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.284965633994011e-05,
+      "loss": 3.5527,
+      "step": 1047552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.284127039242958e-05,
+      "loss": 3.5501,
+      "step": 1048064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.283288444491906e-05,
+      "loss": 3.5536,
+      "step": 1048576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.282449849740854e-05,
+      "loss": 3.5427,
+      "step": 1049088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.281612892870176e-05,
+      "loss": 3.5677,
+      "step": 1049600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.280774298119124e-05,
+      "loss": 3.5594,
+      "step": 1050112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.279935703368072e-05,
+      "loss": 3.559,
+      "step": 1050624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.27909710861702e-05,
+      "loss": 3.5658,
+      "step": 1051136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.278260151746341e-05,
+      "loss": 3.5637,
+      "step": 1051648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.277421556995289e-05,
+      "loss": 3.5531,
+      "step": 1052160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.276582962244237e-05,
+      "loss": 3.5534,
+      "step": 1052672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.275744367493185e-05,
+      "loss": 3.5639,
+      "step": 1053184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.274907410622506e-05,
+      "loss": 3.5546,
+      "step": 1053696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.274068815871454e-05,
+      "loss": 3.5496,
+      "step": 1054208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.273230221120402e-05,
+      "loss": 3.5596,
+      "step": 1054720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.27239162636935e-05,
+      "loss": 3.5529,
+      "step": 1055232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.271554669498671e-05,
+      "loss": 3.5538,
+      "step": 1055744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.270716074747619e-05,
+      "loss": 3.5506,
+      "step": 1056256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.269877479996567e-05,
+      "loss": 3.5522,
+      "step": 1056768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.269038885245515e-05,
+      "loss": 3.5565,
+      "step": 1057280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.268201928374836e-05,
+      "loss": 3.5498,
+      "step": 1057792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.267363333623784e-05,
+      "loss": 3.5639,
+      "step": 1058304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.266524738872732e-05,
+      "loss": 3.5657,
+      "step": 1058816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.26568614412168e-05,
+      "loss": 3.5615,
+      "step": 1059328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.264847549370628e-05,
+      "loss": 3.5508,
+      "step": 1059840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.264010592499949e-05,
+      "loss": 3.5585,
+      "step": 1060352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.263171997748897e-05,
+      "loss": 3.5582,
+      "step": 1060864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.262333402997845e-05,
+      "loss": 3.5582,
+      "step": 1061376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.261494808246793e-05,
+      "loss": 3.5536,
+      "step": 1061888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2606578513761146e-05,
+      "loss": 3.55,
+      "step": 1062400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2598192566250626e-05,
+      "loss": 3.5625,
+      "step": 1062912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2589806618740106e-05,
+      "loss": 3.5753,
+      "step": 1063424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2581420671229586e-05,
+      "loss": 3.5534,
+      "step": 1063936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2573051102522795e-05,
+      "loss": 3.5635,
+      "step": 1064448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2564665155012275e-05,
+      "loss": 3.5647,
+      "step": 1064960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2556279207501755e-05,
+      "loss": 3.5552,
+      "step": 1065472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2547893259991235e-05,
+      "loss": 3.5593,
+      "step": 1065984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2539523691284444e-05,
+      "loss": 3.5582,
+      "step": 1066496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2531137743773924e-05,
+      "loss": 3.5556,
+      "step": 1067008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2522751796263404e-05,
+      "loss": 3.5658,
+      "step": 1067520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2514365848752884e-05,
+      "loss": 3.5461,
+      "step": 1068032
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8501906394958496,
+      "eval_runtime": 306.2388,
+      "eval_samples_per_second": 1246.057,
+      "eval_steps_per_second": 38.94,
+      "step": 1068480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.25059962800461e-05,
+      "loss": 3.5566,
+      "step": 1068544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.249761033253558e-05,
+      "loss": 3.5402,
+      "step": 1069056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.248922438502506e-05,
+      "loss": 3.5525,
+      "step": 1069568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.248083843751454e-05,
+      "loss": 3.5683,
+      "step": 1070080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.247246886880775e-05,
+      "loss": 3.565,
+      "step": 1070592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.246408292129723e-05,
+      "loss": 3.5538,
+      "step": 1071104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.245569697378671e-05,
+      "loss": 3.5571,
+      "step": 1071616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.244731102627619e-05,
+      "loss": 3.5479,
+      "step": 1072128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.24389414575694e-05,
+      "loss": 3.5474,
+      "step": 1072640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.243055551005888e-05,
+      "loss": 3.5605,
+      "step": 1073152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.242216956254836e-05,
+      "loss": 3.5559,
+      "step": 1073664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.241378361503784e-05,
+      "loss": 3.5575,
+      "step": 1074176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2405414046331054e-05,
+      "loss": 3.5592,
+      "step": 1074688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2397028098820534e-05,
+      "loss": 3.5463,
+      "step": 1075200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2388642151310014e-05,
+      "loss": 3.5442,
+      "step": 1075712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2380256203799494e-05,
+      "loss": 3.5373,
+      "step": 1076224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.23718866350927e-05,
+      "loss": 3.5573,
+      "step": 1076736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.236350068758218e-05,
+      "loss": 3.5461,
+      "step": 1077248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.235511474007166e-05,
+      "loss": 3.55,
+      "step": 1077760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.234672879256114e-05,
+      "loss": 3.5512,
+      "step": 1078272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.233835922385435e-05,
+      "loss": 3.5682,
+      "step": 1078784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.232997327634383e-05,
+      "loss": 3.5573,
+      "step": 1079296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.232158732883331e-05,
+      "loss": 3.553,
+      "step": 1079808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.231320138132279e-05,
+      "loss": 3.5533,
+      "step": 1080320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.230483181261601e-05,
+      "loss": 3.5621,
+      "step": 1080832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.229644586510549e-05,
+      "loss": 3.5485,
+      "step": 1081344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.228805991759497e-05,
+      "loss": 3.5441,
+      "step": 1081856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.227967397008445e-05,
+      "loss": 3.5523,
+      "step": 1082368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.227128802257393e-05,
+      "loss": 3.5436,
+      "step": 1082880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2262918453867136e-05,
+      "loss": 3.548,
+      "step": 1083392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2254532506356616e-05,
+      "loss": 3.5507,
+      "step": 1083904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2246146558846096e-05,
+      "loss": 3.5518,
+      "step": 1084416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2237760611335576e-05,
+      "loss": 3.5559,
+      "step": 1084928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2229391042628785e-05,
+      "loss": 3.5564,
+      "step": 1085440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2221005095118265e-05,
+      "loss": 3.5488,
+      "step": 1085952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2212619147607745e-05,
+      "loss": 3.5546,
+      "step": 1086464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2204233200097225e-05,
+      "loss": 3.5568,
+      "step": 1086976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.219586363139044e-05,
+      "loss": 3.534,
+      "step": 1087488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.218747768387992e-05,
+      "loss": 3.5561,
+      "step": 1088000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.21790917363694e-05,
+      "loss": 3.5382,
+      "step": 1088512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.217070578885888e-05,
+      "loss": 3.5444,
+      "step": 1089024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.216233622015209e-05,
+      "loss": 3.5475,
+      "step": 1089536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.215395027264157e-05,
+      "loss": 3.5603,
+      "step": 1090048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.214556432513105e-05,
+      "loss": 3.5521,
+      "step": 1090560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.213717837762053e-05,
+      "loss": 3.5476,
+      "step": 1091072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.212879243011001e-05,
+      "loss": 3.5598,
+      "step": 1091584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.212042286140322e-05,
+      "loss": 3.5522,
+      "step": 1092096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.21120369138927e-05,
+      "loss": 3.5522,
+      "step": 1092608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.210365096638218e-05,
+      "loss": 3.5248,
+      "step": 1093120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.209526501887166e-05,
+      "loss": 3.5492,
+      "step": 1093632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2086895450164875e-05,
+      "loss": 3.5564,
+      "step": 1094144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2078509502654355e-05,
+      "loss": 3.5531,
+      "step": 1094656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2070123555143834e-05,
+      "loss": 3.548,
+      "step": 1095168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.206173760763331e-05,
+      "loss": 3.5404,
+      "step": 1095680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2053368038926524e-05,
+      "loss": 3.544,
+      "step": 1096192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2044982091416003e-05,
+      "loss": 3.5306,
+      "step": 1096704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2036596143905483e-05,
+      "loss": 3.5546,
+      "step": 1097216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2028210196394957e-05,
+      "loss": 3.5301,
+      "step": 1097728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.201984062768817e-05,
+      "loss": 3.553,
+      "step": 1098240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.201145468017765e-05,
+      "loss": 3.5496,
+      "step": 1098752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.200306873266713e-05,
+      "loss": 3.5352,
+      "step": 1099264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.199468278515661e-05,
+      "loss": 3.5488,
+      "step": 1099776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.198631321644983e-05,
+      "loss": 3.5321,
+      "step": 1100288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.197792726893931e-05,
+      "loss": 3.5392,
+      "step": 1100800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.196954132142878e-05,
+      "loss": 3.5369,
+      "step": 1101312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.196115537391826e-05,
+      "loss": 3.5532,
+      "step": 1101824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.195276942640774e-05,
+      "loss": 3.5304,
+      "step": 1102336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.194439985770096e-05,
+      "loss": 3.5434,
+      "step": 1102848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.193601391019043e-05,
+      "loss": 3.5304,
+      "step": 1103360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.192762796267991e-05,
+      "loss": 3.5433,
+      "step": 1103872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.191924201516939e-05,
+      "loss": 3.5519,
+      "step": 1104384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1910872446462606e-05,
+      "loss": 3.5487,
+      "step": 1104896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1902486498952086e-05,
+      "loss": 3.542,
+      "step": 1105408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1894100551441566e-05,
+      "loss": 3.5452,
+      "step": 1105920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1885714603931046e-05,
+      "loss": 3.5593,
+      "step": 1106432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1877345035224255e-05,
+      "loss": 3.5341,
+      "step": 1106944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1868959087713735e-05,
+      "loss": 3.556,
+      "step": 1107456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1860573140203215e-05,
+      "loss": 3.5412,
+      "step": 1107968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1852187192692695e-05,
+      "loss": 3.5393,
+      "step": 1108480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1843817623985904e-05,
+      "loss": 3.5405,
+      "step": 1108992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1835431676475384e-05,
+      "loss": 3.5442,
+      "step": 1109504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1827045728964864e-05,
+      "loss": 3.5554,
+      "step": 1110016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1818659781454344e-05,
+      "loss": 3.549,
+      "step": 1110528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.181029021274756e-05,
+      "loss": 3.5375,
+      "step": 1111040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.180190426523704e-05,
+      "loss": 3.5234,
+      "step": 1111552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.179351831772652e-05,
+      "loss": 3.5337,
+      "step": 1112064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1785132370216e-05,
+      "loss": 3.5496,
+      "step": 1112576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.177674642270548e-05,
+      "loss": 3.5398,
+      "step": 1113088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.176837685399869e-05,
+      "loss": 3.5366,
+      "step": 1113600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.175999090648817e-05,
+      "loss": 3.5455,
+      "step": 1114112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.175160495897765e-05,
+      "loss": 3.5413,
+      "step": 1114624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.174321901146713e-05,
+      "loss": 3.5432,
+      "step": 1115136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.173484944276034e-05,
+      "loss": 3.5358,
+      "step": 1115648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.172646349524982e-05,
+      "loss": 3.5482,
+      "step": 1116160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.17180775477393e-05,
+      "loss": 3.5421,
+      "step": 1116672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.170969160022878e-05,
+      "loss": 3.553,
+      "step": 1117184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1701322031521993e-05,
+      "loss": 3.5322,
+      "step": 1117696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.169293608401147e-05,
+      "loss": 3.5354,
+      "step": 1118208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.168455013650095e-05,
+      "loss": 3.5459,
+      "step": 1118720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.167616418899043e-05,
+      "loss": 3.5411,
+      "step": 1119232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.166779462028364e-05,
+      "loss": 3.5493,
+      "step": 1119744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.165940867277312e-05,
+      "loss": 3.5429,
+      "step": 1120256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.16510227252626e-05,
+      "loss": 3.5367,
+      "step": 1120768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.164263677775208e-05,
+      "loss": 3.5356,
+      "step": 1121280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.163426720904529e-05,
+      "loss": 3.5403,
+      "step": 1121792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.162588126153477e-05,
+      "loss": 3.5347,
+      "step": 1122304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.161749531402425e-05,
+      "loss": 3.5389,
+      "step": 1122816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.160910936651373e-05,
+      "loss": 3.544,
+      "step": 1123328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.160073979780695e-05,
+      "loss": 3.5381,
+      "step": 1123840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.159235385029643e-05,
+      "loss": 3.534,
+      "step": 1124352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.158396790278591e-05,
+      "loss": 3.5432,
+      "step": 1124864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.157558195527539e-05,
+      "loss": 3.527,
+      "step": 1125376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1567212386568596e-05,
+      "loss": 3.5491,
+      "step": 1125888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1558826439058076e-05,
+      "loss": 3.5411,
+      "step": 1126400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1550440491547556e-05,
+      "loss": 3.5449,
+      "step": 1126912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1542054544037036e-05,
+      "loss": 3.5489,
+      "step": 1127424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1533684975330245e-05,
+      "loss": 3.5488,
+      "step": 1127936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1525299027819725e-05,
+      "loss": 3.5397,
+      "step": 1128448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1516913080309205e-05,
+      "loss": 3.5389,
+      "step": 1128960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1508527132798685e-05,
+      "loss": 3.5458,
+      "step": 1129472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.15001575640919e-05,
+      "loss": 3.5396,
+      "step": 1129984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.149177161658138e-05,
+      "loss": 3.5315,
+      "step": 1130496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.148338566907086e-05,
+      "loss": 3.5494,
+      "step": 1131008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.147499972156034e-05,
+      "loss": 3.5365,
+      "step": 1131520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.146663015285355e-05,
+      "loss": 3.5405,
+      "step": 1132032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.145824420534303e-05,
+      "loss": 3.5331,
+      "step": 1132544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.144985825783251e-05,
+      "loss": 3.5342,
+      "step": 1133056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.144147231032199e-05,
+      "loss": 3.5463,
+      "step": 1133568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.143308636281147e-05,
+      "loss": 3.535,
+      "step": 1134080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.142471679410468e-05,
+      "loss": 3.544,
+      "step": 1134592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.141633084659416e-05,
+      "loss": 3.5519,
+      "step": 1135104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.140794489908364e-05,
+      "loss": 3.5438,
+      "step": 1135616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.139955895157312e-05,
+      "loss": 3.5375,
+      "step": 1136128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1391189382866334e-05,
+      "loss": 3.5434,
+      "step": 1136640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1382803435355814e-05,
+      "loss": 3.5454,
+      "step": 1137152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1374417487845294e-05,
+      "loss": 3.5407,
+      "step": 1137664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1366031540334774e-05,
+      "loss": 3.5353,
+      "step": 1138176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.135766197162798e-05,
+      "loss": 3.5385,
+      "step": 1138688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.134927602411746e-05,
+      "loss": 3.5533,
+      "step": 1139200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.134089007660694e-05,
+      "loss": 3.5567,
+      "step": 1139712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.133250412909642e-05,
+      "loss": 3.5422,
+      "step": 1140224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.132413456038963e-05,
+      "loss": 3.5432,
+      "step": 1140736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.131574861287911e-05,
+      "loss": 3.5482,
+      "step": 1141248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.130736266536859e-05,
+      "loss": 3.539,
+      "step": 1141760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.129897671785807e-05,
+      "loss": 3.5418,
+      "step": 1142272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.129060714915129e-05,
+      "loss": 3.5453,
+      "step": 1142784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.128222120164077e-05,
+      "loss": 3.5397,
+      "step": 1143296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.127383525413025e-05,
+      "loss": 3.5496,
+      "step": 1143808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.126544930661973e-05,
+      "loss": 3.5295,
+      "step": 1144320
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8507463932037354,
+      "eval_runtime": 307.069,
+      "eval_samples_per_second": 1242.688,
+      "eval_steps_per_second": 38.835,
+      "step": 1144800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125706335910921e-05,
+      "loss": 3.5227,
+      "step": 1144832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.124867741159869e-05,
+      "loss": 3.5284,
+      "step": 1145344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.124029146408816e-05,
+      "loss": 3.5337,
+      "step": 1145856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.123190551657764e-05,
+      "loss": 3.5545,
+      "step": 1146368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.122353594787086e-05,
+      "loss": 3.5519,
+      "step": 1146880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.121515000036033e-05,
+      "loss": 3.5368,
+      "step": 1147392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.120676405284981e-05,
+      "loss": 3.5405,
+      "step": 1147904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.119837810533929e-05,
+      "loss": 3.5353,
+      "step": 1148416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1190008536632506e-05,
+      "loss": 3.535,
+      "step": 1148928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1181622589121986e-05,
+      "loss": 3.5458,
+      "step": 1149440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1173236641611466e-05,
+      "loss": 3.5364,
+      "step": 1149952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1164850694100946e-05,
+      "loss": 3.5417,
+      "step": 1150464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1156481125394155e-05,
+      "loss": 3.5449,
+      "step": 1150976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.114811155668737e-05,
+      "loss": 3.5319,
+      "step": 1151488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.113972560917685e-05,
+      "loss": 3.5278,
+      "step": 1152000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.113133966166633e-05,
+      "loss": 3.5243,
+      "step": 1152512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1122953714155804e-05,
+      "loss": 3.5373,
+      "step": 1153024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1114567766645284e-05,
+      "loss": 3.5339,
+      "step": 1153536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1106181819134764e-05,
+      "loss": 3.5345,
+      "step": 1154048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1097795871624244e-05,
+      "loss": 3.5353,
+      "step": 1154560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.108940992411373e-05,
+      "loss": 3.5525,
+      "step": 1155072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.108104035540694e-05,
+      "loss": 3.5383,
+      "step": 1155584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.107265440789642e-05,
+      "loss": 3.5421,
+      "step": 1156096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.10642684603859e-05,
+      "loss": 3.5366,
+      "step": 1156608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.105588251287538e-05,
+      "loss": 3.5491,
+      "step": 1157120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.104751294416859e-05,
+      "loss": 3.529,
+      "step": 1157632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.103912699665807e-05,
+      "loss": 3.535,
+      "step": 1158144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.103074104914755e-05,
+      "loss": 3.5342,
+      "step": 1158656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.102235510163703e-05,
+      "loss": 3.5303,
+      "step": 1159168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.101398553293024e-05,
+      "loss": 3.5337,
+      "step": 1159680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.100559958541972e-05,
+      "loss": 3.5301,
+      "step": 1160192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.09972136379092e-05,
+      "loss": 3.5359,
+      "step": 1160704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0988827690398684e-05,
+      "loss": 3.5392,
+      "step": 1161216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0980441742888164e-05,
+      "loss": 3.5397,
+      "step": 1161728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.097207217418137e-05,
+      "loss": 3.5392,
+      "step": 1162240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.096368622667085e-05,
+      "loss": 3.5375,
+      "step": 1162752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.095530027916033e-05,
+      "loss": 3.5435,
+      "step": 1163264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.094691433164981e-05,
+      "loss": 3.5195,
+      "step": 1163776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.093854476294302e-05,
+      "loss": 3.544,
+      "step": 1164288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.09301588154325e-05,
+      "loss": 3.5189,
+      "step": 1164800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.092177286792198e-05,
+      "loss": 3.5325,
+      "step": 1165312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.091338692041146e-05,
+      "loss": 3.5315,
+      "step": 1165824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.090500097290094e-05,
+      "loss": 3.544,
+      "step": 1166336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.089663140419415e-05,
+      "loss": 3.5326,
+      "step": 1166848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.088824545668364e-05,
+      "loss": 3.5367,
+      "step": 1167360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.087985950917312e-05,
+      "loss": 3.5428,
+      "step": 1167872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.08714735616626e-05,
+      "loss": 3.5387,
+      "step": 1168384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.086310399295581e-05,
+      "loss": 3.5349,
+      "step": 1168896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0854718045445287e-05,
+      "loss": 3.512,
+      "step": 1169408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0846332097934767e-05,
+      "loss": 3.5319,
+      "step": 1169920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0837946150424246e-05,
+      "loss": 3.5417,
+      "step": 1170432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0829576581717456e-05,
+      "loss": 3.5371,
+      "step": 1170944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0821190634206936e-05,
+      "loss": 3.5316,
+      "step": 1171456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0812804686696416e-05,
+      "loss": 3.5235,
+      "step": 1171968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0804418739185895e-05,
+      "loss": 3.5275,
+      "step": 1172480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0796049170479105e-05,
+      "loss": 3.5158,
+      "step": 1172992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.078766322296859e-05,
+      "loss": 3.5409,
+      "step": 1173504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.077927727545807e-05,
+      "loss": 3.5152,
+      "step": 1174016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.077089132794755e-05,
+      "loss": 3.5365,
+      "step": 1174528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.076252175924076e-05,
+      "loss": 3.5342,
+      "step": 1175040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.075413581173024e-05,
+      "loss": 3.5236,
+      "step": 1175552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.074574986421972e-05,
+      "loss": 3.53,
+      "step": 1176064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.07373639167092e-05,
+      "loss": 3.5191,
+      "step": 1176576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.072899434800241e-05,
+      "loss": 3.5249,
+      "step": 1177088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.072060840049189e-05,
+      "loss": 3.5149,
+      "step": 1177600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.071222245298137e-05,
+      "loss": 3.5387,
+      "step": 1178112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.070383650547085e-05,
+      "loss": 3.523,
+      "step": 1178624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.069546693676406e-05,
+      "loss": 3.5225,
+      "step": 1179136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0687080989253545e-05,
+      "loss": 3.5184,
+      "step": 1179648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0678695041743025e-05,
+      "loss": 3.5298,
+      "step": 1180160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0670309094232505e-05,
+      "loss": 3.5399,
+      "step": 1180672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0661923146721985e-05,
+      "loss": 3.5305,
+      "step": 1181184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0653553578015194e-05,
+      "loss": 3.5235,
+      "step": 1181696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0645167630504674e-05,
+      "loss": 3.5346,
+      "step": 1182208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0636781682994154e-05,
+      "loss": 3.5448,
+      "step": 1182720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.062841211428736e-05,
+      "loss": 3.5184,
+      "step": 1183232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.062002616677684e-05,
+      "loss": 3.5396,
+      "step": 1183744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.061164021926632e-05,
+      "loss": 3.5288,
+      "step": 1184256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.06032542717558e-05,
+      "loss": 3.5237,
+      "step": 1184768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.059486832424528e-05,
+      "loss": 3.523,
+      "step": 1185280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.058649875553849e-05,
+      "loss": 3.5311,
+      "step": 1185792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.057811280802798e-05,
+      "loss": 3.5387,
+      "step": 1186304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.056972686051746e-05,
+      "loss": 3.5342,
+      "step": 1186816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.056134091300694e-05,
+      "loss": 3.5231,
+      "step": 1187328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.055297134430015e-05,
+      "loss": 3.5057,
+      "step": 1187840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.054458539678963e-05,
+      "loss": 3.5239,
+      "step": 1188352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.053619944927911e-05,
+      "loss": 3.5283,
+      "step": 1188864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.052781350176859e-05,
+      "loss": 3.5314,
+      "step": 1189376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0519443933061797e-05,
+      "loss": 3.5157,
+      "step": 1189888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0511057985551277e-05,
+      "loss": 3.5314,
+      "step": 1190400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.050267203804076e-05,
+      "loss": 3.527,
+      "step": 1190912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0494286090530233e-05,
+      "loss": 3.5278,
+      "step": 1191424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.048591652182345e-05,
+      "loss": 3.5201,
+      "step": 1191936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.047753057431293e-05,
+      "loss": 3.5323,
+      "step": 1192448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.046914462680241e-05,
+      "loss": 3.5248,
+      "step": 1192960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0460758679291885e-05,
+      "loss": 3.5396,
+      "step": 1193472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.04523891105851e-05,
+      "loss": 3.5235,
+      "step": 1193984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.044400316307458e-05,
+      "loss": 3.5185,
+      "step": 1194496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.043561721556406e-05,
+      "loss": 3.53,
+      "step": 1195008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0427231268053534e-05,
+      "loss": 3.5288,
+      "step": 1195520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.041886169934675e-05,
+      "loss": 3.5332,
+      "step": 1196032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.041047575183623e-05,
+      "loss": 3.5307,
+      "step": 1196544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0402089804325707e-05,
+      "loss": 3.521,
+      "step": 1197056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0393703856815187e-05,
+      "loss": 3.5203,
+      "step": 1197568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0385334288108403e-05,
+      "loss": 3.5266,
+      "step": 1198080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0376948340597883e-05,
+      "loss": 3.5139,
+      "step": 1198592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.036856239308736e-05,
+      "loss": 3.5274,
+      "step": 1199104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.036017644557684e-05,
+      "loss": 3.529,
+      "step": 1199616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0351806876870055e-05,
+      "loss": 3.5261,
+      "step": 1200128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0343420929359535e-05,
+      "loss": 3.5136,
+      "step": 1200640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0335034981849008e-05,
+      "loss": 3.5279,
+      "step": 1201152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0326649034338488e-05,
+      "loss": 3.5186,
+      "step": 1201664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0318263086827968e-05,
+      "loss": 3.5322,
+      "step": 1202176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.030989351812118e-05,
+      "loss": 3.5281,
+      "step": 1202688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.030150757061066e-05,
+      "loss": 3.5256,
+      "step": 1203200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.029312162310014e-05,
+      "loss": 3.5353,
+      "step": 1203712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.028473567558962e-05,
+      "loss": 3.5349,
+      "step": 1204224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.027636610688283e-05,
+      "loss": 3.5283,
+      "step": 1204736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0267980159372313e-05,
+      "loss": 3.5217,
+      "step": 1205248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0259594211861793e-05,
+      "loss": 3.5323,
+      "step": 1205760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0251208264351273e-05,
+      "loss": 3.5241,
+      "step": 1206272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0242838695644482e-05,
+      "loss": 3.5185,
+      "step": 1206784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0234452748133962e-05,
+      "loss": 3.5292,
+      "step": 1207296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.022606680062344e-05,
+      "loss": 3.5234,
+      "step": 1207808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.021768085311292e-05,
+      "loss": 3.5277,
+      "step": 1208320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0209311284406134e-05,
+      "loss": 3.5195,
+      "step": 1208832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0200925336895614e-05,
+      "loss": 3.5221,
+      "step": 1209344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0192539389385094e-05,
+      "loss": 3.5273,
+      "step": 1209856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0184153441874574e-05,
+      "loss": 3.5197,
+      "step": 1210368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0175783873167783e-05,
+      "loss": 3.5246,
+      "step": 1210880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0167397925657266e-05,
+      "loss": 3.5398,
+      "step": 1211392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0159011978146746e-05,
+      "loss": 3.5321,
+      "step": 1211904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0150626030636226e-05,
+      "loss": 3.5223,
+      "step": 1212416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0142256461929435e-05,
+      "loss": 3.5265,
+      "step": 1212928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0133870514418915e-05,
+      "loss": 3.5286,
+      "step": 1213440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0125484566908395e-05,
+      "loss": 3.5274,
+      "step": 1213952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0117098619397875e-05,
+      "loss": 3.5249,
+      "step": 1214464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0108729050691088e-05,
+      "loss": 3.5203,
+      "step": 1214976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0100343103180568e-05,
+      "loss": 3.5368,
+      "step": 1215488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0091957155670048e-05,
+      "loss": 3.5385,
+      "step": 1216000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0083571208159528e-05,
+      "loss": 3.5287,
+      "step": 1216512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0075201639452737e-05,
+      "loss": 3.5278,
+      "step": 1217024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0066815691942217e-05,
+      "loss": 3.5345,
+      "step": 1217536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.00584297444317e-05,
+      "loss": 3.5203,
+      "step": 1218048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.005004379692118e-05,
+      "loss": 3.5305,
+      "step": 1218560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.004167422821439e-05,
+      "loss": 3.5292,
+      "step": 1219072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.003328828070387e-05,
+      "loss": 3.525,
+      "step": 1219584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.002490233319335e-05,
+      "loss": 3.5349,
+      "step": 1220096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.001651638568283e-05,
+      "loss": 3.5167,
+      "step": 1220608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.000814681697604e-05,
+      "loss": 3.5249,
+      "step": 1221120
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8520734310150146,
+      "eval_runtime": 303.0188,
+      "eval_samples_per_second": 1259.298,
+      "eval_steps_per_second": 39.354,
+      "step": 1221120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.999976086946552e-05,
+      "loss": 3.5128,
+      "step": 1221632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9991374921955e-05,
+      "loss": 3.518,
+      "step": 1222144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.998298897444448e-05,
+      "loss": 3.5375,
+      "step": 1222656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.997461940573769e-05,
+      "loss": 3.5386,
+      "step": 1223168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.996623345822717e-05,
+      "loss": 3.5257,
+      "step": 1223680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9957847510716654e-05,
+      "loss": 3.5266,
+      "step": 1224192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9949461563206134e-05,
+      "loss": 3.5213,
+      "step": 1224704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9941075615695614e-05,
+      "loss": 3.5154,
+      "step": 1225216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9932706046988823e-05,
+      "loss": 3.5328,
+      "step": 1225728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9924320099478303e-05,
+      "loss": 3.5223,
+      "step": 1226240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9915934151967783e-05,
+      "loss": 3.5291,
+      "step": 1226752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9907548204457263e-05,
+      "loss": 3.5257,
+      "step": 1227264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9899195014554204e-05,
+      "loss": 3.5219,
+      "step": 1227776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9890809067043684e-05,
+      "loss": 3.5157,
+      "step": 1228288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9882423119533164e-05,
+      "loss": 3.5077,
+      "step": 1228800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9874037172022644e-05,
+      "loss": 3.5189,
+      "step": 1229312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9865651224512124e-05,
+      "loss": 3.519,
+      "step": 1229824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9857265277001607e-05,
+      "loss": 3.5216,
+      "step": 1230336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9848879329491087e-05,
+      "loss": 3.5179,
+      "step": 1230848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9840493381980567e-05,
+      "loss": 3.5421,
+      "step": 1231360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9832123813273776e-05,
+      "loss": 3.522,
+      "step": 1231872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9823737865763256e-05,
+      "loss": 3.5247,
+      "step": 1232384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9815351918252736e-05,
+      "loss": 3.5227,
+      "step": 1232896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9806965970742216e-05,
+      "loss": 3.5363,
+      "step": 1233408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.979859640203543e-05,
+      "loss": 3.5093,
+      "step": 1233920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.979021045452491e-05,
+      "loss": 3.52,
+      "step": 1234432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.978182450701439e-05,
+      "loss": 3.5192,
+      "step": 1234944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.977343855950387e-05,
+      "loss": 3.5214,
+      "step": 1235456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9765068990797078e-05,
+      "loss": 3.5183,
+      "step": 1235968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.975668304328656e-05,
+      "loss": 3.5179,
+      "step": 1236480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.974829709577604e-05,
+      "loss": 3.5216,
+      "step": 1236992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.973991114826552e-05,
+      "loss": 3.5282,
+      "step": 1237504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.973154157955873e-05,
+      "loss": 3.5214,
+      "step": 1238016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.972315563204821e-05,
+      "loss": 3.529,
+      "step": 1238528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.971476968453769e-05,
+      "loss": 3.5212,
+      "step": 1239040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.970638373702717e-05,
+      "loss": 3.529,
+      "step": 1239552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9698014168320382e-05,
+      "loss": 3.5055,
+      "step": 1240064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9689628220809862e-05,
+      "loss": 3.5299,
+      "step": 1240576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9681242273299342e-05,
+      "loss": 3.5039,
+      "step": 1241088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9672856325788822e-05,
+      "loss": 3.5196,
+      "step": 1241600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.966448675708203e-05,
+      "loss": 3.5171,
+      "step": 1242112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9656100809571515e-05,
+      "loss": 3.5293,
+      "step": 1242624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9647714862060995e-05,
+      "loss": 3.5167,
+      "step": 1243136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9639328914550475e-05,
+      "loss": 3.5242,
+      "step": 1243648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9630959345843684e-05,
+      "loss": 3.5247,
+      "step": 1244160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9622573398333164e-05,
+      "loss": 3.5232,
+      "step": 1244672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9614187450822644e-05,
+      "loss": 3.5221,
+      "step": 1245184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9605801503312124e-05,
+      "loss": 3.498,
+      "step": 1245696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9597431934605336e-05,
+      "loss": 3.52,
+      "step": 1246208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9589045987094816e-05,
+      "loss": 3.5248,
+      "step": 1246720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9580660039584296e-05,
+      "loss": 3.5211,
+      "step": 1247232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9572274092073776e-05,
+      "loss": 3.5191,
+      "step": 1247744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9563904523366985e-05,
+      "loss": 3.509,
+      "step": 1248256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.955551857585647e-05,
+      "loss": 3.5155,
+      "step": 1248768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.954713262834595e-05,
+      "loss": 3.4994,
+      "step": 1249280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9538746680835428e-05,
+      "loss": 3.5283,
+      "step": 1249792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9530377112128637e-05,
+      "loss": 3.4998,
+      "step": 1250304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9521991164618117e-05,
+      "loss": 3.5235,
+      "step": 1250816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9513605217107597e-05,
+      "loss": 3.5186,
+      "step": 1251328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9505219269597077e-05,
+      "loss": 3.5083,
+      "step": 1251840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.949684970089029e-05,
+      "loss": 3.5151,
+      "step": 1252352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.948846375337977e-05,
+      "loss": 3.5096,
+      "step": 1252864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.948007780586925e-05,
+      "loss": 3.5125,
+      "step": 1253376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.947169185835873e-05,
+      "loss": 3.5011,
+      "step": 1253888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.946332228965194e-05,
+      "loss": 3.5251,
+      "step": 1254400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9454936342141422e-05,
+      "loss": 3.5088,
+      "step": 1254912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9446550394630902e-05,
+      "loss": 3.5066,
+      "step": 1255424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9438164447120382e-05,
+      "loss": 3.5045,
+      "step": 1255936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.942979487841359e-05,
+      "loss": 3.512,
+      "step": 1256448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.942140893090307e-05,
+      "loss": 3.5215,
+      "step": 1256960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.941302298339255e-05,
+      "loss": 3.5181,
+      "step": 1257472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.940463703588203e-05,
+      "loss": 3.5085,
+      "step": 1257984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9396267467175243e-05,
+      "loss": 3.5215,
+      "step": 1258496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9387881519664723e-05,
+      "loss": 3.5275,
+      "step": 1259008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9379495572154203e-05,
+      "loss": 3.5063,
+      "step": 1259520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9371109624643676e-05,
+      "loss": 3.5229,
+      "step": 1260032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9362740055936892e-05,
+      "loss": 3.5166,
+      "step": 1260544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9354354108426372e-05,
+      "loss": 3.5135,
+      "step": 1261056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9345968160915856e-05,
+      "loss": 3.5088,
+      "step": 1261568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.933758221340533e-05,
+      "loss": 3.5163,
+      "step": 1262080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9329212644698545e-05,
+      "loss": 3.5221,
+      "step": 1262592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9320826697188025e-05,
+      "loss": 3.5281,
+      "step": 1263104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9312440749677505e-05,
+      "loss": 3.5064,
+      "step": 1263616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.930405480216698e-05,
+      "loss": 3.4905,
+      "step": 1264128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.929566885465646e-05,
+      "loss": 3.509,
+      "step": 1264640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9287299285949677e-05,
+      "loss": 3.5165,
+      "step": 1265152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.927891333843915e-05,
+      "loss": 3.5174,
+      "step": 1265664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.927052739092863e-05,
+      "loss": 3.5066,
+      "step": 1266176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.926214144341811e-05,
+      "loss": 3.5157,
+      "step": 1266688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9253771874711326e-05,
+      "loss": 3.5128,
+      "step": 1267200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9245385927200803e-05,
+      "loss": 3.5121,
+      "step": 1267712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9236999979690283e-05,
+      "loss": 3.5046,
+      "step": 1268224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9228614032179762e-05,
+      "loss": 3.5196,
+      "step": 1268736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9220244463472975e-05,
+      "loss": 3.5102,
+      "step": 1269248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9211858515962455e-05,
+      "loss": 3.5224,
+      "step": 1269760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9203472568451935e-05,
+      "loss": 3.514,
+      "step": 1270272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9195086620941415e-05,
+      "loss": 3.4999,
+      "step": 1270784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9186717052234624e-05,
+      "loss": 3.516,
+      "step": 1271296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9178331104724104e-05,
+      "loss": 3.5173,
+      "step": 1271808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9169945157213584e-05,
+      "loss": 3.5202,
+      "step": 1272320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9161559209703064e-05,
+      "loss": 3.5158,
+      "step": 1272832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9153189640996276e-05,
+      "loss": 3.5095,
+      "step": 1273344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9144803693485756e-05,
+      "loss": 3.5044,
+      "step": 1273856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9136417745975236e-05,
+      "loss": 3.5152,
+      "step": 1274368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9128048177268445e-05,
+      "loss": 3.5009,
+      "step": 1274880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9119662229757925e-05,
+      "loss": 3.5094,
+      "step": 1275392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.911127628224741e-05,
+      "loss": 3.5154,
+      "step": 1275904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.910289033473689e-05,
+      "loss": 3.5125,
+      "step": 1276416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9094520766030098e-05,
+      "loss": 3.5014,
+      "step": 1276928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9086134818519578e-05,
+      "loss": 3.5109,
+      "step": 1277440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9077748871009058e-05,
+      "loss": 3.5066,
+      "step": 1277952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9069362923498538e-05,
+      "loss": 3.5174,
+      "step": 1278464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9060976975988017e-05,
+      "loss": 3.5114,
+      "step": 1278976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.905260740728123e-05,
+      "loss": 3.5105,
+      "step": 1279488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.904422145977071e-05,
+      "loss": 3.5233,
+      "step": 1280000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.903583551226019e-05,
+      "loss": 3.5234,
+      "step": 1280512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.90274659435534e-05,
+      "loss": 3.5129,
+      "step": 1281024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.901907999604288e-05,
+      "loss": 3.5077,
+      "step": 1281536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9010694048532362e-05,
+      "loss": 3.5163,
+      "step": 1282048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9002308101021842e-05,
+      "loss": 3.5074,
+      "step": 1282560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8993922153511322e-05,
+      "loss": 3.5071,
+      "step": 1283072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.898555258480453e-05,
+      "loss": 3.52,
+      "step": 1283584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.897716663729401e-05,
+      "loss": 3.506,
+      "step": 1284096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.896878068978349e-05,
+      "loss": 3.5134,
+      "step": 1284608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.896039474227297e-05,
+      "loss": 3.5037,
+      "step": 1285120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8952025173566184e-05,
+      "loss": 3.5084,
+      "step": 1285632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8943639226055664e-05,
+      "loss": 3.5149,
+      "step": 1286144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8935253278545144e-05,
+      "loss": 3.5065,
+      "step": 1286656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8926867331034623e-05,
+      "loss": 3.5077,
+      "step": 1287168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8918497762327833e-05,
+      "loss": 3.5298,
+      "step": 1287680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8910111814817316e-05,
+      "loss": 3.5178,
+      "step": 1288192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8901725867306796e-05,
+      "loss": 3.5134,
+      "step": 1288704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8893339919796276e-05,
+      "loss": 3.5095,
+      "step": 1289216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8884970351089485e-05,
+      "loss": 3.5122,
+      "step": 1289728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8876584403578965e-05,
+      "loss": 3.5138,
+      "step": 1290240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8868198456068445e-05,
+      "loss": 3.5128,
+      "step": 1290752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8859812508557925e-05,
+      "loss": 3.509,
+      "step": 1291264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8851442939851137e-05,
+      "loss": 3.5238,
+      "step": 1291776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8843056992340617e-05,
+      "loss": 3.5183,
+      "step": 1292288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8834671044830097e-05,
+      "loss": 3.5184,
+      "step": 1292800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8826285097319577e-05,
+      "loss": 3.513,
+      "step": 1293312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8817915528612786e-05,
+      "loss": 3.5163,
+      "step": 1293824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.880952958110227e-05,
+      "loss": 3.5119,
+      "step": 1294336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.880114363359175e-05,
+      "loss": 3.515,
+      "step": 1294848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.879275768608123e-05,
+      "loss": 3.5173,
+      "step": 1295360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.878438811737444e-05,
+      "loss": 3.5149,
+      "step": 1295872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.877600216986392e-05,
+      "loss": 3.5198,
+      "step": 1296384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.87676162223534e-05,
+      "loss": 3.5021,
+      "step": 1296896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.875923027484288e-05,
+      "loss": 3.5096,
+      "step": 1297408
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.852449893951416,
+      "eval_runtime": 303.693,
+      "eval_samples_per_second": 1256.503,
+      "eval_steps_per_second": 39.267,
+      "step": 1297440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8750844327332362e-05,
+      "loss": 3.5004,
+      "step": 1297920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8742458379821842e-05,
+      "loss": 3.5049,
+      "step": 1298432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8734072432311322e-05,
+      "loss": 3.5275,
+      "step": 1298944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.87256864848008e-05,
+      "loss": 3.518,
+      "step": 1299456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.871731691609401e-05,
+      "loss": 3.5166,
+      "step": 1299968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.870893096858349e-05,
+      "loss": 3.509,
+      "step": 1300480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.870054502107297e-05,
+      "loss": 3.51,
+      "step": 1300992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8692159073562454e-05,
+      "loss": 3.5005,
+      "step": 1301504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8683789504855663e-05,
+      "loss": 3.5163,
+      "step": 1302016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8675403557345143e-05,
+      "loss": 3.5133,
+      "step": 1302528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8667017609834623e-05,
+      "loss": 3.5104,
+      "step": 1303040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8658631662324103e-05,
+      "loss": 3.5146,
+      "step": 1303552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8650278472421045e-05,
+      "loss": 3.5076,
+      "step": 1304064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8641892524910525e-05,
+      "loss": 3.5034,
+      "step": 1304576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8633506577400005e-05,
+      "loss": 3.4988,
+      "step": 1305088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8625120629889484e-05,
+      "loss": 3.4995,
+      "step": 1305600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8616734682378964e-05,
+      "loss": 3.5075,
+      "step": 1306112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8608348734868444e-05,
+      "loss": 3.5049,
+      "step": 1306624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8599962787357924e-05,
+      "loss": 3.5067,
+      "step": 1307136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8591576839847404e-05,
+      "loss": 3.5287,
+      "step": 1307648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8583207271140617e-05,
+      "loss": 3.5101,
+      "step": 1308160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8574821323630097e-05,
+      "loss": 3.5146,
+      "step": 1308672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8566435376119577e-05,
+      "loss": 3.5092,
+      "step": 1309184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8558049428609053e-05,
+      "loss": 3.5185,
+      "step": 1309696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.854967985990227e-05,
+      "loss": 3.5007,
+      "step": 1310208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.854129391239175e-05,
+      "loss": 3.5074,
+      "step": 1310720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.853290796488123e-05,
+      "loss": 3.5044,
+      "step": 1311232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8524522017370702e-05,
+      "loss": 3.5092,
+      "step": 1311744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8516152448663918e-05,
+      "loss": 3.5027,
+      "step": 1312256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8507766501153398e-05,
+      "loss": 3.5008,
+      "step": 1312768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8499380553642878e-05,
+      "loss": 3.5093,
+      "step": 1313280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8490994606132355e-05,
+      "loss": 3.5168,
+      "step": 1313792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.848262503742557e-05,
+      "loss": 3.5083,
+      "step": 1314304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.847423908991505e-05,
+      "loss": 3.5131,
+      "step": 1314816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8465853142404524e-05,
+      "loss": 3.5033,
+      "step": 1315328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8457467194894007e-05,
+      "loss": 3.5144,
+      "step": 1315840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.844909762618722e-05,
+      "loss": 3.496,
+      "step": 1316352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8440711678676703e-05,
+      "loss": 3.512,
+      "step": 1316864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8432325731166176e-05,
+      "loss": 3.4923,
+      "step": 1317376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8423939783655656e-05,
+      "loss": 3.5097,
+      "step": 1317888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8415553836145136e-05,
+      "loss": 3.5015,
+      "step": 1318400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8407184267438352e-05,
+      "loss": 3.5145,
+      "step": 1318912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8398798319927828e-05,
+      "loss": 3.5029,
+      "step": 1319424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8390412372417308e-05,
+      "loss": 3.5097,
+      "step": 1319936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8382026424906788e-05,
+      "loss": 3.5143,
+      "step": 1320448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8373656856199997e-05,
+      "loss": 3.5094,
+      "step": 1320960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8365270908689477e-05,
+      "loss": 3.5112,
+      "step": 1321472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8356884961178957e-05,
+      "loss": 3.4881,
+      "step": 1321984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.834849901366844e-05,
+      "loss": 3.4995,
+      "step": 1322496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.834012944496165e-05,
+      "loss": 3.5111,
+      "step": 1323008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.833174349745113e-05,
+      "loss": 3.5093,
+      "step": 1323520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.832335754994061e-05,
+      "loss": 3.5026,
+      "step": 1324032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.831497160243009e-05,
+      "loss": 3.5006,
+      "step": 1324544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8306602033723302e-05,
+      "loss": 3.5,
+      "step": 1325056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8298216086212782e-05,
+      "loss": 3.4927,
+      "step": 1325568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8289830138702262e-05,
+      "loss": 3.5075,
+      "step": 1326080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8281444191191742e-05,
+      "loss": 3.491,
+      "step": 1326592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.827307462248495e-05,
+      "loss": 3.5039,
+      "step": 1327104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.826468867497443e-05,
+      "loss": 3.5089,
+      "step": 1327616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.825630272746391e-05,
+      "loss": 3.497,
+      "step": 1328128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8247916779953394e-05,
+      "loss": 3.498,
+      "step": 1328640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8239547211246603e-05,
+      "loss": 3.4972,
+      "step": 1329152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8231161263736083e-05,
+      "loss": 3.493,
+      "step": 1329664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8222775316225563e-05,
+      "loss": 3.4907,
+      "step": 1330176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8214389368715043e-05,
+      "loss": 3.5124,
+      "step": 1330688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8206019800008256e-05,
+      "loss": 3.4967,
+      "step": 1331200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8197633852497736e-05,
+      "loss": 3.4936,
+      "step": 1331712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8189247904987216e-05,
+      "loss": 3.4939,
+      "step": 1332224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8180861957476696e-05,
+      "loss": 3.4962,
+      "step": 1332736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8172492388769905e-05,
+      "loss": 3.5087,
+      "step": 1333248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8164106441259385e-05,
+      "loss": 3.5024,
+      "step": 1333760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8155720493748865e-05,
+      "loss": 3.4971,
+      "step": 1334272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8147334546238348e-05,
+      "loss": 3.5078,
+      "step": 1334784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8138948598727828e-05,
+      "loss": 3.5103,
+      "step": 1335296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8130579030021037e-05,
+      "loss": 3.494,
+      "step": 1335808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8122193082510517e-05,
+      "loss": 3.507,
+      "step": 1336320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8113807134999997e-05,
+      "loss": 3.5053,
+      "step": 1336832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8105421187489477e-05,
+      "loss": 3.4994,
+      "step": 1337344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.809705161878269e-05,
+      "loss": 3.4971,
+      "step": 1337856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.808866567127217e-05,
+      "loss": 3.4973,
+      "step": 1338368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.808027972376165e-05,
+      "loss": 3.5093,
+      "step": 1338880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.807189377625113e-05,
+      "loss": 3.5065,
+      "step": 1339392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8063524207544338e-05,
+      "loss": 3.4986,
+      "step": 1339904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8055138260033818e-05,
+      "loss": 3.4767,
+      "step": 1340416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.80467523125233e-05,
+      "loss": 3.4947,
+      "step": 1340928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.803836636501278e-05,
+      "loss": 3.5036,
+      "step": 1341440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.802999679630599e-05,
+      "loss": 3.5021,
+      "step": 1341952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.802161084879547e-05,
+      "loss": 3.5,
+      "step": 1342464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.801322490128495e-05,
+      "loss": 3.4962,
+      "step": 1342976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.800483895377443e-05,
+      "loss": 3.5011,
+      "step": 1343488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7996469385067643e-05,
+      "loss": 3.4947,
+      "step": 1344000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7988083437557123e-05,
+      "loss": 3.4901,
+      "step": 1344512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7979697490046603e-05,
+      "loss": 3.5109,
+      "step": 1345024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7971311542536083e-05,
+      "loss": 3.4928,
+      "step": 1345536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7962941973829292e-05,
+      "loss": 3.5109,
+      "step": 1346048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7954556026318772e-05,
+      "loss": 3.5044,
+      "step": 1346560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7946170078808255e-05,
+      "loss": 3.4848,
+      "step": 1347072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7937784131297735e-05,
+      "loss": 3.503,
+      "step": 1347584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7929414562590944e-05,
+      "loss": 3.4982,
+      "step": 1348096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7921028615080424e-05,
+      "loss": 3.5083,
+      "step": 1348608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7912642667569904e-05,
+      "loss": 3.5032,
+      "step": 1349120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7904256720059384e-05,
+      "loss": 3.4953,
+      "step": 1349632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7895887151352597e-05,
+      "loss": 3.491,
+      "step": 1350144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7887501203842077e-05,
+      "loss": 3.5055,
+      "step": 1350656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7879115256331557e-05,
+      "loss": 3.4874,
+      "step": 1351168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7870729308821036e-05,
+      "loss": 3.4946,
+      "step": 1351680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7862359740114246e-05,
+      "loss": 3.5029,
+      "step": 1352192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7853973792603726e-05,
+      "loss": 3.4983,
+      "step": 1352704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.784558784509321e-05,
+      "loss": 3.4863,
+      "step": 1353216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7837218276386418e-05,
+      "loss": 3.5015,
+      "step": 1353728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7828832328875898e-05,
+      "loss": 3.4944,
+      "step": 1354240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7820446381365378e-05,
+      "loss": 3.5004,
+      "step": 1354752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7812060433854858e-05,
+      "loss": 3.4994,
+      "step": 1355264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7803674486344338e-05,
+      "loss": 3.49,
+      "step": 1355776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.779530491763755e-05,
+      "loss": 3.5169,
+      "step": 1356288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.778691897012703e-05,
+      "loss": 3.5075,
+      "step": 1356800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.777853302261651e-05,
+      "loss": 3.4994,
+      "step": 1357312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.777014707510599e-05,
+      "loss": 3.4937,
+      "step": 1357824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.77617775063992e-05,
+      "loss": 3.5008,
+      "step": 1358336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.775339155888868e-05,
+      "loss": 3.4972,
+      "step": 1358848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7745005611378163e-05,
+      "loss": 3.4927,
+      "step": 1359360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7736619663867643e-05,
+      "loss": 3.5057,
+      "step": 1359872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.772825009516085e-05,
+      "loss": 3.4974,
+      "step": 1360384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.771986414765033e-05,
+      "loss": 3.494,
+      "step": 1360896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.771147820013981e-05,
+      "loss": 3.4941,
+      "step": 1361408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.770309225262929e-05,
+      "loss": 3.4956,
+      "step": 1361920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7694722683922504e-05,
+      "loss": 3.5004,
+      "step": 1362432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7686336736411984e-05,
+      "loss": 3.4931,
+      "step": 1362944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7677950788901464e-05,
+      "loss": 3.4908,
+      "step": 1363456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7669564841390944e-05,
+      "loss": 3.5158,
+      "step": 1363968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7661195272684153e-05,
+      "loss": 3.5071,
+      "step": 1364480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7652809325173633e-05,
+      "loss": 3.4998,
+      "step": 1364992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7644423377663113e-05,
+      "loss": 3.4943,
+      "step": 1365504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7636037430152596e-05,
+      "loss": 3.4947,
+      "step": 1366016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7627667861445805e-05,
+      "loss": 3.503,
+      "step": 1366528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7619281913935285e-05,
+      "loss": 3.4995,
+      "step": 1367040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7610895966424765e-05,
+      "loss": 3.4929,
+      "step": 1367552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7602510018914245e-05,
+      "loss": 3.5117,
+      "step": 1368064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7594140450207458e-05,
+      "loss": 3.5016,
+      "step": 1368576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7585754502696938e-05,
+      "loss": 3.5082,
+      "step": 1369088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7577368555186418e-05,
+      "loss": 3.4982,
+      "step": 1369600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7568982607675897e-05,
+      "loss": 3.506,
+      "step": 1370112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7560613038969107e-05,
+      "loss": 3.4975,
+      "step": 1370624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7552227091458587e-05,
+      "loss": 3.4988,
+      "step": 1371136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7543841143948066e-05,
+      "loss": 3.5022,
+      "step": 1371648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.753545519643755e-05,
+      "loss": 3.5036,
+      "step": 1372160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.752708562773076e-05,
+      "loss": 3.5053,
+      "step": 1372672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.751869968022024e-05,
+      "loss": 3.4888,
+      "step": 1373184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.751031373270972e-05,
+      "loss": 3.4973,
+      "step": 1373696
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8527557849884033,
+      "eval_runtime": 303.2007,
+      "eval_samples_per_second": 1258.542,
+      "eval_steps_per_second": 39.33,
+      "step": 1373760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.75019277851992e-05,
+      "loss": 3.4888,
+      "step": 1374208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.749355821649241e-05,
+      "loss": 3.4913,
+      "step": 1374720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.748517226898189e-05,
+      "loss": 3.5116,
+      "step": 1375232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.747678632147137e-05,
+      "loss": 3.5065,
+      "step": 1375744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.746840037396085e-05,
+      "loss": 3.5029,
+      "step": 1376256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.746003080525406e-05,
+      "loss": 3.4953,
+      "step": 1376768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.745164485774354e-05,
+      "loss": 3.4987,
+      "step": 1377280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.744325891023302e-05,
+      "loss": 3.4858,
+      "step": 1377792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7434872962722497e-05,
+      "loss": 3.5059,
+      "step": 1378304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7426487015211977e-05,
+      "loss": 3.4992,
+      "step": 1378816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7418117446505193e-05,
+      "loss": 3.4951,
+      "step": 1379328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7409731498994673e-05,
+      "loss": 3.5001,
+      "step": 1379840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.740134555148415e-05,
+      "loss": 3.4918,
+      "step": 1380352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.739295960397363e-05,
+      "loss": 3.4914,
+      "step": 1380864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7384590035266845e-05,
+      "loss": 3.4887,
+      "step": 1381376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7376204087756325e-05,
+      "loss": 3.4878,
+      "step": 1381888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7367818140245798e-05,
+      "loss": 3.4894,
+      "step": 1382400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7359432192735278e-05,
+      "loss": 3.4923,
+      "step": 1382912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7351062624028494e-05,
+      "loss": 3.495,
+      "step": 1383424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.734267667651797e-05,
+      "loss": 3.519,
+      "step": 1383936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.733429072900745e-05,
+      "loss": 3.4957,
+      "step": 1384448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.732590478149693e-05,
+      "loss": 3.4986,
+      "step": 1384960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7317535212790146e-05,
+      "loss": 3.4983,
+      "step": 1385472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.730914926527962e-05,
+      "loss": 3.503,
+      "step": 1385984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7300763317769103e-05,
+      "loss": 3.4856,
+      "step": 1386496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7292377370258583e-05,
+      "loss": 3.4988,
+      "step": 1387008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7284007801551792e-05,
+      "loss": 3.4898,
+      "step": 1387520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7275621854041272e-05,
+      "loss": 3.4938,
+      "step": 1388032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7267235906530752e-05,
+      "loss": 3.4882,
+      "step": 1388544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.725884995902023e-05,
+      "loss": 3.489,
+      "step": 1389056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7250480390313444e-05,
+      "loss": 3.4935,
+      "step": 1389568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7242094442802924e-05,
+      "loss": 3.5035,
+      "step": 1390080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7233708495292404e-05,
+      "loss": 3.4944,
+      "step": 1390592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7225322547781884e-05,
+      "loss": 3.503,
+      "step": 1391104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7216952979075093e-05,
+      "loss": 3.4852,
+      "step": 1391616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7208567031564573e-05,
+      "loss": 3.5052,
+      "step": 1392128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7200181084054056e-05,
+      "loss": 3.4804,
+      "step": 1392640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7191795136543536e-05,
+      "loss": 3.4987,
+      "step": 1393152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7183425567836745e-05,
+      "loss": 3.4789,
+      "step": 1393664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7175039620326225e-05,
+      "loss": 3.4952,
+      "step": 1394176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7166653672815705e-05,
+      "loss": 3.4864,
+      "step": 1394688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7158267725305185e-05,
+      "loss": 3.5058,
+      "step": 1395200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7149898156598398e-05,
+      "loss": 3.4912,
+      "step": 1395712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7141512209087878e-05,
+      "loss": 3.4942,
+      "step": 1396224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7133126261577358e-05,
+      "loss": 3.4993,
+      "step": 1396736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7124740314066838e-05,
+      "loss": 3.499,
+      "step": 1397248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7116370745360047e-05,
+      "loss": 3.4951,
+      "step": 1397760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7107984797849527e-05,
+      "loss": 3.48,
+      "step": 1398272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.709959885033901e-05,
+      "loss": 3.483,
+      "step": 1398784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.709121290282849e-05,
+      "loss": 3.5012,
+      "step": 1399296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.70828433341217e-05,
+      "loss": 3.4979,
+      "step": 1399808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.707445738661118e-05,
+      "loss": 3.4876,
+      "step": 1400320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.706607143910066e-05,
+      "loss": 3.4887,
+      "step": 1400832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.705768549159014e-05,
+      "loss": 3.4882,
+      "step": 1401344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.704931592288335e-05,
+      "loss": 3.4768,
+      "step": 1401856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.704092997537283e-05,
+      "loss": 3.4918,
+      "step": 1402368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.703254402786231e-05,
+      "loss": 3.4793,
+      "step": 1402880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.702415808035179e-05,
+      "loss": 3.4922,
+      "step": 1403392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7015788511645e-05,
+      "loss": 3.4971,
+      "step": 1403904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.700740256413448e-05,
+      "loss": 3.4794,
+      "step": 1404416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6999016616623964e-05,
+      "loss": 3.4857,
+      "step": 1404928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6990630669113444e-05,
+      "loss": 3.4847,
+      "step": 1405440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6982244721602924e-05,
+      "loss": 3.4802,
+      "step": 1405952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6973875152896133e-05,
+      "loss": 3.474,
+      "step": 1406464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6965489205385613e-05,
+      "loss": 3.4994,
+      "step": 1406976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6957103257875093e-05,
+      "loss": 3.4854,
+      "step": 1407488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6948717310364573e-05,
+      "loss": 3.478,
+      "step": 1408000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6940347741657785e-05,
+      "loss": 3.4803,
+      "step": 1408512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6931961794147265e-05,
+      "loss": 3.4875,
+      "step": 1409024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6923575846636745e-05,
+      "loss": 3.4943,
+      "step": 1409536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6915189899126225e-05,
+      "loss": 3.4914,
+      "step": 1410048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6906820330419434e-05,
+      "loss": 3.4809,
+      "step": 1410560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6898434382908917e-05,
+      "loss": 3.4973,
+      "step": 1411072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6890048435398397e-05,
+      "loss": 3.4927,
+      "step": 1411584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6881662487887877e-05,
+      "loss": 3.4843,
+      "step": 1412096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6873276540377357e-05,
+      "loss": 3.4921,
+      "step": 1412608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6864906971670566e-05,
+      "loss": 3.496,
+      "step": 1413120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6856521024160046e-05,
+      "loss": 3.4871,
+      "step": 1413632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6848135076649526e-05,
+      "loss": 3.4855,
+      "step": 1414144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.683974912913901e-05,
+      "loss": 3.4851,
+      "step": 1414656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.683137956043222e-05,
+      "loss": 3.495,
+      "step": 1415168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.68229936129217e-05,
+      "loss": 3.4946,
+      "step": 1415680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.681460766541118e-05,
+      "loss": 3.4826,
+      "step": 1416192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.680622171790066e-05,
+      "loss": 3.4722,
+      "step": 1416704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.679785214919387e-05,
+      "loss": 3.4773,
+      "step": 1417216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.678946620168335e-05,
+      "loss": 3.489,
+      "step": 1417728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.678108025417283e-05,
+      "loss": 3.4934,
+      "step": 1418240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.677271068546604e-05,
+      "loss": 3.4848,
+      "step": 1418752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.676432473795552e-05,
+      "loss": 3.4835,
+      "step": 1419264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6755938790445e-05,
+      "loss": 3.4912,
+      "step": 1419776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.674755284293448e-05,
+      "loss": 3.4802,
+      "step": 1420288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6739183274227692e-05,
+      "loss": 3.4788,
+      "step": 1420800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6730797326717172e-05,
+      "loss": 3.4929,
+      "step": 1421312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6722411379206652e-05,
+      "loss": 3.4813,
+      "step": 1421824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6714025431696132e-05,
+      "loss": 3.5012,
+      "step": 1422336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.670565586298934e-05,
+      "loss": 3.4879,
+      "step": 1422848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.669726991547882e-05,
+      "loss": 3.4741,
+      "step": 1423360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6688883967968305e-05,
+      "loss": 3.4901,
+      "step": 1423872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6680498020457785e-05,
+      "loss": 3.4851,
+      "step": 1424384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6672128451750994e-05,
+      "loss": 3.5009,
+      "step": 1424896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6663742504240474e-05,
+      "loss": 3.4909,
+      "step": 1425408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6655356556729954e-05,
+      "loss": 3.4814,
+      "step": 1425920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6646970609219434e-05,
+      "loss": 3.4735,
+      "step": 1426432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6638601040512646e-05,
+      "loss": 3.491,
+      "step": 1426944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6630215093002126e-05,
+      "loss": 3.4745,
+      "step": 1427456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6621829145491606e-05,
+      "loss": 3.4864,
+      "step": 1427968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6613443197981086e-05,
+      "loss": 3.4883,
+      "step": 1428480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6605057250470566e-05,
+      "loss": 3.4838,
+      "step": 1428992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6596687681763775e-05,
+      "loss": 3.4738,
+      "step": 1429504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.658830173425326e-05,
+      "loss": 3.4942,
+      "step": 1430016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.657991578674274e-05,
+      "loss": 3.4813,
+      "step": 1430528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6571529839232218e-05,
+      "loss": 3.483,
+      "step": 1431040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6563160270525427e-05,
+      "loss": 3.4883,
+      "step": 1431552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6554774323014907e-05,
+      "loss": 3.4787,
+      "step": 1432064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6546388375504387e-05,
+      "loss": 3.5017,
+      "step": 1432576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6538002427993867e-05,
+      "loss": 3.4933,
+      "step": 1433088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.652963285928708e-05,
+      "loss": 3.4888,
+      "step": 1433600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.652124691177656e-05,
+      "loss": 3.4796,
+      "step": 1434112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.651286096426604e-05,
+      "loss": 3.4888,
+      "step": 1434624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.650447501675552e-05,
+      "loss": 3.4869,
+      "step": 1435136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.649610544804873e-05,
+      "loss": 3.4747,
+      "step": 1435648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6487719500538212e-05,
+      "loss": 3.491,
+      "step": 1436160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6479333553027692e-05,
+      "loss": 3.4879,
+      "step": 1436672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6470947605517172e-05,
+      "loss": 3.4805,
+      "step": 1437184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.646257803681038e-05,
+      "loss": 3.483,
+      "step": 1437696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.645419208929986e-05,
+      "loss": 3.4846,
+      "step": 1438208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.644580614178934e-05,
+      "loss": 3.48,
+      "step": 1438720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6437420194278818e-05,
+      "loss": 3.485,
+      "step": 1439232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6429050625572033e-05,
+      "loss": 3.4802,
+      "step": 1439744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6420664678061513e-05,
+      "loss": 3.4986,
+      "step": 1440256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6412278730550993e-05,
+      "loss": 3.5005,
+      "step": 1440768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6403892783040466e-05,
+      "loss": 3.4837,
+      "step": 1441280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6395523214333682e-05,
+      "loss": 3.4819,
+      "step": 1441792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6387137266823166e-05,
+      "loss": 3.4818,
+      "step": 1442304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6378751319312646e-05,
+      "loss": 3.49,
+      "step": 1442816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.637036537180212e-05,
+      "loss": 3.4884,
+      "step": 1443328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6361995803095335e-05,
+      "loss": 3.4815,
+      "step": 1443840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6353609855584815e-05,
+      "loss": 3.4962,
+      "step": 1444352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.634522390807429e-05,
+      "loss": 3.4884,
+      "step": 1444864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.633683796056377e-05,
+      "loss": 3.4958,
+      "step": 1445376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6328468391856987e-05,
+      "loss": 3.4831,
+      "step": 1445888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6320082444346467e-05,
+      "loss": 3.4955,
+      "step": 1446400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.631169649683594e-05,
+      "loss": 3.4874,
+      "step": 1446912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.630331054932542e-05,
+      "loss": 3.4816,
+      "step": 1447424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6294940980618636e-05,
+      "loss": 3.4932,
+      "step": 1447936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.628655503310812e-05,
+      "loss": 3.4918,
+      "step": 1448448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6278169085597593e-05,
+      "loss": 3.4904,
+      "step": 1448960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6269783138087073e-05,
+      "loss": 3.4749,
+      "step": 1449472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.626141356938029e-05,
+      "loss": 3.4814,
+      "step": 1449984
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8544392585754395,
+      "eval_runtime": 303.1261,
+      "eval_samples_per_second": 1258.852,
+      "eval_steps_per_second": 39.34,
+      "step": 1450080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6253027621869765e-05,
+      "loss": 3.4771,
+      "step": 1450496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6244641674359245e-05,
+      "loss": 3.4783,
+      "step": 1451008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6236255726848725e-05,
+      "loss": 3.4976,
+      "step": 1451520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6227869779338205e-05,
+      "loss": 3.4944,
+      "step": 1452032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6219500210631414e-05,
+      "loss": 3.4973,
+      "step": 1452544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6211114263120894e-05,
+      "loss": 3.4774,
+      "step": 1453056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6202728315610374e-05,
+      "loss": 3.4893,
+      "step": 1453568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6194342368099857e-05,
+      "loss": 3.468,
+      "step": 1454080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6185972799393066e-05,
+      "loss": 3.4931,
+      "step": 1454592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6177586851882546e-05,
+      "loss": 3.4894,
+      "step": 1455104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6169200904372026e-05,
+      "loss": 3.4782,
+      "step": 1455616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6160814956861506e-05,
+      "loss": 3.4902,
+      "step": 1456128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.615244538815472e-05,
+      "loss": 3.4765,
+      "step": 1456640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.61440594406442e-05,
+      "loss": 3.4793,
+      "step": 1457152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.613567349313368e-05,
+      "loss": 3.4763,
+      "step": 1457664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.612728754562316e-05,
+      "loss": 3.4779,
+      "step": 1458176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6118917976916368e-05,
+      "loss": 3.4784,
+      "step": 1458688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6110532029405848e-05,
+      "loss": 3.4792,
+      "step": 1459200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6102146081895327e-05,
+      "loss": 3.482,
+      "step": 1459712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.609376013438481e-05,
+      "loss": 3.501,
+      "step": 1460224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.608539056567802e-05,
+      "loss": 3.4815,
+      "step": 1460736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.60770046181675e-05,
+      "loss": 3.488,
+      "step": 1461248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.606861867065698e-05,
+      "loss": 3.486,
+      "step": 1461760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.606023272314646e-05,
+      "loss": 3.4895,
+      "step": 1462272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6051863154439672e-05,
+      "loss": 3.4756,
+      "step": 1462784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6043477206929152e-05,
+      "loss": 3.4845,
+      "step": 1463296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6035091259418632e-05,
+      "loss": 3.4785,
+      "step": 1463808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6026705311908112e-05,
+      "loss": 3.4779,
+      "step": 1464320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.601833574320132e-05,
+      "loss": 3.4745,
+      "step": 1464832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.60099497956908e-05,
+      "loss": 3.4818,
+      "step": 1465344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.600156384818028e-05,
+      "loss": 3.4811,
+      "step": 1465856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5993177900669765e-05,
+      "loss": 3.49,
+      "step": 1466368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5984808331962974e-05,
+      "loss": 3.4833,
+      "step": 1466880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5976422384452454e-05,
+      "loss": 3.4857,
+      "step": 1467392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5968036436941934e-05,
+      "loss": 3.4748,
+      "step": 1467904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5959650489431413e-05,
+      "loss": 3.4892,
+      "step": 1468416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5951280920724626e-05,
+      "loss": 3.4738,
+      "step": 1468928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5942894973214106e-05,
+      "loss": 3.4835,
+      "step": 1469440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5934509025703586e-05,
+      "loss": 3.4671,
+      "step": 1469952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5926123078193066e-05,
+      "loss": 3.4798,
+      "step": 1470464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5917753509486275e-05,
+      "loss": 3.4763,
+      "step": 1470976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5909367561975755e-05,
+      "loss": 3.4932,
+      "step": 1471488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5900981614465235e-05,
+      "loss": 3.4805,
+      "step": 1472000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5892595666954718e-05,
+      "loss": 3.4812,
+      "step": 1472512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5884226098247927e-05,
+      "loss": 3.4866,
+      "step": 1473024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5875840150737407e-05,
+      "loss": 3.4879,
+      "step": 1473536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5867454203226887e-05,
+      "loss": 3.481,
+      "step": 1474048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5859068255716367e-05,
+      "loss": 3.4714,
+      "step": 1474560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.585069868700958e-05,
+      "loss": 3.4643,
+      "step": 1475072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.584231273949906e-05,
+      "loss": 3.4886,
+      "step": 1475584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.583392679198854e-05,
+      "loss": 3.4872,
+      "step": 1476096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.582554084447802e-05,
+      "loss": 3.4776,
+      "step": 1476608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.581717127577123e-05,
+      "loss": 3.4712,
+      "step": 1477120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.580878532826071e-05,
+      "loss": 3.4792,
+      "step": 1477632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.580039938075019e-05,
+      "loss": 3.46,
+      "step": 1478144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.579201343323967e-05,
+      "loss": 3.4824,
+      "step": 1478656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5783627485729152e-05,
+      "loss": 3.465,
+      "step": 1479168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.577525791702236e-05,
+      "loss": 3.4789,
+      "step": 1479680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.576687196951184e-05,
+      "loss": 3.4835,
+      "step": 1480192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.575848602200132e-05,
+      "loss": 3.4741,
+      "step": 1480704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5750116453294533e-05,
+      "loss": 3.4659,
+      "step": 1481216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5741730505784013e-05,
+      "loss": 3.4753,
+      "step": 1481728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5733344558273493e-05,
+      "loss": 3.4668,
+      "step": 1482240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5724958610762973e-05,
+      "loss": 3.4658,
+      "step": 1482752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5716572663252453e-05,
+      "loss": 3.4824,
+      "step": 1483264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5708203094545662e-05,
+      "loss": 3.4739,
+      "step": 1483776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5699817147035142e-05,
+      "loss": 3.4682,
+      "step": 1484288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5691431199524622e-05,
+      "loss": 3.4656,
+      "step": 1484800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5683045252014105e-05,
+      "loss": 3.4743,
+      "step": 1485312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5674675683307315e-05,
+      "loss": 3.4772,
+      "step": 1485824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5666289735796795e-05,
+      "loss": 3.4795,
+      "step": 1486336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5657903788286274e-05,
+      "loss": 3.4732,
+      "step": 1486848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5649517840775754e-05,
+      "loss": 3.4838,
+      "step": 1487360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5641148272068967e-05,
+      "loss": 3.4838,
+      "step": 1487872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5632762324558447e-05,
+      "loss": 3.4744,
+      "step": 1488384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5624376377047927e-05,
+      "loss": 3.4765,
+      "step": 1488896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5615990429537407e-05,
+      "loss": 3.4837,
+      "step": 1489408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5607620860830616e-05,
+      "loss": 3.4781,
+      "step": 1489920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5599234913320096e-05,
+      "loss": 3.4722,
+      "step": 1490432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5590848965809576e-05,
+      "loss": 3.4738,
+      "step": 1490944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.558246301829906e-05,
+      "loss": 3.4805,
+      "step": 1491456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5574093449592268e-05,
+      "loss": 3.483,
+      "step": 1491968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5565707502081748e-05,
+      "loss": 3.4739,
+      "step": 1492480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5557321554571228e-05,
+      "loss": 3.4607,
+      "step": 1492992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5548935607060708e-05,
+      "loss": 3.461,
+      "step": 1493504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.554056603835392e-05,
+      "loss": 3.4735,
+      "step": 1494016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.55321800908434e-05,
+      "loss": 3.4818,
+      "step": 1494528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.552379414333288e-05,
+      "loss": 3.4718,
+      "step": 1495040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.551540819582236e-05,
+      "loss": 3.4662,
+      "step": 1495552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.550703862711557e-05,
+      "loss": 3.4777,
+      "step": 1496064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.549865267960505e-05,
+      "loss": 3.4707,
+      "step": 1496576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.549026673209453e-05,
+      "loss": 3.468,
+      "step": 1497088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5481880784584013e-05,
+      "loss": 3.4756,
+      "step": 1497600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5473511215877222e-05,
+      "loss": 3.4778,
+      "step": 1498112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5465125268366702e-05,
+      "loss": 3.489,
+      "step": 1498624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5456739320856182e-05,
+      "loss": 3.4714,
+      "step": 1499136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5448353373345662e-05,
+      "loss": 3.4598,
+      "step": 1499648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5439983804638874e-05,
+      "loss": 3.48,
+      "step": 1500160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5431597857128354e-05,
+      "loss": 3.4703,
+      "step": 1500672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5423211909617834e-05,
+      "loss": 3.4848,
+      "step": 1501184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5414825962107314e-05,
+      "loss": 3.4811,
+      "step": 1501696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5406456393400523e-05,
+      "loss": 3.4692,
+      "step": 1502208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5398070445890003e-05,
+      "loss": 3.4629,
+      "step": 1502720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5389684498379483e-05,
+      "loss": 3.478,
+      "step": 1503232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5381298550868966e-05,
+      "loss": 3.4618,
+      "step": 1503744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5372928982162176e-05,
+      "loss": 3.4744,
+      "step": 1504256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5364543034651656e-05,
+      "loss": 3.471,
+      "step": 1504768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5356157087141135e-05,
+      "loss": 3.4777,
+      "step": 1505280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5347771139630612e-05,
+      "loss": 3.461,
+      "step": 1505792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5339401570923828e-05,
+      "loss": 3.479,
+      "step": 1506304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5331015623413308e-05,
+      "loss": 3.4674,
+      "step": 1506816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5322629675902788e-05,
+      "loss": 3.4706,
+      "step": 1507328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.531424372839226e-05,
+      "loss": 3.4772,
+      "step": 1507840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5305874159685477e-05,
+      "loss": 3.4685,
+      "step": 1508352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5297488212174957e-05,
+      "loss": 3.4888,
+      "step": 1508864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5289102264664437e-05,
+      "loss": 3.4829,
+      "step": 1509376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5280716317153913e-05,
+      "loss": 3.4745,
+      "step": 1509888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.527234674844713e-05,
+      "loss": 3.4701,
+      "step": 1510400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.526396080093661e-05,
+      "loss": 3.4762,
+      "step": 1510912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5255574853426082e-05,
+      "loss": 3.4712,
+      "step": 1511424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5247188905915566e-05,
+      "loss": 3.466,
+      "step": 1511936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.523881933720878e-05,
+      "loss": 3.4754,
+      "step": 1512448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.523043338969826e-05,
+      "loss": 3.4726,
+      "step": 1512960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5222047442187735e-05,
+      "loss": 3.4702,
+      "step": 1513472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5213661494677215e-05,
+      "loss": 3.4696,
+      "step": 1513984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.520529192597043e-05,
+      "loss": 3.4738,
+      "step": 1514496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.519690597845991e-05,
+      "loss": 3.4687,
+      "step": 1515008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5188520030949387e-05,
+      "loss": 3.475,
+      "step": 1515520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5180134083438867e-05,
+      "loss": 3.4699,
+      "step": 1516032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5171764514732083e-05,
+      "loss": 3.4834,
+      "step": 1516544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5163378567221556e-05,
+      "loss": 3.481,
+      "step": 1517056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5154992619711036e-05,
+      "loss": 3.4779,
+      "step": 1517568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.514660667220052e-05,
+      "loss": 3.4691,
+      "step": 1518080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5138237103493735e-05,
+      "loss": 3.4705,
+      "step": 1518592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.512985115598321e-05,
+      "loss": 3.4775,
+      "step": 1519104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.512146520847269e-05,
+      "loss": 3.4783,
+      "step": 1519616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.511307926096217e-05,
+      "loss": 3.4649,
+      "step": 1520128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5104709692255384e-05,
+      "loss": 3.481,
+      "step": 1520640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.509632374474486e-05,
+      "loss": 3.4738,
+      "step": 1521152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.508793779723434e-05,
+      "loss": 3.4857,
+      "step": 1521664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.507955184972382e-05,
+      "loss": 3.4757,
+      "step": 1522176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.507118228101703e-05,
+      "loss": 3.4817,
+      "step": 1522688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.506279633350651e-05,
+      "loss": 3.4782,
+      "step": 1523200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.505441038599599e-05,
+      "loss": 3.468,
+      "step": 1523712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5046024438485473e-05,
+      "loss": 3.4777,
+      "step": 1524224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5037638490974953e-05,
+      "loss": 3.4861,
+      "step": 1524736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5029268922268162e-05,
+      "loss": 3.4748,
+      "step": 1525248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5020882974757642e-05,
+      "loss": 3.4599,
+      "step": 1525760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5012497027247122e-05,
+      "loss": 3.4748,
+      "step": 1526272
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.856945037841797,
+      "eval_runtime": 304.2365,
+      "eval_samples_per_second": 1254.258,
+      "eval_steps_per_second": 39.196,
+      "step": 1526400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5004111079736602e-05,
+      "loss": 3.4675,
+      "step": 1526784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4995741511029814e-05,
+      "loss": 3.4661,
+      "step": 1527296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4987355563519298e-05,
+      "loss": 3.4826,
+      "step": 1527808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4978969616008774e-05,
+      "loss": 3.4821,
+      "step": 1528320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4970583668498254e-05,
+      "loss": 3.4809,
+      "step": 1528832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4962214099791467e-05,
+      "loss": 3.4736,
+      "step": 1529344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4953828152280947e-05,
+      "loss": 3.4731,
+      "step": 1529856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4945442204770427e-05,
+      "loss": 3.4546,
+      "step": 1530368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4937056257259907e-05,
+      "loss": 3.4812,
+      "step": 1530880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.492868668855312e-05,
+      "loss": 3.4774,
+      "step": 1531392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4920300741042596e-05,
+      "loss": 3.4676,
+      "step": 1531904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4911914793532076e-05,
+      "loss": 3.4771,
+      "step": 1532416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4903545224825288e-05,
+      "loss": 3.4656,
+      "step": 1532928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4895159277314768e-05,
+      "loss": 3.4689,
+      "step": 1533440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4886773329804248e-05,
+      "loss": 3.4635,
+      "step": 1533952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4878387382293728e-05,
+      "loss": 3.4672,
+      "step": 1534464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4870001434783208e-05,
+      "loss": 3.4657,
+      "step": 1534976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4861615487272688e-05,
+      "loss": 3.4659,
+      "step": 1535488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4853229539762168e-05,
+      "loss": 3.47,
+      "step": 1536000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4844843592251648e-05,
+      "loss": 3.4867,
+      "step": 1536512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.483647402354486e-05,
+      "loss": 3.4721,
+      "step": 1537024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.482808807603434e-05,
+      "loss": 3.4775,
+      "step": 1537536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.481970212852382e-05,
+      "loss": 3.4735,
+      "step": 1538048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.48113161810133e-05,
+      "loss": 3.476,
+      "step": 1538560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.480294661230651e-05,
+      "loss": 3.4654,
+      "step": 1539072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.479456066479599e-05,
+      "loss": 3.4708,
+      "step": 1539584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.478617471728547e-05,
+      "loss": 3.4695,
+      "step": 1540096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4777788769774953e-05,
+      "loss": 3.4632,
+      "step": 1540608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.476941920106816e-05,
+      "loss": 3.4603,
+      "step": 1541120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.476103325355764e-05,
+      "loss": 3.4714,
+      "step": 1541632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.475264730604712e-05,
+      "loss": 3.4681,
+      "step": 1542144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4744261358536598e-05,
+      "loss": 3.477,
+      "step": 1542656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4735891789829814e-05,
+      "loss": 3.4716,
+      "step": 1543168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4727505842319294e-05,
+      "loss": 3.4763,
+      "step": 1543680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4719119894808774e-05,
+      "loss": 3.4609,
+      "step": 1544192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4710750326101983e-05,
+      "loss": 3.478,
+      "step": 1544704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4702364378591463e-05,
+      "loss": 3.463,
+      "step": 1545216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4693978431080943e-05,
+      "loss": 3.4693,
+      "step": 1545728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4685592483570423e-05,
+      "loss": 3.4553,
+      "step": 1546240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4677206536059903e-05,
+      "loss": 3.4687,
+      "step": 1546752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4668836967353115e-05,
+      "loss": 3.4664,
+      "step": 1547264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4660451019842595e-05,
+      "loss": 3.4806,
+      "step": 1547776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4652065072332072e-05,
+      "loss": 3.467,
+      "step": 1548288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4643679124821552e-05,
+      "loss": 3.4716,
+      "step": 1548800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4635309556114768e-05,
+      "loss": 3.4753,
+      "step": 1549312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4626923608604248e-05,
+      "loss": 3.4751,
+      "step": 1549824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4618537661093724e-05,
+      "loss": 3.4703,
+      "step": 1550336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4610151713583204e-05,
+      "loss": 3.458,
+      "step": 1550848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4601782144876417e-05,
+      "loss": 3.4519,
+      "step": 1551360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4593396197365897e-05,
+      "loss": 3.4799,
+      "step": 1551872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4585010249855377e-05,
+      "loss": 3.4738,
+      "step": 1552384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4576624302344856e-05,
+      "loss": 3.4646,
+      "step": 1552896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.456825473363807e-05,
+      "loss": 3.4643,
+      "step": 1553408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4559868786127546e-05,
+      "loss": 3.4665,
+      "step": 1553920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4551482838617026e-05,
+      "loss": 3.4472,
+      "step": 1554432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4543096891106505e-05,
+      "loss": 3.4639,
+      "step": 1554944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.453472732239972e-05,
+      "loss": 3.4558,
+      "step": 1555456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4526341374889198e-05,
+      "loss": 3.4644,
+      "step": 1555968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4517955427378678e-05,
+      "loss": 3.4739,
+      "step": 1556480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4509569479868158e-05,
+      "loss": 3.4621,
+      "step": 1556992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.450119991116137e-05,
+      "loss": 3.4538,
+      "step": 1557504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.449281396365085e-05,
+      "loss": 3.4681,
+      "step": 1558016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.448442801614033e-05,
+      "loss": 3.4535,
+      "step": 1558528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.447604206862981e-05,
+      "loss": 3.4571,
+      "step": 1559040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.446767249992302e-05,
+      "loss": 3.4688,
+      "step": 1559552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.44592865524125e-05,
+      "loss": 3.465,
+      "step": 1560064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.445090060490198e-05,
+      "loss": 3.4557,
+      "step": 1560576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.444251465739146e-05,
+      "loss": 3.4563,
+      "step": 1561088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.443414508868467e-05,
+      "loss": 3.4587,
+      "step": 1561600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.442575914117415e-05,
+      "loss": 3.4657,
+      "step": 1562112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.441737319366363e-05,
+      "loss": 3.4644,
+      "step": 1562624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.440898724615311e-05,
+      "loss": 3.4601,
+      "step": 1563136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.440061767744632e-05,
+      "loss": 3.4726,
+      "step": 1563648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.43922317299358e-05,
+      "loss": 3.471,
+      "step": 1564160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4383845782425284e-05,
+      "loss": 3.4626,
+      "step": 1564672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4375459834914764e-05,
+      "loss": 3.4586,
+      "step": 1565184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4367090266207973e-05,
+      "loss": 3.4783,
+      "step": 1565696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4358704318697453e-05,
+      "loss": 3.4664,
+      "step": 1566208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4350318371186933e-05,
+      "loss": 3.4564,
+      "step": 1566720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4341932423676413e-05,
+      "loss": 3.464,
+      "step": 1567232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4333562854969625e-05,
+      "loss": 3.4679,
+      "step": 1567744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4325176907459105e-05,
+      "loss": 3.4734,
+      "step": 1568256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4316790959948585e-05,
+      "loss": 3.4573,
+      "step": 1568768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4308405012438065e-05,
+      "loss": 3.4498,
+      "step": 1569280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4300035443731274e-05,
+      "loss": 3.4552,
+      "step": 1569792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4291649496220754e-05,
+      "loss": 3.458,
+      "step": 1570304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4283263548710238e-05,
+      "loss": 3.469,
+      "step": 1570816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4274877601199717e-05,
+      "loss": 3.4634,
+      "step": 1571328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4266508032492927e-05,
+      "loss": 3.4563,
+      "step": 1571840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4258122084982407e-05,
+      "loss": 3.4648,
+      "step": 1572352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4249736137471887e-05,
+      "loss": 3.4592,
+      "step": 1572864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4241350189961366e-05,
+      "loss": 3.4591,
+      "step": 1573376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.423298062125458e-05,
+      "loss": 3.4581,
+      "step": 1573888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.422459467374406e-05,
+      "loss": 3.4653,
+      "step": 1574400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.421620872623354e-05,
+      "loss": 3.4772,
+      "step": 1574912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.420782277872302e-05,
+      "loss": 3.4657,
+      "step": 1575424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4199453210016228e-05,
+      "loss": 3.445,
+      "step": 1575936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4191067262505708e-05,
+      "loss": 3.4653,
+      "step": 1576448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.418268131499519e-05,
+      "loss": 3.4573,
+      "step": 1576960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.417429536748467e-05,
+      "loss": 3.4691,
+      "step": 1577472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.416592579877788e-05,
+      "loss": 3.4714,
+      "step": 1577984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.415753985126736e-05,
+      "loss": 3.4591,
+      "step": 1578496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.414915390375684e-05,
+      "loss": 3.4526,
+      "step": 1579008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.414076795624632e-05,
+      "loss": 3.4645,
+      "step": 1579520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4132398387539533e-05,
+      "loss": 3.4471,
+      "step": 1580032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4124012440029013e-05,
+      "loss": 3.4641,
+      "step": 1580544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4115626492518493e-05,
+      "loss": 3.4579,
+      "step": 1581056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4107240545007972e-05,
+      "loss": 3.466,
+      "step": 1581568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.409887097630118e-05,
+      "loss": 3.4502,
+      "step": 1582080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.409048502879066e-05,
+      "loss": 3.4653,
+      "step": 1582592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4082099081280145e-05,
+      "loss": 3.4538,
+      "step": 1583104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.407371313376962e-05,
+      "loss": 3.4583,
+      "step": 1583616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4065343565062834e-05,
+      "loss": 3.4686,
+      "step": 1584128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4056957617552314e-05,
+      "loss": 3.461,
+      "step": 1584640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4048571670041794e-05,
+      "loss": 3.4746,
+      "step": 1585152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4040185722531274e-05,
+      "loss": 3.4713,
+      "step": 1585664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4031816153824486e-05,
+      "loss": 3.4597,
+      "step": 1586176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4023430206313966e-05,
+      "loss": 3.4645,
+      "step": 1586688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4015044258803443e-05,
+      "loss": 3.4593,
+      "step": 1587200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4006658311292923e-05,
+      "loss": 3.4632,
+      "step": 1587712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3998288742586135e-05,
+      "loss": 3.4525,
+      "step": 1588224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3989902795075615e-05,
+      "loss": 3.4676,
+      "step": 1588736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3981516847565095e-05,
+      "loss": 3.4562,
+      "step": 1589248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3973130900054575e-05,
+      "loss": 3.4571,
+      "step": 1589760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3964761331347788e-05,
+      "loss": 3.4561,
+      "step": 1590272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3956375383837268e-05,
+      "loss": 3.4662,
+      "step": 1590784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3947989436326744e-05,
+      "loss": 3.4572,
+      "step": 1591296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3939603488816224e-05,
+      "loss": 3.4622,
+      "step": 1591808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.393123392010944e-05,
+      "loss": 3.457,
+      "step": 1592320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3922847972598917e-05,
+      "loss": 3.4697,
+      "step": 1592832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3914462025088396e-05,
+      "loss": 3.4664,
+      "step": 1593344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3906076077577876e-05,
+      "loss": 3.465,
+      "step": 1593856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.389770650887109e-05,
+      "loss": 3.4587,
+      "step": 1594368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.388932056136057e-05,
+      "loss": 3.4592,
+      "step": 1594880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.388093461385005e-05,
+      "loss": 3.4677,
+      "step": 1595392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.387254866633953e-05,
+      "loss": 3.4612,
+      "step": 1595904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.386417909763274e-05,
+      "loss": 3.4529,
+      "step": 1596416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3855793150122218e-05,
+      "loss": 3.4697,
+      "step": 1596928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3847407202611698e-05,
+      "loss": 3.4642,
+      "step": 1597440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3839021255101178e-05,
+      "loss": 3.4764,
+      "step": 1597952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.383065168639439e-05,
+      "loss": 3.4632,
+      "step": 1598464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.382226573888387e-05,
+      "loss": 3.4687,
+      "step": 1598976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.381387979137335e-05,
+      "loss": 3.4657,
+      "step": 1599488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.380549384386283e-05,
+      "loss": 3.4569,
+      "step": 1600000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.379712427515604e-05,
+      "loss": 3.4672,
+      "step": 1600512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3788738327645523e-05,
+      "loss": 3.4698,
+      "step": 1601024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3780352380135002e-05,
+      "loss": 3.465,
+      "step": 1601536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3771966432624482e-05,
+      "loss": 3.4467,
+      "step": 1602048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.376359686391769e-05,
+      "loss": 3.466,
+      "step": 1602560
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8589024543762207,
+      "eval_runtime": 304.3605,
+      "eval_samples_per_second": 1253.747,
+      "eval_steps_per_second": 39.181,
+      "step": 1602720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.375521091640717e-05,
+      "loss": 3.454,
+      "step": 1603072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.374682496889665e-05,
+      "loss": 3.4528,
+      "step": 1603584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.373843902138613e-05,
+      "loss": 3.4683,
+      "step": 1604096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3730053073875615e-05,
+      "loss": 3.4724,
+      "step": 1604608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3721683505168824e-05,
+      "loss": 3.4683,
+      "step": 1605120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3713297557658304e-05,
+      "loss": 3.462,
+      "step": 1605632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3704911610147784e-05,
+      "loss": 3.4609,
+      "step": 1606144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3696525662637264e-05,
+      "loss": 3.4467,
+      "step": 1606656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3688156093930476e-05,
+      "loss": 3.465,
+      "step": 1607168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3679770146419956e-05,
+      "loss": 3.4674,
+      "step": 1607680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3671384198909436e-05,
+      "loss": 3.4572,
+      "step": 1608192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3662998251398916e-05,
+      "loss": 3.4651,
+      "step": 1608704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3654628682692125e-05,
+      "loss": 3.4548,
+      "step": 1609216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3646242735181605e-05,
+      "loss": 3.4571,
+      "step": 1609728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3637856787671085e-05,
+      "loss": 3.4509,
+      "step": 1610240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.362947084016057e-05,
+      "loss": 3.4535,
+      "step": 1610752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3621101271453778e-05,
+      "loss": 3.4518,
+      "step": 1611264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3612715323943257e-05,
+      "loss": 3.4556,
+      "step": 1611776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3604329376432737e-05,
+      "loss": 3.458,
+      "step": 1612288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3595943428922217e-05,
+      "loss": 3.4724,
+      "step": 1612800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.358757386021543e-05,
+      "loss": 3.4652,
+      "step": 1613312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.357918791270491e-05,
+      "loss": 3.4631,
+      "step": 1613824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.357080196519439e-05,
+      "loss": 3.4612,
+      "step": 1614336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.356241601768387e-05,
+      "loss": 3.4667,
+      "step": 1614848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.355404644897708e-05,
+      "loss": 3.4508,
+      "step": 1615360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.354566050146656e-05,
+      "loss": 3.4644,
+      "step": 1615872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.353727455395604e-05,
+      "loss": 3.4546,
+      "step": 1616384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.352888860644552e-05,
+      "loss": 3.4508,
+      "step": 1616896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.352051903773873e-05,
+      "loss": 3.4515,
+      "step": 1617408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.351213309022821e-05,
+      "loss": 3.4545,
+      "step": 1617920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.350374714271769e-05,
+      "loss": 3.4604,
+      "step": 1618432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3495361195207168e-05,
+      "loss": 3.4625,
+      "step": 1618944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3486991626500384e-05,
+      "loss": 3.4643,
+      "step": 1619456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3478605678989863e-05,
+      "loss": 3.4654,
+      "step": 1619968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3470219731479343e-05,
+      "loss": 3.4508,
+      "step": 1620480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.346183378396882e-05,
+      "loss": 3.4625,
+      "step": 1620992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3453464215262032e-05,
+      "loss": 3.4549,
+      "step": 1621504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3445078267751512e-05,
+      "loss": 3.4526,
+      "step": 1622016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3436692320240992e-05,
+      "loss": 3.4499,
+      "step": 1622528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3428306372730472e-05,
+      "loss": 3.454,
+      "step": 1623040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3419936804023685e-05,
+      "loss": 3.4573,
+      "step": 1623552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3411550856513165e-05,
+      "loss": 3.4651,
+      "step": 1624064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.340316490900264e-05,
+      "loss": 3.4573,
+      "step": 1624576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.339477896149212e-05,
+      "loss": 3.4554,
+      "step": 1625088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3386409392785337e-05,
+      "loss": 3.4657,
+      "step": 1625600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3378023445274814e-05,
+      "loss": 3.4621,
+      "step": 1626112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3369637497764294e-05,
+      "loss": 3.4611,
+      "step": 1626624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3361251550253774e-05,
+      "loss": 3.4474,
+      "step": 1627136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3352881981546986e-05,
+      "loss": 3.4361,
+      "step": 1627648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3344496034036463e-05,
+      "loss": 3.4693,
+      "step": 1628160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3336110086525946e-05,
+      "loss": 3.4633,
+      "step": 1628672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3327724139015426e-05,
+      "loss": 3.4544,
+      "step": 1629184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.331935457030864e-05,
+      "loss": 3.4546,
+      "step": 1629696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3310968622798115e-05,
+      "loss": 3.4488,
+      "step": 1630208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3302582675287595e-05,
+      "loss": 3.4406,
+      "step": 1630720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3294196727777075e-05,
+      "loss": 3.4505,
+      "step": 1631232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3285827159070287e-05,
+      "loss": 3.444,
+      "step": 1631744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3277441211559767e-05,
+      "loss": 3.4535,
+      "step": 1632256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3269055264049247e-05,
+      "loss": 3.4635,
+      "step": 1632768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3260669316538727e-05,
+      "loss": 3.4476,
+      "step": 1633280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3252299747831936e-05,
+      "loss": 3.4418,
+      "step": 1633792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3243913800321416e-05,
+      "loss": 3.4579,
+      "step": 1634304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.32355278528109e-05,
+      "loss": 3.4408,
+      "step": 1634816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.322714190530038e-05,
+      "loss": 3.4409,
+      "step": 1635328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.321877233659359e-05,
+      "loss": 3.4585,
+      "step": 1635840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.321038638908307e-05,
+      "loss": 3.4555,
+      "step": 1636352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.320200044157255e-05,
+      "loss": 3.4435,
+      "step": 1636864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.319361449406203e-05,
+      "loss": 3.445,
+      "step": 1637376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.318524492535524e-05,
+      "loss": 3.4475,
+      "step": 1637888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.317685897784472e-05,
+      "loss": 3.4519,
+      "step": 1638400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.31684730303342e-05,
+      "loss": 3.4579,
+      "step": 1638912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.316008708282368e-05,
+      "loss": 3.4462,
+      "step": 1639424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.315171751411689e-05,
+      "loss": 3.4599,
+      "step": 1639936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.314333156660637e-05,
+      "loss": 3.4592,
+      "step": 1640448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3134945619095853e-05,
+      "loss": 3.459,
+      "step": 1640960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3126559671585333e-05,
+      "loss": 3.4437,
+      "step": 1641472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3118190102878542e-05,
+      "loss": 3.4624,
+      "step": 1641984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3109804155368022e-05,
+      "loss": 3.4565,
+      "step": 1642496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3101418207857502e-05,
+      "loss": 3.4436,
+      "step": 1643008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3093032260346982e-05,
+      "loss": 3.4582,
+      "step": 1643520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3084662691640195e-05,
+      "loss": 3.4531,
+      "step": 1644032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3076276744129675e-05,
+      "loss": 3.4635,
+      "step": 1644544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3067890796619155e-05,
+      "loss": 3.4486,
+      "step": 1645056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3059504849108635e-05,
+      "loss": 3.4338,
+      "step": 1645568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3051135280401844e-05,
+      "loss": 3.4462,
+      "step": 1646080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3042749332891324e-05,
+      "loss": 3.4499,
+      "step": 1646592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3034363385380807e-05,
+      "loss": 3.455,
+      "step": 1647104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3025977437870287e-05,
+      "loss": 3.4497,
+      "step": 1647616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3017607869163496e-05,
+      "loss": 3.4463,
+      "step": 1648128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3009221921652976e-05,
+      "loss": 3.4511,
+      "step": 1648640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3000835974142456e-05,
+      "loss": 3.451,
+      "step": 1649152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2992450026631936e-05,
+      "loss": 3.4448,
+      "step": 1649664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.298408045792515e-05,
+      "loss": 3.4508,
+      "step": 1650176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.297569451041463e-05,
+      "loss": 3.4581,
+      "step": 1650688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.296730856290411e-05,
+      "loss": 3.465,
+      "step": 1651200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.295892261539359e-05,
+      "loss": 3.4511,
+      "step": 1651712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2950553046686797e-05,
+      "loss": 3.437,
+      "step": 1652224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2942167099176277e-05,
+      "loss": 3.4547,
+      "step": 1652736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.293378115166576e-05,
+      "loss": 3.4472,
+      "step": 1653248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.292539520415524e-05,
+      "loss": 3.4559,
+      "step": 1653760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.291702563544845e-05,
+      "loss": 3.4602,
+      "step": 1654272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.290863968793793e-05,
+      "loss": 3.4431,
+      "step": 1654784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.290025374042741e-05,
+      "loss": 3.4436,
+      "step": 1655296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2891867792916886e-05,
+      "loss": 3.4541,
+      "step": 1655808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2883498224210102e-05,
+      "loss": 3.4373,
+      "step": 1656320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2875112276699582e-05,
+      "loss": 3.4548,
+      "step": 1656832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2866726329189062e-05,
+      "loss": 3.4397,
+      "step": 1657344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.285834038167854e-05,
+      "loss": 3.455,
+      "step": 1657856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.284997081297175e-05,
+      "loss": 3.4393,
+      "step": 1658368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.284158486546123e-05,
+      "loss": 3.4524,
+      "step": 1658880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2833198917950714e-05,
+      "loss": 3.4488,
+      "step": 1659392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.282481297044019e-05,
+      "loss": 3.4433,
+      "step": 1659904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2816443401733403e-05,
+      "loss": 3.4545,
+      "step": 1660416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2808057454222883e-05,
+      "loss": 3.4483,
+      "step": 1660928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.279967150671236e-05,
+      "loss": 3.4644,
+      "step": 1661440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.279128555920184e-05,
+      "loss": 3.4624,
+      "step": 1661952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2782915990495056e-05,
+      "loss": 3.4458,
+      "step": 1662464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2774530042984536e-05,
+      "loss": 3.4538,
+      "step": 1662976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2766144095474012e-05,
+      "loss": 3.4482,
+      "step": 1663488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2757758147963492e-05,
+      "loss": 3.4521,
+      "step": 1664000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2749388579256705e-05,
+      "loss": 3.4444,
+      "step": 1664512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2741002631746185e-05,
+      "loss": 3.4552,
+      "step": 1665024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2732616684235665e-05,
+      "loss": 3.4467,
+      "step": 1665536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2724230736725145e-05,
+      "loss": 3.4449,
+      "step": 1666048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2715861168018357e-05,
+      "loss": 3.4475,
+      "step": 1666560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2707475220507834e-05,
+      "loss": 3.4531,
+      "step": 1667072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2699089272997314e-05,
+      "loss": 3.4433,
+      "step": 1667584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2690703325486794e-05,
+      "loss": 3.451,
+      "step": 1668096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.268233375678001e-05,
+      "loss": 3.4447,
+      "step": 1668608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2673947809269486e-05,
+      "loss": 3.466,
+      "step": 1669120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2665561861758966e-05,
+      "loss": 3.4483,
+      "step": 1669632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2657175914248446e-05,
+      "loss": 3.4601,
+      "step": 1670144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2648806345541655e-05,
+      "loss": 3.4465,
+      "step": 1670656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.264042039803114e-05,
+      "loss": 3.4476,
+      "step": 1671168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.263203445052062e-05,
+      "loss": 3.4545,
+      "step": 1671680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.26236485030101e-05,
+      "loss": 3.4497,
+      "step": 1672192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2615278934303307e-05,
+      "loss": 3.4434,
+      "step": 1672704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2606892986792787e-05,
+      "loss": 3.4587,
+      "step": 1673216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2598507039282267e-05,
+      "loss": 3.452,
+      "step": 1673728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2590121091771747e-05,
+      "loss": 3.4636,
+      "step": 1674240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.258175152306496e-05,
+      "loss": 3.4488,
+      "step": 1674752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.257336557555444e-05,
+      "loss": 3.4567,
+      "step": 1675264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.256497962804392e-05,
+      "loss": 3.4531,
+      "step": 1675776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.25565936805334e-05,
+      "loss": 3.4482,
+      "step": 1676288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.254822411182661e-05,
+      "loss": 3.4572,
+      "step": 1676800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2539838164316092e-05,
+      "loss": 3.4573,
+      "step": 1677312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2531452216805572e-05,
+      "loss": 3.4589,
+      "step": 1677824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2523066269295052e-05,
+      "loss": 3.4279,
+      "step": 1678336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.251469670058826e-05,
+      "loss": 3.4584,
+      "step": 1678848
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8598439693450928,
+      "eval_runtime": 302.7402,
+      "eval_samples_per_second": 1260.457,
+      "eval_steps_per_second": 39.39,
+      "step": 1679040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.250631075307774e-05,
+      "loss": 3.4482,
+      "step": 1679360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.249792480556722e-05,
+      "loss": 3.442,
+      "step": 1679872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.24895388580567e-05,
+      "loss": 3.4556,
+      "step": 1680384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2481152910546184e-05,
+      "loss": 3.4617,
+      "step": 1680896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2472783341839393e-05,
+      "loss": 3.4558,
+      "step": 1681408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2464397394328873e-05,
+      "loss": 3.4471,
+      "step": 1681920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2456011446818353e-05,
+      "loss": 3.45,
+      "step": 1682432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2447625499307833e-05,
+      "loss": 3.4356,
+      "step": 1682944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2439255930601046e-05,
+      "loss": 3.4529,
+      "step": 1683456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2430869983090526e-05,
+      "loss": 3.4535,
+      "step": 1683968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2422484035580006e-05,
+      "loss": 3.4514,
+      "step": 1684480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2414098088069486e-05,
+      "loss": 3.4497,
+      "step": 1684992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2405728519362695e-05,
+      "loss": 3.446,
+      "step": 1685504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2397342571852175e-05,
+      "loss": 3.4449,
+      "step": 1686016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2388956624341655e-05,
+      "loss": 3.4398,
+      "step": 1686528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2380570676831138e-05,
+      "loss": 3.4383,
+      "step": 1687040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2372201108124347e-05,
+      "loss": 3.4438,
+      "step": 1687552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2363815160613827e-05,
+      "loss": 3.4432,
+      "step": 1688064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2355429213103307e-05,
+      "loss": 3.4482,
+      "step": 1688576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2347059644396516e-05,
+      "loss": 3.4609,
+      "step": 1689088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2338673696886e-05,
+      "loss": 3.4578,
+      "step": 1689600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.233028774937548e-05,
+      "loss": 3.4505,
+      "step": 1690112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.232190180186496e-05,
+      "loss": 3.4492,
+      "step": 1690624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.231353223315817e-05,
+      "loss": 3.455,
+      "step": 1691136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.230514628564765e-05,
+      "loss": 3.4402,
+      "step": 1691648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.229676033813713e-05,
+      "loss": 3.4567,
+      "step": 1692160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2288374390626608e-05,
+      "loss": 3.4438,
+      "step": 1692672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.228000482191982e-05,
+      "loss": 3.4368,
+      "step": 1693184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.22716188744093e-05,
+      "loss": 3.4453,
+      "step": 1693696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.226323292689878e-05,
+      "loss": 3.4453,
+      "step": 1694208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2254846979388257e-05,
+      "loss": 3.4462,
+      "step": 1694720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.224647741068147e-05,
+      "loss": 3.4546,
+      "step": 1695232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2238091463170953e-05,
+      "loss": 3.4505,
+      "step": 1695744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2229705515660433e-05,
+      "loss": 3.4524,
+      "step": 1696256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.222131956814991e-05,
+      "loss": 3.437,
+      "step": 1696768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2212949999443122e-05,
+      "loss": 3.4557,
+      "step": 1697280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2204564051932602e-05,
+      "loss": 3.4429,
+      "step": 1697792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2196178104422082e-05,
+      "loss": 3.4375,
+      "step": 1698304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2187792156911562e-05,
+      "loss": 3.4445,
+      "step": 1698816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2179422588204774e-05,
+      "loss": 3.4374,
+      "step": 1699328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2171036640694254e-05,
+      "loss": 3.4477,
+      "step": 1699840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.216265069318373e-05,
+      "loss": 3.4574,
+      "step": 1700352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.215426474567321e-05,
+      "loss": 3.4438,
+      "step": 1700864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2145895176966423e-05,
+      "loss": 3.4466,
+      "step": 1701376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2137509229455903e-05,
+      "loss": 3.4541,
+      "step": 1701888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2129123281945383e-05,
+      "loss": 3.4534,
+      "step": 1702400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2120737334434863e-05,
+      "loss": 3.4507,
+      "step": 1702912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2112367765728076e-05,
+      "loss": 3.4378,
+      "step": 1703424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2103981818217552e-05,
+      "loss": 3.4177,
+      "step": 1703936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2095595870707032e-05,
+      "loss": 3.4612,
+      "step": 1704448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2087209923196516e-05,
+      "loss": 3.4526,
+      "step": 1704960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2078840354489728e-05,
+      "loss": 3.4471,
+      "step": 1705472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2070454406979205e-05,
+      "loss": 3.4414,
+      "step": 1705984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2062068459468685e-05,
+      "loss": 3.4398,
+      "step": 1706496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2053682511958165e-05,
+      "loss": 3.4296,
+      "step": 1707008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2045312943251377e-05,
+      "loss": 3.4382,
+      "step": 1707520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2036926995740857e-05,
+      "loss": 3.4365,
+      "step": 1708032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2028541048230337e-05,
+      "loss": 3.4415,
+      "step": 1708544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2020155100719817e-05,
+      "loss": 3.4526,
+      "step": 1709056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2011769153209297e-05,
+      "loss": 3.4363,
+      "step": 1709568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2003399584502506e-05,
+      "loss": 3.4327,
+      "step": 1710080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1995013636991986e-05,
+      "loss": 3.4465,
+      "step": 1710592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.198662768948147e-05,
+      "loss": 3.4342,
+      "step": 1711104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.197824174197095e-05,
+      "loss": 3.4325,
+      "step": 1711616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.196987217326416e-05,
+      "loss": 3.4439,
+      "step": 1712128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1961486225753638e-05,
+      "loss": 3.4439,
+      "step": 1712640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1953100278243118e-05,
+      "loss": 3.4321,
+      "step": 1713152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1944714330732598e-05,
+      "loss": 3.4327,
+      "step": 1713664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.193634476202581e-05,
+      "loss": 3.4342,
+      "step": 1714176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.192795881451529e-05,
+      "loss": 3.4461,
+      "step": 1714688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.191957286700477e-05,
+      "loss": 3.4418,
+      "step": 1715200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.191118691949425e-05,
+      "loss": 3.4387,
+      "step": 1715712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.190281735078746e-05,
+      "loss": 3.444,
+      "step": 1716224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.189443140327694e-05,
+      "loss": 3.4468,
+      "step": 1716736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1886045455766423e-05,
+      "loss": 3.4498,
+      "step": 1717248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1877659508255903e-05,
+      "loss": 3.4347,
+      "step": 1717760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1869289939549112e-05,
+      "loss": 3.4487,
+      "step": 1718272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1860903992038592e-05,
+      "loss": 3.4487,
+      "step": 1718784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1852518044528072e-05,
+      "loss": 3.4297,
+      "step": 1719296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1844132097017552e-05,
+      "loss": 3.4473,
+      "step": 1719808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1835762528310764e-05,
+      "loss": 3.4428,
+      "step": 1720320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1827376580800244e-05,
+      "loss": 3.448,
+      "step": 1720832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1818990633289724e-05,
+      "loss": 3.439,
+      "step": 1721344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1810604685779204e-05,
+      "loss": 3.4235,
+      "step": 1721856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1802235117072413e-05,
+      "loss": 3.4338,
+      "step": 1722368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1793849169561893e-05,
+      "loss": 3.4361,
+      "step": 1722880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1785463222051373e-05,
+      "loss": 3.4457,
+      "step": 1723392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1777077274540857e-05,
+      "loss": 3.4372,
+      "step": 1723904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1768707705834066e-05,
+      "loss": 3.4366,
+      "step": 1724416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1760321758323546e-05,
+      "loss": 3.44,
+      "step": 1724928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1751935810813026e-05,
+      "loss": 3.4432,
+      "step": 1725440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1743549863302506e-05,
+      "loss": 3.4298,
+      "step": 1725952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1735180294595718e-05,
+      "loss": 3.4377,
+      "step": 1726464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1726794347085198e-05,
+      "loss": 3.4481,
+      "step": 1726976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1718408399574678e-05,
+      "loss": 3.4504,
+      "step": 1727488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1710022452064154e-05,
+      "loss": 3.4424,
+      "step": 1728000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1701652883357367e-05,
+      "loss": 3.4269,
+      "step": 1728512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1693266935846847e-05,
+      "loss": 3.4418,
+      "step": 1729024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1684880988336327e-05,
+      "loss": 3.4324,
+      "step": 1729536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1676495040825807e-05,
+      "loss": 3.4485,
+      "step": 1730048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.166812547211902e-05,
+      "loss": 3.4497,
+      "step": 1730560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.16597395246085e-05,
+      "loss": 3.437,
+      "step": 1731072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.165135357709798e-05,
+      "loss": 3.4322,
+      "step": 1731584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1642967629587456e-05,
+      "loss": 3.4404,
+      "step": 1732096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.163459806088067e-05,
+      "loss": 3.4265,
+      "step": 1732608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.162621211337015e-05,
+      "loss": 3.4445,
+      "step": 1733120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1617826165859628e-05,
+      "loss": 3.427,
+      "step": 1733632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1609440218349108e-05,
+      "loss": 3.4453,
+      "step": 1734144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.160107064964232e-05,
+      "loss": 3.4287,
+      "step": 1734656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.15926847021318e-05,
+      "loss": 3.4386,
+      "step": 1735168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.158429875462128e-05,
+      "loss": 3.4372,
+      "step": 1735680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.157591280711076e-05,
+      "loss": 3.4371,
+      "step": 1736192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1567543238403973e-05,
+      "loss": 3.4437,
+      "step": 1736704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1559157290893453e-05,
+      "loss": 3.4349,
+      "step": 1737216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.155077134338293e-05,
+      "loss": 3.4539,
+      "step": 1737728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.154238539587241e-05,
+      "loss": 3.449,
+      "step": 1738240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1534015827165625e-05,
+      "loss": 3.4385,
+      "step": 1738752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1525629879655102e-05,
+      "loss": 3.4404,
+      "step": 1739264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1517243932144582e-05,
+      "loss": 3.4361,
+      "step": 1739776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1508857984634062e-05,
+      "loss": 3.4402,
+      "step": 1740288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1500488415927274e-05,
+      "loss": 3.4374,
+      "step": 1740800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1492102468416754e-05,
+      "loss": 3.4389,
+      "step": 1741312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1483716520906234e-05,
+      "loss": 3.4379,
+      "step": 1741824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1475330573395714e-05,
+      "loss": 3.4373,
+      "step": 1742336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1466961004688927e-05,
+      "loss": 3.4374,
+      "step": 1742848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1458575057178403e-05,
+      "loss": 3.4386,
+      "step": 1743360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1450189109667883e-05,
+      "loss": 3.4362,
+      "step": 1743872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1441803162157363e-05,
+      "loss": 3.4394,
+      "step": 1744384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1433433593450576e-05,
+      "loss": 3.4319,
+      "step": 1744896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1425047645940056e-05,
+      "loss": 3.4533,
+      "step": 1745408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1416661698429536e-05,
+      "loss": 3.4412,
+      "step": 1745920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1408275750919015e-05,
+      "loss": 3.4505,
+      "step": 1746432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1399906182212225e-05,
+      "loss": 3.4291,
+      "step": 1746944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1391520234701708e-05,
+      "loss": 3.4401,
+      "step": 1747456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1383134287191188e-05,
+      "loss": 3.4433,
+      "step": 1747968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1374748339680668e-05,
+      "loss": 3.4427,
+      "step": 1748480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1366378770973877e-05,
+      "loss": 3.4287,
+      "step": 1748992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1357992823463357e-05,
+      "loss": 3.4494,
+      "step": 1749504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1349606875952837e-05,
+      "loss": 3.4383,
+      "step": 1750016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1341220928442317e-05,
+      "loss": 3.4548,
+      "step": 1750528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.133285135973553e-05,
+      "loss": 3.4343,
+      "step": 1751040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.132446541222501e-05,
+      "loss": 3.4469,
+      "step": 1751552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.131607946471449e-05,
+      "loss": 3.4421,
+      "step": 1752064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.130769351720397e-05,
+      "loss": 3.4369,
+      "step": 1752576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1299323948497178e-05,
+      "loss": 3.4432,
+      "step": 1753088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.129093800098666e-05,
+      "loss": 3.4461,
+      "step": 1753600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.128255205347614e-05,
+      "loss": 3.4475,
+      "step": 1754112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.127416610596562e-05,
+      "loss": 3.4229,
+      "step": 1754624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.126579653725883e-05,
+      "loss": 3.4465,
+      "step": 1755136
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8604748249053955,
+      "eval_runtime": 303.0497,
+      "eval_samples_per_second": 1259.17,
+      "eval_steps_per_second": 39.35,
+      "step": 1755360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.125741058974831e-05,
+      "loss": 3.4355,
+      "step": 1755648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.124902464223779e-05,
+      "loss": 3.4318,
+      "step": 1756160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.124063869472727e-05,
+      "loss": 3.4414,
+      "step": 1756672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.123225274721675e-05,
+      "loss": 3.4493,
+      "step": 1757184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1223883178509963e-05,
+      "loss": 3.4464,
+      "step": 1757696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1215497230999443e-05,
+      "loss": 3.4379,
+      "step": 1758208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1207111283488923e-05,
+      "loss": 3.442,
+      "step": 1758720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1198725335978403e-05,
+      "loss": 3.4269,
+      "step": 1759232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1190355767271612e-05,
+      "loss": 3.4426,
+      "step": 1759744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1181969819761095e-05,
+      "loss": 3.4415,
+      "step": 1760256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1173583872250575e-05,
+      "loss": 3.4405,
+      "step": 1760768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1165197924740052e-05,
+      "loss": 3.4411,
+      "step": 1761280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1156828356033264e-05,
+      "loss": 3.4299,
+      "step": 1761792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1148442408522744e-05,
+      "loss": 3.431,
+      "step": 1762304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1140056461012224e-05,
+      "loss": 3.4335,
+      "step": 1762816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1131670513501704e-05,
+      "loss": 3.4273,
+      "step": 1763328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1123300944794917e-05,
+      "loss": 3.4309,
+      "step": 1763840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1114914997284397e-05,
+      "loss": 3.432,
+      "step": 1764352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1106529049773876e-05,
+      "loss": 3.4375,
+      "step": 1764864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1098143102263353e-05,
+      "loss": 3.4476,
+      "step": 1765376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1089773533556566e-05,
+      "loss": 3.4485,
+      "step": 1765888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.108138758604605e-05,
+      "loss": 3.4429,
+      "step": 1766400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1073001638535525e-05,
+      "loss": 3.4341,
+      "step": 1766912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1064615691025005e-05,
+      "loss": 3.4413,
+      "step": 1767424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1056246122318218e-05,
+      "loss": 3.4305,
+      "step": 1767936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1047860174807698e-05,
+      "loss": 3.4445,
+      "step": 1768448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1039474227297178e-05,
+      "loss": 3.4347,
+      "step": 1768960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1031088279786658e-05,
+      "loss": 3.4262,
+      "step": 1769472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.102271871107987e-05,
+      "loss": 3.4344,
+      "step": 1769984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.101433276356935e-05,
+      "loss": 3.4285,
+      "step": 1770496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1005946816058827e-05,
+      "loss": 3.4325,
+      "step": 1771008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0997560868548307e-05,
+      "loss": 3.4443,
+      "step": 1771520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.098919129984152e-05,
+      "loss": 3.4388,
+      "step": 1772032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0980805352331e-05,
+      "loss": 3.4398,
+      "step": 1772544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.097241940482048e-05,
+      "loss": 3.4263,
+      "step": 1773056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.096403345730996e-05,
+      "loss": 3.4468,
+      "step": 1773568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.095566388860317e-05,
+      "loss": 3.4361,
+      "step": 1774080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0947277941092648e-05,
+      "loss": 3.4262,
+      "step": 1774592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.093889199358213e-05,
+      "loss": 3.4336,
+      "step": 1775104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.093050604607161e-05,
+      "loss": 3.4275,
+      "step": 1775616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0922136477364824e-05,
+      "loss": 3.4346,
+      "step": 1776128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.09137505298543e-05,
+      "loss": 3.4402,
+      "step": 1776640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.090536458234378e-05,
+      "loss": 3.4364,
+      "step": 1777152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.089697863483326e-05,
+      "loss": 3.4322,
+      "step": 1777664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0888609066126473e-05,
+      "loss": 3.4442,
+      "step": 1778176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0880223118615953e-05,
+      "loss": 3.4417,
+      "step": 1778688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0871837171105433e-05,
+      "loss": 3.443,
+      "step": 1779200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0863451223594913e-05,
+      "loss": 3.4255,
+      "step": 1779712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0855081654888122e-05,
+      "loss": 3.406,
+      "step": 1780224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0846695707377602e-05,
+      "loss": 3.4521,
+      "step": 1780736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0838309759867082e-05,
+      "loss": 3.4356,
+      "step": 1781248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0829923812356565e-05,
+      "loss": 3.4368,
+      "step": 1781760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0821554243649774e-05,
+      "loss": 3.4342,
+      "step": 1782272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0813168296139254e-05,
+      "loss": 3.4226,
+      "step": 1782784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0804782348628734e-05,
+      "loss": 3.4214,
+      "step": 1783296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0796396401118214e-05,
+      "loss": 3.4284,
+      "step": 1783808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0788026832411427e-05,
+      "loss": 3.4227,
+      "step": 1784320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0779640884900907e-05,
+      "loss": 3.4275,
+      "step": 1784832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0771254937390386e-05,
+      "loss": 3.4442,
+      "step": 1785344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0762868989879866e-05,
+      "loss": 3.4222,
+      "step": 1785856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0754499421173076e-05,
+      "loss": 3.4209,
+      "step": 1786368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0746113473662555e-05,
+      "loss": 3.4424,
+      "step": 1786880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0737727526152035e-05,
+      "loss": 3.418,
+      "step": 1787392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.072934157864152e-05,
+      "loss": 3.4207,
+      "step": 1787904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0720972009934728e-05,
+      "loss": 3.4332,
+      "step": 1788416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0712586062424208e-05,
+      "loss": 3.4329,
+      "step": 1788928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0704200114913688e-05,
+      "loss": 3.4201,
+      "step": 1789440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0695814167403168e-05,
+      "loss": 3.4275,
+      "step": 1789952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.068744459869638e-05,
+      "loss": 3.4197,
+      "step": 1790464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.067905865118586e-05,
+      "loss": 3.4335,
+      "step": 1790976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.067067270367534e-05,
+      "loss": 3.4319,
+      "step": 1791488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.066228675616482e-05,
+      "loss": 3.4259,
+      "step": 1792000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.065391718745803e-05,
+      "loss": 3.4353,
+      "step": 1792512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.064553123994751e-05,
+      "loss": 3.433,
+      "step": 1793024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.063714529243699e-05,
+      "loss": 3.4433,
+      "step": 1793536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0628759344926472e-05,
+      "loss": 3.4237,
+      "step": 1794048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.062038977621968e-05,
+      "loss": 3.4371,
+      "step": 1794560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.061200382870916e-05,
+      "loss": 3.4405,
+      "step": 1795072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.060361788119864e-05,
+      "loss": 3.4152,
+      "step": 1795584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.059523193368812e-05,
+      "loss": 3.4385,
+      "step": 1796096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0586862364981334e-05,
+      "loss": 3.4315,
+      "step": 1796608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0578476417470814e-05,
+      "loss": 3.4385,
+      "step": 1797120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0570090469960294e-05,
+      "loss": 3.4298,
+      "step": 1797632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0561704522449774e-05,
+      "loss": 3.4154,
+      "step": 1798144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0553334953742983e-05,
+      "loss": 3.4171,
+      "step": 1798656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0544949006232463e-05,
+      "loss": 3.4263,
+      "step": 1799168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0536563058721943e-05,
+      "loss": 3.4379,
+      "step": 1799680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0528177111211423e-05,
+      "loss": 3.4222,
+      "step": 1800192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0519807542504635e-05,
+      "loss": 3.4283,
+      "step": 1800704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0511421594994115e-05,
+      "loss": 3.427,
+      "step": 1801216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0503035647483595e-05,
+      "loss": 3.4308,
+      "step": 1801728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.049464969997307e-05,
+      "loss": 3.4221,
+      "step": 1802240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0486280131266288e-05,
+      "loss": 3.4243,
+      "step": 1802752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0477894183755768e-05,
+      "loss": 3.4377,
+      "step": 1803264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0469508236245247e-05,
+      "loss": 3.4392,
+      "step": 1803776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0461122288734724e-05,
+      "loss": 3.4343,
+      "step": 1804288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0452752720027937e-05,
+      "loss": 3.4167,
+      "step": 1804800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0444366772517416e-05,
+      "loss": 3.4311,
+      "step": 1805312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0435980825006896e-05,
+      "loss": 3.4204,
+      "step": 1805824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0427594877496376e-05,
+      "loss": 3.4384,
+      "step": 1806336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.041922530878959e-05,
+      "loss": 3.4326,
+      "step": 1806848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.041083936127907e-05,
+      "loss": 3.4303,
+      "step": 1807360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0402453413768545e-05,
+      "loss": 3.4256,
+      "step": 1807872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0394067466258025e-05,
+      "loss": 3.4231,
+      "step": 1808384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.038569789755124e-05,
+      "loss": 3.4201,
+      "step": 1808896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.037731195004072e-05,
+      "loss": 3.432,
+      "step": 1809408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0368926002530198e-05,
+      "loss": 3.4189,
+      "step": 1809920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0360540055019678e-05,
+      "loss": 3.4355,
+      "step": 1810432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.035217048631289e-05,
+      "loss": 3.4214,
+      "step": 1810944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.034378453880237e-05,
+      "loss": 3.4266,
+      "step": 1811456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.033539859129185e-05,
+      "loss": 3.4222,
+      "step": 1811968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.032701264378133e-05,
+      "loss": 3.4223,
+      "step": 1812480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0318643075074543e-05,
+      "loss": 3.4357,
+      "step": 1812992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.031025712756402e-05,
+      "loss": 3.425,
+      "step": 1813504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.03018711800535e-05,
+      "loss": 3.4428,
+      "step": 1814016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.029348523254298e-05,
+      "loss": 3.4397,
+      "step": 1814528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0285115663836195e-05,
+      "loss": 3.4261,
+      "step": 1815040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.027672971632567e-05,
+      "loss": 3.4293,
+      "step": 1815552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.026834376881515e-05,
+      "loss": 3.4262,
+      "step": 1816064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.025995782130463e-05,
+      "loss": 3.4296,
+      "step": 1816576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.025158825259784e-05,
+      "loss": 3.4288,
+      "step": 1817088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.024320230508732e-05,
+      "loss": 3.4264,
+      "step": 1817600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0234816357576804e-05,
+      "loss": 3.4261,
+      "step": 1818112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0226430410066284e-05,
+      "loss": 3.426,
+      "step": 1818624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0218060841359493e-05,
+      "loss": 3.4279,
+      "step": 1819136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0209674893848973e-05,
+      "loss": 3.4285,
+      "step": 1819648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0201288946338453e-05,
+      "loss": 3.4228,
+      "step": 1820160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0192902998827933e-05,
+      "loss": 3.4293,
+      "step": 1820672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0184533430121145e-05,
+      "loss": 3.4257,
+      "step": 1821184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0176147482610625e-05,
+      "loss": 3.4404,
+      "step": 1821696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0167761535100105e-05,
+      "loss": 3.4306,
+      "step": 1822208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0159375587589585e-05,
+      "loss": 3.4356,
+      "step": 1822720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0151006018882794e-05,
+      "loss": 3.4204,
+      "step": 1823232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0142620071372274e-05,
+      "loss": 3.4308,
+      "step": 1823744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0134234123861757e-05,
+      "loss": 3.4317,
+      "step": 1824256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0125848176351237e-05,
+      "loss": 3.4292,
+      "step": 1824768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0117478607644446e-05,
+      "loss": 3.4216,
+      "step": 1825280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0109092660133926e-05,
+      "loss": 3.4333,
+      "step": 1825792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0100706712623406e-05,
+      "loss": 3.4277,
+      "step": 1826304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0092320765112886e-05,
+      "loss": 3.4484,
+      "step": 1826816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.00839511964061e-05,
+      "loss": 3.423,
+      "step": 1827328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.007556524889558e-05,
+      "loss": 3.4375,
+      "step": 1827840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.006717930138506e-05,
+      "loss": 3.432,
+      "step": 1828352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.005879335387454e-05,
+      "loss": 3.426,
+      "step": 1828864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0050423785167748e-05,
+      "loss": 3.4313,
+      "step": 1829376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0042037837657228e-05,
+      "loss": 3.4352,
+      "step": 1829888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.003365189014671e-05,
+      "loss": 3.4342,
+      "step": 1830400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.002526594263619e-05,
+      "loss": 3.4111,
+      "step": 1830912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.00168963739294e-05,
+      "loss": 3.4372,
+      "step": 1831424
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8622753620147705,
+      "eval_runtime": 303.6666,
+      "eval_samples_per_second": 1256.612,
+      "eval_steps_per_second": 39.27,
+      "step": 1831680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000851042641888e-05,
+      "loss": 3.4221,
+      "step": 1831936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000012447890836e-05,
+      "loss": 3.4207,
+      "step": 1832448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999173853139784e-05,
+      "loss": 3.4296,
+      "step": 1832960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9983368962691052e-05,
+      "loss": 3.4404,
+      "step": 1833472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9974983015180532e-05,
+      "loss": 3.4358,
+      "step": 1833984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9966597067670012e-05,
+      "loss": 3.4249,
+      "step": 1834496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9958211120159492e-05,
+      "loss": 3.4346,
+      "step": 1835008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.99498415514527e-05,
+      "loss": 3.4175,
+      "step": 1835520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.994145560394218e-05,
+      "loss": 3.4265,
+      "step": 1836032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9933069656431665e-05,
+      "loss": 3.4342,
+      "step": 1836544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9924683708921145e-05,
+      "loss": 3.4295,
+      "step": 1837056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9916314140214354e-05,
+      "loss": 3.428,
+      "step": 1837568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9907944571507563e-05,
+      "loss": 3.4264,
+      "step": 1838080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9899558623997043e-05,
+      "loss": 3.4192,
+      "step": 1838592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9891172676486526e-05,
+      "loss": 3.4206,
+      "step": 1839104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9882786728976006e-05,
+      "loss": 3.4158,
+      "step": 1839616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9874400781465486e-05,
+      "loss": 3.4195,
+      "step": 1840128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9866014833954966e-05,
+      "loss": 3.4236,
+      "step": 1840640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9857628886444443e-05,
+      "loss": 3.4236,
+      "step": 1841152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9849242938933923e-05,
+      "loss": 3.4362,
+      "step": 1841664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9840873370227135e-05,
+      "loss": 3.4367,
+      "step": 1842176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.983248742271662e-05,
+      "loss": 3.4327,
+      "step": 1842688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9824101475206095e-05,
+      "loss": 3.4271,
+      "step": 1843200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9815715527695575e-05,
+      "loss": 3.429,
+      "step": 1843712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9807345958988787e-05,
+      "loss": 3.4276,
+      "step": 1844224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9798960011478264e-05,
+      "loss": 3.4265,
+      "step": 1844736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9790574063967744e-05,
+      "loss": 3.4255,
+      "step": 1845248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9782188116457227e-05,
+      "loss": 3.4206,
+      "step": 1845760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.977381854775044e-05,
+      "loss": 3.4226,
+      "step": 1846272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9765432600239916e-05,
+      "loss": 3.4217,
+      "step": 1846784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9757046652729396e-05,
+      "loss": 3.4194,
+      "step": 1847296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9748660705218876e-05,
+      "loss": 3.4319,
+      "step": 1847808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.974029113651209e-05,
+      "loss": 3.4318,
+      "step": 1848320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.973190518900157e-05,
+      "loss": 3.4292,
+      "step": 1848832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.972351924149105e-05,
+      "loss": 3.4197,
+      "step": 1849344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.971513329398053e-05,
+      "loss": 3.4295,
+      "step": 1849856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9706763725273738e-05,
+      "loss": 3.4252,
+      "step": 1850368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9698377777763218e-05,
+      "loss": 3.4178,
+      "step": 1850880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9689991830252698e-05,
+      "loss": 3.4238,
+      "step": 1851392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.968160588274218e-05,
+      "loss": 3.4133,
+      "step": 1851904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.967323631403539e-05,
+      "loss": 3.421,
+      "step": 1852416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966485036652487e-05,
+      "loss": 3.4318,
+      "step": 1852928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.965646441901435e-05,
+      "loss": 3.4238,
+      "step": 1853440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.964807847150383e-05,
+      "loss": 3.4246,
+      "step": 1853952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9639708902797042e-05,
+      "loss": 3.4346,
+      "step": 1854464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9631322955286522e-05,
+      "loss": 3.4287,
+      "step": 1854976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9622937007776002e-05,
+      "loss": 3.4293,
+      "step": 1855488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9614551060265482e-05,
+      "loss": 3.4206,
+      "step": 1856000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.960618149155869e-05,
+      "loss": 3.3957,
+      "step": 1856512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.959779554404817e-05,
+      "loss": 3.4387,
+      "step": 1857024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.958940959653765e-05,
+      "loss": 3.4238,
+      "step": 1857536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9581040027830864e-05,
+      "loss": 3.4277,
+      "step": 1858048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9572654080320344e-05,
+      "loss": 3.4244,
+      "step": 1858560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9564268132809824e-05,
+      "loss": 3.4135,
+      "step": 1859072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9555882185299304e-05,
+      "loss": 3.4078,
+      "step": 1859584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9547512616592513e-05,
+      "loss": 3.4194,
+      "step": 1860096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9539126669081996e-05,
+      "loss": 3.4132,
+      "step": 1860608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9530740721571476e-05,
+      "loss": 3.4166,
+      "step": 1861120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9522354774060956e-05,
+      "loss": 3.4342,
+      "step": 1861632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9513985205354165e-05,
+      "loss": 3.4167,
+      "step": 1862144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9505599257843645e-05,
+      "loss": 3.4105,
+      "step": 1862656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9497213310333125e-05,
+      "loss": 3.4331,
+      "step": 1863168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9488827362822605e-05,
+      "loss": 3.4035,
+      "step": 1863680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9480457794115817e-05,
+      "loss": 3.4076,
+      "step": 1864192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9472071846605297e-05,
+      "loss": 3.4255,
+      "step": 1864704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9463685899094777e-05,
+      "loss": 3.4234,
+      "step": 1865216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9455299951584257e-05,
+      "loss": 3.4096,
+      "step": 1865728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9446930382877466e-05,
+      "loss": 3.4117,
+      "step": 1866240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.943854443536695e-05,
+      "loss": 3.4112,
+      "step": 1866752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.943015848785643e-05,
+      "loss": 3.422,
+      "step": 1867264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.942177254034591e-05,
+      "loss": 3.4216,
+      "step": 1867776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.941340297163912e-05,
+      "loss": 3.4162,
+      "step": 1868288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.94050170241286e-05,
+      "loss": 3.4224,
+      "step": 1868800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.939663107661808e-05,
+      "loss": 3.4227,
+      "step": 1869312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.938824512910756e-05,
+      "loss": 3.4344,
+      "step": 1869824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.937987556040077e-05,
+      "loss": 3.4119,
+      "step": 1870336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.937148961289025e-05,
+      "loss": 3.4279,
+      "step": 1870848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.936310366537973e-05,
+      "loss": 3.428,
+      "step": 1871360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.935471771786921e-05,
+      "loss": 3.4034,
+      "step": 1871872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.934634814916242e-05,
+      "loss": 3.4273,
+      "step": 1872384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9337962201651903e-05,
+      "loss": 3.4242,
+      "step": 1872896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9329576254141383e-05,
+      "loss": 3.4273,
+      "step": 1873408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9321190306630863e-05,
+      "loss": 3.42,
+      "step": 1873920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9312820737924072e-05,
+      "loss": 3.4084,
+      "step": 1874432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9304434790413552e-05,
+      "loss": 3.4019,
+      "step": 1874944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9296048842903032e-05,
+      "loss": 3.4148,
+      "step": 1875456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9287662895392512e-05,
+      "loss": 3.4257,
+      "step": 1875968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9279293326685725e-05,
+      "loss": 3.4159,
+      "step": 1876480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9270907379175205e-05,
+      "loss": 3.4128,
+      "step": 1876992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9262521431664685e-05,
+      "loss": 3.4135,
+      "step": 1877504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.925413548415416e-05,
+      "loss": 3.4257,
+      "step": 1878016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9245765915447374e-05,
+      "loss": 3.4124,
+      "step": 1878528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9237379967936857e-05,
+      "loss": 3.4149,
+      "step": 1879040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9228994020426337e-05,
+      "loss": 3.4237,
+      "step": 1879552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9220608072915814e-05,
+      "loss": 3.4292,
+      "step": 1880064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9212238504209026e-05,
+      "loss": 3.4214,
+      "step": 1880576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9203852556698506e-05,
+      "loss": 3.408,
+      "step": 1881088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9195466609187986e-05,
+      "loss": 3.4159,
+      "step": 1881600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9187080661677466e-05,
+      "loss": 3.4156,
+      "step": 1882112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.917871109297068e-05,
+      "loss": 3.4253,
+      "step": 1882624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.917032514546016e-05,
+      "loss": 3.4233,
+      "step": 1883136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9161939197949635e-05,
+      "loss": 3.4217,
+      "step": 1883648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9153553250439115e-05,
+      "loss": 3.4123,
+      "step": 1884160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9145183681732327e-05,
+      "loss": 3.4113,
+      "step": 1884672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.913679773422181e-05,
+      "loss": 3.4142,
+      "step": 1885184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9128411786711287e-05,
+      "loss": 3.4178,
+      "step": 1885696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9120025839200767e-05,
+      "loss": 3.4075,
+      "step": 1886208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.911165627049398e-05,
+      "loss": 3.4267,
+      "step": 1886720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.910327032298346e-05,
+      "loss": 3.4121,
+      "step": 1887232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9094884375472936e-05,
+      "loss": 3.4123,
+      "step": 1887744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.908649842796242e-05,
+      "loss": 3.4121,
+      "step": 1888256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9078128859255632e-05,
+      "loss": 3.4086,
+      "step": 1888768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.906974291174511e-05,
+      "loss": 3.4284,
+      "step": 1889280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.906135696423459e-05,
+      "loss": 3.4115,
+      "step": 1889792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.905297101672407e-05,
+      "loss": 3.4317,
+      "step": 1890304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.904460144801728e-05,
+      "loss": 3.4289,
+      "step": 1890816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.903621550050676e-05,
+      "loss": 3.4186,
+      "step": 1891328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.902782955299624e-05,
+      "loss": 3.4164,
+      "step": 1891840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.901944360548572e-05,
+      "loss": 3.4204,
+      "step": 1892352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9011074036778933e-05,
+      "loss": 3.4176,
+      "step": 1892864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.900268808926841e-05,
+      "loss": 3.4207,
+      "step": 1893376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.899430214175789e-05,
+      "loss": 3.4143,
+      "step": 1893888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8985916194247373e-05,
+      "loss": 3.4174,
+      "step": 1894400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8977546625540582e-05,
+      "loss": 3.4126,
+      "step": 1894912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8969160678030062e-05,
+      "loss": 3.4176,
+      "step": 1895424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8960774730519542e-05,
+      "loss": 3.4163,
+      "step": 1895936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8952388783009022e-05,
+      "loss": 3.4128,
+      "step": 1896448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8944019214302235e-05,
+      "loss": 3.4175,
+      "step": 1896960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8935633266791715e-05,
+      "loss": 3.413,
+      "step": 1897472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8927247319281195e-05,
+      "loss": 3.4309,
+      "step": 1897984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8918861371770675e-05,
+      "loss": 3.4177,
+      "step": 1898496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8910491803063884e-05,
+      "loss": 3.4236,
+      "step": 1899008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8902105855553364e-05,
+      "loss": 3.4099,
+      "step": 1899520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8893719908042844e-05,
+      "loss": 3.4246,
+      "step": 1900032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8885350339336056e-05,
+      "loss": 3.4148,
+      "step": 1900544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8876964391825536e-05,
+      "loss": 3.4194,
+      "step": 1901056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8868578444315016e-05,
+      "loss": 3.4129,
+      "step": 1901568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8860192496804496e-05,
+      "loss": 3.4236,
+      "step": 1902080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8851822928097705e-05,
+      "loss": 3.4175,
+      "step": 1902592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.884343698058719e-05,
+      "loss": 3.4392,
+      "step": 1903104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.883505103307667e-05,
+      "loss": 3.4133,
+      "step": 1903616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.882666508556615e-05,
+      "loss": 3.4247,
+      "step": 1904128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8818295516859357e-05,
+      "loss": 3.4215,
+      "step": 1904640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8809909569348837e-05,
+      "loss": 3.4198,
+      "step": 1905152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8801523621838317e-05,
+      "loss": 3.4162,
+      "step": 1905664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8793137674327797e-05,
+      "loss": 3.4267,
+      "step": 1906176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.878476810562101e-05,
+      "loss": 3.4222,
+      "step": 1906688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.877638215811049e-05,
+      "loss": 3.4109,
+      "step": 1907200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.876799621059997e-05,
+      "loss": 3.4203,
+      "step": 1907712
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.86281681060791,
+      "eval_runtime": 303.7952,
+      "eval_samples_per_second": 1256.08,
+      "eval_steps_per_second": 39.253,
+      "step": 1908000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.875961026308945e-05,
+      "loss": 3.4165,
+      "step": 1908224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.875122431557893e-05,
+      "loss": 3.4071,
+      "step": 1908736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.874283836806841e-05,
+      "loss": 3.4196,
+      "step": 1909248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.873445242055789e-05,
+      "loss": 3.4327,
+      "step": 1909760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8726082851851102e-05,
+      "loss": 3.4264,
+      "step": 1910272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8717696904340582e-05,
+      "loss": 3.4113,
+      "step": 1910784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.870931095683006e-05,
+      "loss": 3.4259,
+      "step": 1911296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.870092500931954e-05,
+      "loss": 3.4044,
+      "step": 1911808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.869255544061275e-05,
+      "loss": 3.417,
+      "step": 1912320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8684169493102234e-05,
+      "loss": 3.4267,
+      "step": 1912832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.867578354559171e-05,
+      "loss": 3.4134,
+      "step": 1913344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.866739759808119e-05,
+      "loss": 3.416,
+      "step": 1913856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8659044408178132e-05,
+      "loss": 3.4181,
+      "step": 1914368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8650658460667612e-05,
+      "loss": 3.4094,
+      "step": 1914880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8642272513157096e-05,
+      "loss": 3.4089,
+      "step": 1915392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8633886565646576e-05,
+      "loss": 3.4057,
+      "step": 1915904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8625500618136056e-05,
+      "loss": 3.4092,
+      "step": 1916416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8617114670625532e-05,
+      "loss": 3.4134,
+      "step": 1916928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8608728723115012e-05,
+      "loss": 3.4106,
+      "step": 1917440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8600342775604492e-05,
+      "loss": 3.4258,
+      "step": 1917952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8591973206897705e-05,
+      "loss": 3.427,
+      "step": 1918464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8583587259387185e-05,
+      "loss": 3.4227,
+      "step": 1918976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8575201311876665e-05,
+      "loss": 3.4183,
+      "step": 1919488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8566815364366144e-05,
+      "loss": 3.4125,
+      "step": 1920000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8558445795659357e-05,
+      "loss": 3.4206,
+      "step": 1920512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8550059848148834e-05,
+      "loss": 3.4156,
+      "step": 1921024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8541673900638313e-05,
+      "loss": 3.412,
+      "step": 1921536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8533287953127797e-05,
+      "loss": 3.4128,
+      "step": 1922048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8524918384421006e-05,
+      "loss": 3.4115,
+      "step": 1922560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8516532436910486e-05,
+      "loss": 3.4116,
+      "step": 1923072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8508146489399966e-05,
+      "loss": 3.4102,
+      "step": 1923584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8499760541889446e-05,
+      "loss": 3.419,
+      "step": 1924096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8491390973182658e-05,
+      "loss": 3.4253,
+      "step": 1924608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8483005025672138e-05,
+      "loss": 3.4149,
+      "step": 1925120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8474619078161618e-05,
+      "loss": 3.4119,
+      "step": 1925632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8466233130651098e-05,
+      "loss": 3.4109,
+      "step": 1926144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8457863561944307e-05,
+      "loss": 3.4198,
+      "step": 1926656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8449477614433787e-05,
+      "loss": 3.4105,
+      "step": 1927168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8441091666923267e-05,
+      "loss": 3.4122,
+      "step": 1927680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.843270571941275e-05,
+      "loss": 3.4032,
+      "step": 1928192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.842433615070596e-05,
+      "loss": 3.4166,
+      "step": 1928704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.841595020319544e-05,
+      "loss": 3.4168,
+      "step": 1929216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.840756425568492e-05,
+      "loss": 3.4152,
+      "step": 1929728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.83991783081744e-05,
+      "loss": 3.417,
+      "step": 1930240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8390808739467612e-05,
+      "loss": 3.42,
+      "step": 1930752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8382422791957092e-05,
+      "loss": 3.4187,
+      "step": 1931264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8374036844446572e-05,
+      "loss": 3.418,
+      "step": 1931776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8365650896936052e-05,
+      "loss": 3.4123,
+      "step": 1932288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.835728132822926e-05,
+      "loss": 3.3843,
+      "step": 1932800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.834889538071874e-05,
+      "loss": 3.4252,
+      "step": 1933312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.834050943320822e-05,
+      "loss": 3.4153,
+      "step": 1933824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8332123485697704e-05,
+      "loss": 3.4188,
+      "step": 1934336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8323753916990913e-05,
+      "loss": 3.4139,
+      "step": 1934848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8315367969480393e-05,
+      "loss": 3.4055,
+      "step": 1935360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8306982021969873e-05,
+      "loss": 3.3958,
+      "step": 1935872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8298612453263082e-05,
+      "loss": 3.4047,
+      "step": 1936384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8290226505752566e-05,
+      "loss": 3.407,
+      "step": 1936896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8281840558242046e-05,
+      "loss": 3.4033,
+      "step": 1937408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8273454610731526e-05,
+      "loss": 3.4241,
+      "step": 1937920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8265085042024735e-05,
+      "loss": 3.4069,
+      "step": 1938432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8256699094514215e-05,
+      "loss": 3.3926,
+      "step": 1938944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8248313147003695e-05,
+      "loss": 3.4237,
+      "step": 1939456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8239927199493174e-05,
+      "loss": 3.3964,
+      "step": 1939968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8231557630786387e-05,
+      "loss": 3.3986,
+      "step": 1940480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8223171683275867e-05,
+      "loss": 3.4151,
+      "step": 1940992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8214785735765347e-05,
+      "loss": 3.4103,
+      "step": 1941504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8206399788254827e-05,
+      "loss": 3.4011,
+      "step": 1942016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8198030219548036e-05,
+      "loss": 3.4016,
+      "step": 1942528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.818964427203752e-05,
+      "loss": 3.4021,
+      "step": 1943040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181258324527e-05,
+      "loss": 3.413,
+      "step": 1943552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.817287237701648e-05,
+      "loss": 3.4087,
+      "step": 1944064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8164502808309688e-05,
+      "loss": 3.4087,
+      "step": 1944576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8156116860799168e-05,
+      "loss": 3.4092,
+      "step": 1945088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8147730913288648e-05,
+      "loss": 3.4117,
+      "step": 1945600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8139344965778128e-05,
+      "loss": 3.425,
+      "step": 1946112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.813097539707134e-05,
+      "loss": 3.4,
+      "step": 1946624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.812258944956082e-05,
+      "loss": 3.4175,
+      "step": 1947136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.81142035020503e-05,
+      "loss": 3.4173,
+      "step": 1947648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.810581755453978e-05,
+      "loss": 3.3931,
+      "step": 1948160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.809744798583299e-05,
+      "loss": 3.4118,
+      "step": 1948672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8089062038322473e-05,
+      "loss": 3.4152,
+      "step": 1949184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8080676090811953e-05,
+      "loss": 3.415,
+      "step": 1949696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.807229014330143e-05,
+      "loss": 3.4152,
+      "step": 1950208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8063920574594642e-05,
+      "loss": 3.3939,
+      "step": 1950720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8055534627084122e-05,
+      "loss": 3.3908,
+      "step": 1951232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8047148679573602e-05,
+      "loss": 3.4085,
+      "step": 1951744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8038762732063082e-05,
+      "loss": 3.4162,
+      "step": 1952256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8030393163356294e-05,
+      "loss": 3.4013,
+      "step": 1952768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8022007215845774e-05,
+      "loss": 3.4066,
+      "step": 1953280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8013621268335254e-05,
+      "loss": 3.4028,
+      "step": 1953792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.800523532082473e-05,
+      "loss": 3.414,
+      "step": 1954304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7996865752117943e-05,
+      "loss": 3.4054,
+      "step": 1954816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7988479804607423e-05,
+      "loss": 3.4053,
+      "step": 1955328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7980093857096903e-05,
+      "loss": 3.4095,
+      "step": 1955840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7971707909586383e-05,
+      "loss": 3.4175,
+      "step": 1956352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7963338340879596e-05,
+      "loss": 3.4138,
+      "step": 1956864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7954952393369076e-05,
+      "loss": 3.3981,
+      "step": 1957376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7946566445858552e-05,
+      "loss": 3.4047,
+      "step": 1957888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7938180498348035e-05,
+      "loss": 3.4057,
+      "step": 1958400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7929810929641248e-05,
+      "loss": 3.4168,
+      "step": 1958912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7921424982130728e-05,
+      "loss": 3.4122,
+      "step": 1959424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7913039034620205e-05,
+      "loss": 3.4105,
+      "step": 1959936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7904653087109684e-05,
+      "loss": 3.4037,
+      "step": 1960448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7896283518402897e-05,
+      "loss": 3.3986,
+      "step": 1960960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7887897570892377e-05,
+      "loss": 3.4068,
+      "step": 1961472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7879511623381857e-05,
+      "loss": 3.4033,
+      "step": 1961984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7871125675871337e-05,
+      "loss": 3.3999,
+      "step": 1962496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.786275610716455e-05,
+      "loss": 3.4164,
+      "step": 1963008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7854370159654026e-05,
+      "loss": 3.4034,
+      "step": 1963520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7845984212143506e-05,
+      "loss": 3.4025,
+      "step": 1964032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.783759826463299e-05,
+      "loss": 3.4019,
+      "step": 1964544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.78292286959262e-05,
+      "loss": 3.3987,
+      "step": 1965056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7820842748415678e-05,
+      "loss": 3.4159,
+      "step": 1965568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7812456800905158e-05,
+      "loss": 3.4047,
+      "step": 1966080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7804070853394638e-05,
+      "loss": 3.4192,
+      "step": 1966592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.779570128468785e-05,
+      "loss": 3.4193,
+      "step": 1967104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.778731533717733e-05,
+      "loss": 3.4112,
+      "step": 1967616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.777892938966681e-05,
+      "loss": 3.4059,
+      "step": 1968128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.777054344215629e-05,
+      "loss": 3.4117,
+      "step": 1968640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.77621738734495e-05,
+      "loss": 3.4023,
+      "step": 1969152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.775378792593898e-05,
+      "loss": 3.4121,
+      "step": 1969664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.774540197842846e-05,
+      "loss": 3.408,
+      "step": 1970176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7737016030917943e-05,
+      "loss": 3.4046,
+      "step": 1970688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7728646462211152e-05,
+      "loss": 3.4028,
+      "step": 1971200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7720260514700632e-05,
+      "loss": 3.4058,
+      "step": 1971712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7711874567190112e-05,
+      "loss": 3.4077,
+      "step": 1972224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7703488619679592e-05,
+      "loss": 3.4013,
+      "step": 1972736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7695119050972804e-05,
+      "loss": 3.4102,
+      "step": 1973248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7686733103462284e-05,
+      "loss": 3.397,
+      "step": 1973760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7678347155951764e-05,
+      "loss": 3.4261,
+      "step": 1974272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7669961208441244e-05,
+      "loss": 3.4071,
+      "step": 1974784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7661591639734453e-05,
+      "loss": 3.4141,
+      "step": 1975296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7653205692223933e-05,
+      "loss": 3.4019,
+      "step": 1975808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7644819744713413e-05,
+      "loss": 3.4136,
+      "step": 1976320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7636433797202893e-05,
+      "loss": 3.4036,
+      "step": 1976832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7628064228496106e-05,
+      "loss": 3.41,
+      "step": 1977344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7619678280985586e-05,
+      "loss": 3.405,
+      "step": 1977856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7611292333475066e-05,
+      "loss": 3.4062,
+      "step": 1978368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7602922764768275e-05,
+      "loss": 3.4091,
+      "step": 1978880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7594536817257758e-05,
+      "loss": 3.4293,
+      "step": 1979392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7586150869747238e-05,
+      "loss": 3.402,
+      "step": 1979904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7577764922236718e-05,
+      "loss": 3.4156,
+      "step": 1980416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7569395353529927e-05,
+      "loss": 3.4136,
+      "step": 1980928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7561009406019407e-05,
+      "loss": 3.4073,
+      "step": 1981440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7552623458508887e-05,
+      "loss": 3.405,
+      "step": 1981952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7544237510998367e-05,
+      "loss": 3.4132,
+      "step": 1982464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.753586794229158e-05,
+      "loss": 3.4108,
+      "step": 1982976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.752748199478106e-05,
+      "loss": 3.4031,
+      "step": 1983488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.751909604727054e-05,
+      "loss": 3.4095,
+      "step": 1984000
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.864938974380493,
+      "eval_runtime": 302.0028,
+      "eval_samples_per_second": 1263.534,
+      "eval_steps_per_second": 39.486,
+      "step": 1984320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.751071009976002e-05,
+      "loss": 3.4047,
+      "step": 1984512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7502340531053228e-05,
+      "loss": 3.3949,
+      "step": 1985024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7493954583542708e-05,
+      "loss": 3.4089,
+      "step": 1985536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.748556863603219e-05,
+      "loss": 3.4197,
+      "step": 1986048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.747718268852167e-05,
+      "loss": 3.4199,
+      "step": 1986560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.746881311981488e-05,
+      "loss": 3.4023,
+      "step": 1987072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.746042717230436e-05,
+      "loss": 3.4107,
+      "step": 1987584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.745204122479384e-05,
+      "loss": 3.3996,
+      "step": 1988096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.744365527728332e-05,
+      "loss": 3.4023,
+      "step": 1988608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7435285708576533e-05,
+      "loss": 3.4143,
+      "step": 1989120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7426899761066013e-05,
+      "loss": 3.4064,
+      "step": 1989632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7418513813555493e-05,
+      "loss": 3.4089,
+      "step": 1990144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7410144244848702e-05,
+      "loss": 3.408,
+      "step": 1990656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7401758297338182e-05,
+      "loss": 3.3967,
+      "step": 1991168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7393372349827662e-05,
+      "loss": 3.3973,
+      "step": 1991680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7384986402317145e-05,
+      "loss": 3.3917,
+      "step": 1992192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7376600454806625e-05,
+      "loss": 3.4024,
+      "step": 1992704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7368214507296102e-05,
+      "loss": 3.4065,
+      "step": 1993216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7359828559785582e-05,
+      "loss": 3.3981,
+      "step": 1993728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.735144261227506e-05,
+      "loss": 3.4144,
+      "step": 1994240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7343073043568274e-05,
+      "loss": 3.4159,
+      "step": 1994752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7334687096057754e-05,
+      "loss": 3.41,
+      "step": 1995264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7326301148547234e-05,
+      "loss": 3.4092,
+      "step": 1995776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7317915201036714e-05,
+      "loss": 3.4048,
+      "step": 1996288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7309545632329923e-05,
+      "loss": 3.411,
+      "step": 1996800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7301159684819403e-05,
+      "loss": 3.406,
+      "step": 1997312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7292773737308883e-05,
+      "loss": 3.4002,
+      "step": 1997824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7284387789798366e-05,
+      "loss": 3.3995,
+      "step": 1998336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7276018221091575e-05,
+      "loss": 3.4036,
+      "step": 1998848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7267632273581055e-05,
+      "loss": 3.4024,
+      "step": 1999360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7259246326070535e-05,
+      "loss": 3.4017,
+      "step": 1999872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7250860378560015e-05,
+      "loss": 3.4077,
+      "step": 2000384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7242490809853228e-05,
+      "loss": 3.4115,
+      "step": 2000896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7234104862342708e-05,
+      "loss": 3.4105,
+      "step": 2001408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7225718914832188e-05,
+      "loss": 3.3975,
+      "step": 2001920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7217332967321668e-05,
+      "loss": 3.4045,
+      "step": 2002432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7208963398614877e-05,
+      "loss": 3.4118,
+      "step": 2002944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7200577451104357e-05,
+      "loss": 3.3933,
+      "step": 2003456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7192191503593837e-05,
+      "loss": 3.4053,
+      "step": 2003968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.718382193488705e-05,
+      "loss": 3.3921,
+      "step": 2004480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.717543598737653e-05,
+      "loss": 3.4063,
+      "step": 2004992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.716705003986601e-05,
+      "loss": 3.4029,
+      "step": 2005504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.715866409235549e-05,
+      "loss": 3.4052,
+      "step": 2006016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.715027814484497e-05,
+      "loss": 3.4085,
+      "step": 2006528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.714189219733445e-05,
+      "loss": 3.412,
+      "step": 2007040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.713350624982393e-05,
+      "loss": 3.4089,
+      "step": 2007552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.712512030231341e-05,
+      "loss": 3.4113,
+      "step": 2008064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.711675073360662e-05,
+      "loss": 3.3986,
+      "step": 2008576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.71083647860961e-05,
+      "loss": 3.3768,
+      "step": 2009088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7099978838585578e-05,
+      "loss": 3.416,
+      "step": 2009600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7091592891075058e-05,
+      "loss": 3.4041,
+      "step": 2010112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.708322332236827e-05,
+      "loss": 3.4083,
+      "step": 2010624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7074837374857754e-05,
+      "loss": 3.4016,
+      "step": 2011136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.706645142734723e-05,
+      "loss": 3.3957,
+      "step": 2011648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.705806547983671e-05,
+      "loss": 3.385,
+      "step": 2012160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7049695911129923e-05,
+      "loss": 3.3949,
+      "step": 2012672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.70413099636194e-05,
+      "loss": 3.3996,
+      "step": 2013184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7032924016108883e-05,
+      "loss": 3.3883,
+      "step": 2013696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7024538068598363e-05,
+      "loss": 3.4148,
+      "step": 2014208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7016168499891575e-05,
+      "loss": 3.3994,
+      "step": 2014720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.700778255238105e-05,
+      "loss": 3.3847,
+      "step": 2015232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.699939660487053e-05,
+      "loss": 3.4094,
+      "step": 2015744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.699101065736001e-05,
+      "loss": 3.3869,
+      "step": 2016256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6982641088653224e-05,
+      "loss": 3.3901,
+      "step": 2016768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6974255141142704e-05,
+      "loss": 3.4033,
+      "step": 2017280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6965869193632184e-05,
+      "loss": 3.401,
+      "step": 2017792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6957499624925396e-05,
+      "loss": 3.3908,
+      "step": 2018304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6949113677414873e-05,
+      "loss": 3.3914,
+      "step": 2018816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6940727729904353e-05,
+      "loss": 3.3936,
+      "step": 2019328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6932341782393836e-05,
+      "loss": 3.3985,
+      "step": 2019840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.692397221368705e-05,
+      "loss": 3.4049,
+      "step": 2020352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6915586266176525e-05,
+      "loss": 3.3952,
+      "step": 2020864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6907200318666005e-05,
+      "loss": 3.3997,
+      "step": 2021376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6898814371155485e-05,
+      "loss": 3.3958,
+      "step": 2021888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6890444802448698e-05,
+      "loss": 3.4161,
+      "step": 2022400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6882058854938178e-05,
+      "loss": 3.3934,
+      "step": 2022912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6873672907427658e-05,
+      "loss": 3.4084,
+      "step": 2023424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6865286959917138e-05,
+      "loss": 3.4009,
+      "step": 2023936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6856917391210347e-05,
+      "loss": 3.3877,
+      "step": 2024448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6848531443699827e-05,
+      "loss": 3.4011,
+      "step": 2024960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6840145496189307e-05,
+      "loss": 3.4058,
+      "step": 2025472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.683175954867879e-05,
+      "loss": 3.4024,
+      "step": 2025984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6823389979972e-05,
+      "loss": 3.4062,
+      "step": 2026496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.681500403246148e-05,
+      "loss": 3.3783,
+      "step": 2027008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.680661808495096e-05,
+      "loss": 3.3825,
+      "step": 2027520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.679823213744044e-05,
+      "loss": 3.3954,
+      "step": 2028032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.678986256873365e-05,
+      "loss": 3.406,
+      "step": 2028544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.678147662122313e-05,
+      "loss": 3.3908,
+      "step": 2029056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.677309067371261e-05,
+      "loss": 3.3978,
+      "step": 2029568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.676470472620209e-05,
+      "loss": 3.3931,
+      "step": 2030080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.67563351574953e-05,
+      "loss": 3.4037,
+      "step": 2030592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.674794920998478e-05,
+      "loss": 3.3963,
+      "step": 2031104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.673956326247426e-05,
+      "loss": 3.3939,
+      "step": 2031616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.673117731496374e-05,
+      "loss": 3.4021,
+      "step": 2032128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6722807746256953e-05,
+      "loss": 3.4059,
+      "step": 2032640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6714421798746433e-05,
+      "loss": 3.4064,
+      "step": 2033152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6706035851235913e-05,
+      "loss": 3.3856,
+      "step": 2033664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6697649903725393e-05,
+      "loss": 3.3927,
+      "step": 2034176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.66892803350186e-05,
+      "loss": 3.3963,
+      "step": 2034688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6680894387508085e-05,
+      "loss": 3.4078,
+      "step": 2035200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6672508439997565e-05,
+      "loss": 3.4001,
+      "step": 2035712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6664122492487045e-05,
+      "loss": 3.4018,
+      "step": 2036224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6655752923780254e-05,
+      "loss": 3.3931,
+      "step": 2036736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6647366976269734e-05,
+      "loss": 3.3886,
+      "step": 2037248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6638981028759214e-05,
+      "loss": 3.3954,
+      "step": 2037760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6630595081248694e-05,
+      "loss": 3.3937,
+      "step": 2038272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6622225512541906e-05,
+      "loss": 3.3921,
+      "step": 2038784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6613839565031386e-05,
+      "loss": 3.4051,
+      "step": 2039296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6605453617520866e-05,
+      "loss": 3.3923,
+      "step": 2039808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6597067670010346e-05,
+      "loss": 3.3924,
+      "step": 2040320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6588698101303555e-05,
+      "loss": 3.3904,
+      "step": 2040832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.658031215379304e-05,
+      "loss": 3.3889,
+      "step": 2041344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.657192620628252e-05,
+      "loss": 3.4045,
+      "step": 2041856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6563540258772e-05,
+      "loss": 3.3998,
+      "step": 2042368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6555170690065208e-05,
+      "loss": 3.4036,
+      "step": 2042880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6546784742554688e-05,
+      "loss": 3.4076,
+      "step": 2043392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6538398795044168e-05,
+      "loss": 3.4028,
+      "step": 2043904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6530012847533648e-05,
+      "loss": 3.3961,
+      "step": 2044416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.652164327882686e-05,
+      "loss": 3.3999,
+      "step": 2044928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.651325733131634e-05,
+      "loss": 3.3984,
+      "step": 2045440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.650487138380582e-05,
+      "loss": 3.4005,
+      "step": 2045952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6496485436295296e-05,
+      "loss": 3.3948,
+      "step": 2046464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.648811586758851e-05,
+      "loss": 3.3991,
+      "step": 2046976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6479729920077992e-05,
+      "loss": 3.389,
+      "step": 2047488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6471343972567472e-05,
+      "loss": 3.3984,
+      "step": 2048000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.646295802505695e-05,
+      "loss": 3.3959,
+      "step": 2048512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.645458845635016e-05,
+      "loss": 3.3941,
+      "step": 2049024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.644620250883964e-05,
+      "loss": 3.3953,
+      "step": 2049536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.643781656132912e-05,
+      "loss": 3.3861,
+      "step": 2050048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.64294306138186e-05,
+      "loss": 3.4198,
+      "step": 2050560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6421061045111814e-05,
+      "loss": 3.3964,
+      "step": 2051072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6412675097601294e-05,
+      "loss": 3.4084,
+      "step": 2051584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.640428915009077e-05,
+      "loss": 3.3908,
+      "step": 2052096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.639590320258025e-05,
+      "loss": 3.4021,
+      "step": 2052608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6387533633873463e-05,
+      "loss": 3.396,
+      "step": 2053120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6379147686362946e-05,
+      "loss": 3.4001,
+      "step": 2053632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6370761738852423e-05,
+      "loss": 3.3926,
+      "step": 2054144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6362375791341903e-05,
+      "loss": 3.3945,
+      "step": 2054656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6354006222635115e-05,
+      "loss": 3.4013,
+      "step": 2055168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6345620275124595e-05,
+      "loss": 3.422,
+      "step": 2055680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6337234327614075e-05,
+      "loss": 3.3885,
+      "step": 2056192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6328864758907287e-05,
+      "loss": 3.4071,
+      "step": 2056704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6320478811396767e-05,
+      "loss": 3.4056,
+      "step": 2057216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6312092863886244e-05,
+      "loss": 3.3937,
+      "step": 2057728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6303706916375724e-05,
+      "loss": 3.3956,
+      "step": 2058240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6295337347668936e-05,
+      "loss": 3.402,
+      "step": 2058752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6286951400158416e-05,
+      "loss": 3.4011,
+      "step": 2059264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6278565452647896e-05,
+      "loss": 3.3963,
+      "step": 2059776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6270179505137376e-05,
+      "loss": 3.3955,
+      "step": 2060288
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.865971803665161,
+      "eval_runtime": 301.9586,
+      "eval_samples_per_second": 1263.72,
+      "eval_steps_per_second": 39.492,
+      "step": 2060640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6261793557626856e-05,
+      "loss": 3.3913,
+      "step": 2060800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6253407610116336e-05,
+      "loss": 3.387,
+      "step": 2061312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6245021662605816e-05,
+      "loss": 3.3972,
+      "step": 2061824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6236635715095296e-05,
+      "loss": 3.4074,
+      "step": 2062336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.622826614638851e-05,
+      "loss": 3.4103,
+      "step": 2062848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.621988019887799e-05,
+      "loss": 3.3935,
+      "step": 2063360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.621149425136747e-05,
+      "loss": 3.4022,
+      "step": 2063872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.620310830385695e-05,
+      "loss": 3.3914,
+      "step": 2064384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6194738735150157e-05,
+      "loss": 3.3902,
+      "step": 2064896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6186352787639637e-05,
+      "loss": 3.4002,
+      "step": 2065408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6177966840129117e-05,
+      "loss": 3.4009,
+      "step": 2065920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.61695808926186e-05,
+      "loss": 3.4006,
+      "step": 2066432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.616122770271554e-05,
+      "loss": 3.3937,
+      "step": 2066944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.615284175520502e-05,
+      "loss": 3.3897,
+      "step": 2067456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.61444558076945e-05,
+      "loss": 3.3881,
+      "step": 2067968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.613606986018398e-05,
+      "loss": 3.3786,
+      "step": 2068480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6127683912673462e-05,
+      "loss": 3.3961,
+      "step": 2068992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6119297965162942e-05,
+      "loss": 3.3958,
+      "step": 2069504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6110912017652422e-05,
+      "loss": 3.3897,
+      "step": 2070016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.61025260701419e-05,
+      "loss": 3.4016,
+      "step": 2070528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.609415650143511e-05,
+      "loss": 3.4066,
+      "step": 2071040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.608577055392459e-05,
+      "loss": 3.398,
+      "step": 2071552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.607738460641407e-05,
+      "loss": 3.3999,
+      "step": 2072064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.606899865890355e-05,
+      "loss": 3.3929,
+      "step": 2072576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6060629090196764e-05,
+      "loss": 3.4017,
+      "step": 2073088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6052243142686243e-05,
+      "loss": 3.3982,
+      "step": 2073600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.604385719517572e-05,
+      "loss": 3.3891,
+      "step": 2074112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.60354712476652e-05,
+      "loss": 3.3895,
+      "step": 2074624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6027101678958416e-05,
+      "loss": 3.3913,
+      "step": 2075136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6018715731447896e-05,
+      "loss": 3.3914,
+      "step": 2075648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6010329783937372e-05,
+      "loss": 3.3956,
+      "step": 2076160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6001943836426852e-05,
+      "loss": 3.3964,
+      "step": 2076672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5993574267720065e-05,
+      "loss": 3.3981,
+      "step": 2077184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5985188320209545e-05,
+      "loss": 3.405,
+      "step": 2077696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5976802372699025e-05,
+      "loss": 3.3885,
+      "step": 2078208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5968416425188505e-05,
+      "loss": 3.3947,
+      "step": 2078720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5960046856481717e-05,
+      "loss": 3.4011,
+      "step": 2079232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5951660908971194e-05,
+      "loss": 3.3821,
+      "step": 2079744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5943274961460674e-05,
+      "loss": 3.3973,
+      "step": 2080256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5934905392753886e-05,
+      "loss": 3.385,
+      "step": 2080768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.592651944524337e-05,
+      "loss": 3.3888,
+      "step": 2081280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5918133497732846e-05,
+      "loss": 3.3916,
+      "step": 2081792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5909747550222326e-05,
+      "loss": 3.3957,
+      "step": 2082304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5901361602711806e-05,
+      "loss": 3.4009,
+      "step": 2082816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5892975655201286e-05,
+      "loss": 3.3989,
+      "step": 2083328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5884589707690766e-05,
+      "loss": 3.4004,
+      "step": 2083840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5876203760180246e-05,
+      "loss": 3.3993,
+      "step": 2084352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.586783419147346e-05,
+      "loss": 3.3915,
+      "step": 2084864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.585944824396294e-05,
+      "loss": 3.3687,
+      "step": 2085376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5851062296452418e-05,
+      "loss": 3.405,
+      "step": 2085888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5842676348941898e-05,
+      "loss": 3.3929,
+      "step": 2086400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5834306780235107e-05,
+      "loss": 3.3983,
+      "step": 2086912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5825920832724587e-05,
+      "loss": 3.3914,
+      "step": 2087424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.581753488521407e-05,
+      "loss": 3.3816,
+      "step": 2087936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.580914893770355e-05,
+      "loss": 3.3846,
+      "step": 2088448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.580077936899676e-05,
+      "loss": 3.3786,
+      "step": 2088960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.579239342148624e-05,
+      "loss": 3.3922,
+      "step": 2089472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.578400747397572e-05,
+      "loss": 3.375,
+      "step": 2089984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.57756215264652e-05,
+      "loss": 3.4054,
+      "step": 2090496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5767251957758412e-05,
+      "loss": 3.3914,
+      "step": 2091008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5758866010247892e-05,
+      "loss": 3.373,
+      "step": 2091520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5750480062737372e-05,
+      "loss": 3.3991,
+      "step": 2092032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.574209411522685e-05,
+      "loss": 3.3773,
+      "step": 2092544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.573372454652006e-05,
+      "loss": 3.3818,
+      "step": 2093056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.572533859900954e-05,
+      "loss": 3.3889,
+      "step": 2093568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5716952651499024e-05,
+      "loss": 3.3907,
+      "step": 2094080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5708583082792233e-05,
+      "loss": 3.3811,
+      "step": 2094592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5700197135281713e-05,
+      "loss": 3.3855,
+      "step": 2095104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5691811187771193e-05,
+      "loss": 3.3782,
+      "step": 2095616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5683425240260673e-05,
+      "loss": 3.3891,
+      "step": 2096128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5675055671553886e-05,
+      "loss": 3.3957,
+      "step": 2096640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5666669724043366e-05,
+      "loss": 3.3835,
+      "step": 2097152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5658283776532846e-05,
+      "loss": 3.393,
+      "step": 2097664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5649897829022322e-05,
+      "loss": 3.3815,
+      "step": 2098176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5641528260315535e-05,
+      "loss": 3.4103,
+      "step": 2098688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5633142312805015e-05,
+      "loss": 3.3789,
+      "step": 2099200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5624756365294495e-05,
+      "loss": 3.3982,
+      "step": 2099712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5616370417783975e-05,
+      "loss": 3.3862,
+      "step": 2100224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5608000849077187e-05,
+      "loss": 3.3821,
+      "step": 2100736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5599614901566667e-05,
+      "loss": 3.3896,
+      "step": 2101248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5591228954056147e-05,
+      "loss": 3.395,
+      "step": 2101760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5582843006545624e-05,
+      "loss": 3.3961,
+      "step": 2102272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.557447343783884e-05,
+      "loss": 3.3949,
+      "step": 2102784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.556608749032832e-05,
+      "loss": 3.3699,
+      "step": 2103296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5557701542817796e-05,
+      "loss": 3.3733,
+      "step": 2103808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5549315595307276e-05,
+      "loss": 3.3846,
+      "step": 2104320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.554094602660049e-05,
+      "loss": 3.3965,
+      "step": 2104832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.553256007908997e-05,
+      "loss": 3.3818,
+      "step": 2105344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5524174131579448e-05,
+      "loss": 3.3835,
+      "step": 2105856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5515788184068928e-05,
+      "loss": 3.3843,
+      "step": 2106368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.550741861536214e-05,
+      "loss": 3.3881,
+      "step": 2106880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5499032667851617e-05,
+      "loss": 3.387,
+      "step": 2107392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5490646720341097e-05,
+      "loss": 3.3844,
+      "step": 2107904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5482260772830577e-05,
+      "loss": 3.3912,
+      "step": 2108416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5473891204123793e-05,
+      "loss": 3.3979,
+      "step": 2108928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.546550525661327e-05,
+      "loss": 3.3982,
+      "step": 2109440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.545711930910275e-05,
+      "loss": 3.3769,
+      "step": 2109952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.544873336159223e-05,
+      "loss": 3.3793,
+      "step": 2110464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5440363792885442e-05,
+      "loss": 3.3863,
+      "step": 2110976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5431977845374922e-05,
+      "loss": 3.3922,
+      "step": 2111488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5423591897864402e-05,
+      "loss": 3.3931,
+      "step": 2112000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5415205950353882e-05,
+      "loss": 3.3904,
+      "step": 2112512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.540683638164709e-05,
+      "loss": 3.3843,
+      "step": 2113024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.539845043413657e-05,
+      "loss": 3.3802,
+      "step": 2113536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.539006448662605e-05,
+      "loss": 3.3861,
+      "step": 2114048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.538167853911553e-05,
+      "loss": 3.3849,
+      "step": 2114560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5373308970408743e-05,
+      "loss": 3.3823,
+      "step": 2115072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5364923022898223e-05,
+      "loss": 3.392,
+      "step": 2115584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5356537075387703e-05,
+      "loss": 3.386,
+      "step": 2116096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5348151127877183e-05,
+      "loss": 3.38,
+      "step": 2116608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5339781559170392e-05,
+      "loss": 3.3825,
+      "step": 2117120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5331395611659872e-05,
+      "loss": 3.3773,
+      "step": 2117632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5323009664149356e-05,
+      "loss": 3.3943,
+      "step": 2118144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5314623716638836e-05,
+      "loss": 3.3894,
+      "step": 2118656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5306254147932045e-05,
+      "loss": 3.3912,
+      "step": 2119168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5297868200421525e-05,
+      "loss": 3.393,
+      "step": 2119680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5289482252911005e-05,
+      "loss": 3.3974,
+      "step": 2120192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5281096305400485e-05,
+      "loss": 3.3885,
+      "step": 2120704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5272726736693697e-05,
+      "loss": 3.3834,
+      "step": 2121216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5264340789183177e-05,
+      "loss": 3.3931,
+      "step": 2121728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5255954841672657e-05,
+      "loss": 3.3855,
+      "step": 2122240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5247568894162137e-05,
+      "loss": 3.3838,
+      "step": 2122752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5239199325455348e-05,
+      "loss": 3.3906,
+      "step": 2123264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5230813377944828e-05,
+      "loss": 3.3811,
+      "step": 2123776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5222427430434308e-05,
+      "loss": 3.3843,
+      "step": 2124288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5214041482923788e-05,
+      "loss": 3.3842,
+      "step": 2124800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5205671914216998e-05,
+      "loss": 3.3855,
+      "step": 2125312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5197285966706478e-05,
+      "loss": 3.3848,
+      "step": 2125824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5188900019195958e-05,
+      "loss": 3.3803,
+      "step": 2126336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5180530450489169e-05,
+      "loss": 3.4059,
+      "step": 2126848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5172144502978649e-05,
+      "loss": 3.388,
+      "step": 2127360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.516375855546813e-05,
+      "loss": 3.3962,
+      "step": 2127872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.515537260795761e-05,
+      "loss": 3.3796,
+      "step": 2128384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.514700303925082e-05,
+      "loss": 3.3905,
+      "step": 2128896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5138617091740301e-05,
+      "loss": 3.3869,
+      "step": 2129408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5130231144229781e-05,
+      "loss": 3.3892,
+      "step": 2129920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5121845196719261e-05,
+      "loss": 3.3826,
+      "step": 2130432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5113475628012472e-05,
+      "loss": 3.383,
+      "step": 2130944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5105089680501952e-05,
+      "loss": 3.3874,
+      "step": 2131456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5096703732991432e-05,
+      "loss": 3.4111,
+      "step": 2131968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5088317785480912e-05,
+      "loss": 3.3827,
+      "step": 2132480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5079948216774123e-05,
+      "loss": 3.3966,
+      "step": 2132992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5071562269263603e-05,
+      "loss": 3.3964,
+      "step": 2133504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5063176321753084e-05,
+      "loss": 3.3804,
+      "step": 2134016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5054790374242564e-05,
+      "loss": 3.3933,
+      "step": 2134528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5046420805535773e-05,
+      "loss": 3.3899,
+      "step": 2135040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5038034858025255e-05,
+      "loss": 3.3906,
+      "step": 2135552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5029648910514735e-05,
+      "loss": 3.3874,
+      "step": 2136064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5021262963004215e-05,
+      "loss": 3.385,
+      "step": 2136576
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.867809534072876,
+      "eval_runtime": 304.8577,
+      "eval_samples_per_second": 1251.702,
+      "eval_steps_per_second": 39.117,
+      "step": 2136960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5012893394297426e-05,
+      "loss": 3.3871,
+      "step": 2137088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5004507446786906e-05,
+      "loss": 3.3719,
+      "step": 2137600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4996121499276386e-05,
+      "loss": 3.3854,
+      "step": 2138112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4987735551765866e-05,
+      "loss": 3.4,
+      "step": 2138624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4979349604255344e-05,
+      "loss": 3.396,
+      "step": 2139136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4970963656744824e-05,
+      "loss": 3.3859,
+      "step": 2139648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4962577709234304e-05,
+      "loss": 3.397,
+      "step": 2140160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4954191761723785e-05,
+      "loss": 3.3795,
+      "step": 2140672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4945822193016994e-05,
+      "loss": 3.379,
+      "step": 2141184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4937436245506476e-05,
+      "loss": 3.3927,
+      "step": 2141696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4929050297995956e-05,
+      "loss": 3.3899,
+      "step": 2142208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4920664350485436e-05,
+      "loss": 3.3875,
+      "step": 2142720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.491231116058238e-05,
+      "loss": 3.3896,
+      "step": 2143232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.490392521307186e-05,
+      "loss": 3.3779,
+      "step": 2143744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.489553926556134e-05,
+      "loss": 3.3767,
+      "step": 2144256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4887153318050818e-05,
+      "loss": 3.3701,
+      "step": 2144768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4878767370540298e-05,
+      "loss": 3.3843,
+      "step": 2145280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4870381423029777e-05,
+      "loss": 3.3869,
+      "step": 2145792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4861995475519257e-05,
+      "loss": 3.3814,
+      "step": 2146304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4853609528008739e-05,
+      "loss": 3.3852,
+      "step": 2146816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4845239959301948e-05,
+      "loss": 3.4046,
+      "step": 2147328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.483685401179143e-05,
+      "loss": 3.3838,
+      "step": 2147840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.482846806428091e-05,
+      "loss": 3.391,
+      "step": 2148352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.482008211677039e-05,
+      "loss": 3.3845,
+      "step": 2148864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.48117125480636e-05,
+      "loss": 3.3952,
+      "step": 2149376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.480332660055308e-05,
+      "loss": 3.3818,
+      "step": 2149888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.479494065304256e-05,
+      "loss": 3.38,
+      "step": 2150400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.478655470553204e-05,
+      "loss": 3.3793,
+      "step": 2150912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4778185136825251e-05,
+      "loss": 3.3839,
+      "step": 2151424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4769799189314731e-05,
+      "loss": 3.3772,
+      "step": 2151936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4761413241804211e-05,
+      "loss": 3.3853,
+      "step": 2152448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4753027294293693e-05,
+      "loss": 3.3831,
+      "step": 2152960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4744657725586902e-05,
+      "loss": 3.3924,
+      "step": 2153472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4736271778076382e-05,
+      "loss": 3.3917,
+      "step": 2153984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4727885830565863e-05,
+      "loss": 3.3798,
+      "step": 2154496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4719499883055343e-05,
+      "loss": 3.386,
+      "step": 2155008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4711130314348554e-05,
+      "loss": 3.3906,
+      "step": 2155520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4702744366838034e-05,
+      "loss": 3.3699,
+      "step": 2156032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4694358419327514e-05,
+      "loss": 3.3872,
+      "step": 2156544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4685988850620725e-05,
+      "loss": 3.3778,
+      "step": 2157056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4677602903110205e-05,
+      "loss": 3.3783,
+      "step": 2157568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4669216955599685e-05,
+      "loss": 3.3846,
+      "step": 2158080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4660831008089165e-05,
+      "loss": 3.3885,
+      "step": 2158592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4652445060578643e-05,
+      "loss": 3.3883,
+      "step": 2159104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4644059113068123e-05,
+      "loss": 3.3856,
+      "step": 2159616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4635673165557603e-05,
+      "loss": 3.3913,
+      "step": 2160128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4627287218047085e-05,
+      "loss": 3.3875,
+      "step": 2160640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4618917649340294e-05,
+      "loss": 3.3834,
+      "step": 2161152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4610531701829774e-05,
+      "loss": 3.3582,
+      "step": 2161664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4602145754319255e-05,
+      "loss": 3.3913,
+      "step": 2162176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4593759806808735e-05,
+      "loss": 3.3857,
+      "step": 2162688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4585390238101946e-05,
+      "loss": 3.3885,
+      "step": 2163200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4577004290591426e-05,
+      "loss": 3.3829,
+      "step": 2163712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4568618343080906e-05,
+      "loss": 3.3763,
+      "step": 2164224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4560232395570386e-05,
+      "loss": 3.37,
+      "step": 2164736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4551862826863597e-05,
+      "loss": 3.3722,
+      "step": 2165248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4543476879353077e-05,
+      "loss": 3.3783,
+      "step": 2165760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4535090931842557e-05,
+      "loss": 3.366,
+      "step": 2166272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4526704984332038e-05,
+      "loss": 3.3988,
+      "step": 2166784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4518335415625247e-05,
+      "loss": 3.3818,
+      "step": 2167296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4509949468114727e-05,
+      "loss": 3.3643,
+      "step": 2167808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4501563520604209e-05,
+      "loss": 3.385,
+      "step": 2168320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4493177573093689e-05,
+      "loss": 3.3705,
+      "step": 2168832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.44848080043869e-05,
+      "loss": 3.3708,
+      "step": 2169344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.447642205687638e-05,
+      "loss": 3.3782,
+      "step": 2169856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.446803610936586e-05,
+      "loss": 3.3831,
+      "step": 2170368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.445966654065907e-05,
+      "loss": 3.365,
+      "step": 2170880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.445128059314855e-05,
+      "loss": 3.3813,
+      "step": 2171392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.444289464563803e-05,
+      "loss": 3.3675,
+      "step": 2171904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.443450869812751e-05,
+      "loss": 3.3795,
+      "step": 2172416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4426139129420721e-05,
+      "loss": 3.3848,
+      "step": 2172928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4417753181910201e-05,
+      "loss": 3.3761,
+      "step": 2173440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4409367234399681e-05,
+      "loss": 3.3767,
+      "step": 2173952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4400981286889163e-05,
+      "loss": 3.3758,
+      "step": 2174464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4392611718182372e-05,
+      "loss": 3.3982,
+      "step": 2174976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4384225770671852e-05,
+      "loss": 3.3679,
+      "step": 2175488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4375839823161333e-05,
+      "loss": 3.392,
+      "step": 2176000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4367453875650813e-05,
+      "loss": 3.3763,
+      "step": 2176512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4359084306944024e-05,
+      "loss": 3.3728,
+      "step": 2177024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4350714738237237e-05,
+      "loss": 3.3814,
+      "step": 2177536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4342328790726715e-05,
+      "loss": 3.3815,
+      "step": 2178048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4333942843216195e-05,
+      "loss": 3.3917,
+      "step": 2178560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4325556895705675e-05,
+      "loss": 3.3882,
+      "step": 2179072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4317170948195155e-05,
+      "loss": 3.3601,
+      "step": 2179584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4308785000684635e-05,
+      "loss": 3.3644,
+      "step": 2180096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4300399053174116e-05,
+      "loss": 3.3703,
+      "step": 2180608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4292013105663593e-05,
+      "loss": 3.387,
+      "step": 2181120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4283643536956805e-05,
+      "loss": 3.3729,
+      "step": 2181632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4275257589446287e-05,
+      "loss": 3.373,
+      "step": 2182144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4266871641935767e-05,
+      "loss": 3.3823,
+      "step": 2182656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4258502073228978e-05,
+      "loss": 3.3752,
+      "step": 2183168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4250116125718458e-05,
+      "loss": 3.3782,
+      "step": 2183680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4241730178207938e-05,
+      "loss": 3.3719,
+      "step": 2184192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4233344230697418e-05,
+      "loss": 3.3797,
+      "step": 2184704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4224974661990628e-05,
+      "loss": 3.384,
+      "step": 2185216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4216588714480108e-05,
+      "loss": 3.389,
+      "step": 2185728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4208202766969588e-05,
+      "loss": 3.3664,
+      "step": 2186240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4199816819459067e-05,
+      "loss": 3.3717,
+      "step": 2186752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4191447250752279e-05,
+      "loss": 3.3798,
+      "step": 2187264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4183061303241759e-05,
+      "loss": 3.3797,
+      "step": 2187776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.417467535573124e-05,
+      "loss": 3.3871,
+      "step": 2188288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4166289408220717e-05,
+      "loss": 3.3777,
+      "step": 2188800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4157919839513931e-05,
+      "loss": 3.3739,
+      "step": 2189312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4149533892003411e-05,
+      "loss": 3.3731,
+      "step": 2189824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4141147944492891e-05,
+      "loss": 3.3707,
+      "step": 2190336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.413276199698237e-05,
+      "loss": 3.3765,
+      "step": 2190848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4124392428275582e-05,
+      "loss": 3.374,
+      "step": 2191360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4116006480765062e-05,
+      "loss": 3.3815,
+      "step": 2191872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.410762053325454e-05,
+      "loss": 3.3728,
+      "step": 2192384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.409923458574402e-05,
+      "loss": 3.3697,
+      "step": 2192896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4090865017037233e-05,
+      "loss": 3.3754,
+      "step": 2193408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4082479069526713e-05,
+      "loss": 3.3622,
+      "step": 2193920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4074093122016191e-05,
+      "loss": 3.385,
+      "step": 2194432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4065707174505671e-05,
+      "loss": 3.384,
+      "step": 2194944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4057337605798885e-05,
+      "loss": 3.3778,
+      "step": 2195456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4048951658288365e-05,
+      "loss": 3.3879,
+      "step": 2195968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4040565710777842e-05,
+      "loss": 3.3864,
+      "step": 2196480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4032179763267323e-05,
+      "loss": 3.3738,
+      "step": 2196992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4023810194560536e-05,
+      "loss": 3.3733,
+      "step": 2197504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4015424247050012e-05,
+      "loss": 3.3865,
+      "step": 2198016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4007038299539494e-05,
+      "loss": 3.3721,
+      "step": 2198528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3998652352028974e-05,
+      "loss": 3.3745,
+      "step": 2199040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3990282783322186e-05,
+      "loss": 3.3803,
+      "step": 2199552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3981896835811665e-05,
+      "loss": 3.3713,
+      "step": 2200064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3973510888301145e-05,
+      "loss": 3.3756,
+      "step": 2200576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3965124940790625e-05,
+      "loss": 3.3703,
+      "step": 2201088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3956755372083837e-05,
+      "loss": 3.3765,
+      "step": 2201600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3948369424573315e-05,
+      "loss": 3.3756,
+      "step": 2202112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3939983477062795e-05,
+      "loss": 3.3711,
+      "step": 2202624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3931597529552275e-05,
+      "loss": 3.3925,
+      "step": 2203136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3923227960845486e-05,
+      "loss": 3.3798,
+      "step": 2203648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3914842013334966e-05,
+      "loss": 3.3851,
+      "step": 2204160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3906456065824448e-05,
+      "loss": 3.3725,
+      "step": 2204672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3898070118313928e-05,
+      "loss": 3.3774,
+      "step": 2205184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3889700549607138e-05,
+      "loss": 3.38,
+      "step": 2205696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3881314602096618e-05,
+      "loss": 3.382,
+      "step": 2206208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3872928654586098e-05,
+      "loss": 3.3745,
+      "step": 2206720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3864542707075578e-05,
+      "loss": 3.369,
+      "step": 2207232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3856173138368789e-05,
+      "loss": 3.3829,
+      "step": 2207744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3847787190858269e-05,
+      "loss": 3.3986,
+      "step": 2208256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3839401243347749e-05,
+      "loss": 3.3779,
+      "step": 2208768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3831015295837229e-05,
+      "loss": 3.3815,
+      "step": 2209280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.382264572713044e-05,
+      "loss": 3.3878,
+      "step": 2209792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.381425977961992e-05,
+      "loss": 3.3735,
+      "step": 2210304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3805873832109401e-05,
+      "loss": 3.3826,
+      "step": 2210816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3797487884598881e-05,
+      "loss": 3.3794,
+      "step": 2211328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.378911831589209e-05,
+      "loss": 3.3797,
+      "step": 2211840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3780732368381572e-05,
+      "loss": 3.3804,
+      "step": 2212352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3772346420871052e-05,
+      "loss": 3.3718,
+      "step": 2212864
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.86861515045166,
+      "eval_runtime": 304.2795,
+      "eval_samples_per_second": 1254.081,
+      "eval_steps_per_second": 39.191,
+      "step": 2213280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3763976852164263e-05,
+      "loss": 3.3876,
+      "step": 2213376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3755590904653743e-05,
+      "loss": 3.3628,
+      "step": 2213888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3747204957143223e-05,
+      "loss": 3.3757,
+      "step": 2214400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3738819009632703e-05,
+      "loss": 3.3942,
+      "step": 2214912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3730449440925913e-05,
+      "loss": 3.3835,
+      "step": 2215424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3722063493415393e-05,
+      "loss": 3.3774,
+      "step": 2215936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3713677545904873e-05,
+      "loss": 3.3809,
+      "step": 2216448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3705291598394355e-05,
+      "loss": 3.3728,
+      "step": 2216960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3696922029687564e-05,
+      "loss": 3.3687,
+      "step": 2217472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3688536082177044e-05,
+      "loss": 3.3858,
+      "step": 2217984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3680150134666526e-05,
+      "loss": 3.3763,
+      "step": 2218496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3671764187156006e-05,
+      "loss": 3.3788,
+      "step": 2219008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3663410997252949e-05,
+      "loss": 3.3802,
+      "step": 2219520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3655025049742429e-05,
+      "loss": 3.3666,
+      "step": 2220032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3646639102231905e-05,
+      "loss": 3.3674,
+      "step": 2220544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3638253154721387e-05,
+      "loss": 3.3591,
+      "step": 2221056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3629867207210867e-05,
+      "loss": 3.377,
+      "step": 2221568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3621481259700347e-05,
+      "loss": 3.3742,
+      "step": 2222080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3613095312189827e-05,
+      "loss": 3.3731,
+      "step": 2222592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3604709364679309e-05,
+      "loss": 3.3738,
+      "step": 2223104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3596339795972518e-05,
+      "loss": 3.3931,
+      "step": 2223616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3587953848461998e-05,
+      "loss": 3.3749,
+      "step": 2224128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.357956790095148e-05,
+      "loss": 3.3808,
+      "step": 2224640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.357118195344096e-05,
+      "loss": 3.3781,
+      "step": 2225152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3562828763537903e-05,
+      "loss": 3.3859,
+      "step": 2225664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3554442816027379e-05,
+      "loss": 3.3714,
+      "step": 2226176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3546056868516859e-05,
+      "loss": 3.3695,
+      "step": 2226688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.353767092100634e-05,
+      "loss": 3.3748,
+      "step": 2227200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.352928497349582e-05,
+      "loss": 3.3695,
+      "step": 2227712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.35208990259853e-05,
+      "loss": 3.3672,
+      "step": 2228224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.351251307847478e-05,
+      "loss": 3.3762,
+      "step": 2228736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.350412713096426e-05,
+      "loss": 3.3706,
+      "step": 2229248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3495757562257471e-05,
+      "loss": 3.3854,
+      "step": 2229760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3487371614746951e-05,
+      "loss": 3.3773,
+      "step": 2230272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3478985667236433e-05,
+      "loss": 3.3718,
+      "step": 2230784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.347059971972591e-05,
+      "loss": 3.3793,
+      "step": 2231296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3462230151019124e-05,
+      "loss": 3.3771,
+      "step": 2231808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3453844203508604e-05,
+      "loss": 3.3604,
+      "step": 2232320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3445458255998084e-05,
+      "loss": 3.3759,
+      "step": 2232832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3437072308487562e-05,
+      "loss": 3.3678,
+      "step": 2233344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3428702739780774e-05,
+      "loss": 3.3684,
+      "step": 2233856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3420316792270254e-05,
+      "loss": 3.3744,
+      "step": 2234368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3411930844759734e-05,
+      "loss": 3.3813,
+      "step": 2234880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3403544897249213e-05,
+      "loss": 3.375,
+      "step": 2235392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3395175328542425e-05,
+      "loss": 3.3777,
+      "step": 2235904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3386789381031905e-05,
+      "loss": 3.3799,
+      "step": 2236416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3378403433521383e-05,
+      "loss": 3.3774,
+      "step": 2236928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3370017486010863e-05,
+      "loss": 3.3727,
+      "step": 2237440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3361647917304076e-05,
+      "loss": 3.352,
+      "step": 2237952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3353261969793557e-05,
+      "loss": 3.3783,
+      "step": 2238464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3344876022283034e-05,
+      "loss": 3.3779,
+      "step": 2238976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3336490074772514e-05,
+      "loss": 3.3745,
+      "step": 2239488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3328120506065728e-05,
+      "loss": 3.3732,
+      "step": 2240000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3319734558555208e-05,
+      "loss": 3.3673,
+      "step": 2240512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3311348611044686e-05,
+      "loss": 3.3607,
+      "step": 2241024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3302962663534166e-05,
+      "loss": 3.3592,
+      "step": 2241536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3294593094827379e-05,
+      "loss": 3.3737,
+      "step": 2242048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3286207147316857e-05,
+      "loss": 3.3594,
+      "step": 2242560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3277821199806337e-05,
+      "loss": 3.3796,
+      "step": 2243072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3269435252295817e-05,
+      "loss": 3.3748,
+      "step": 2243584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.326106568358903e-05,
+      "loss": 3.3586,
+      "step": 2244096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3252679736078508e-05,
+      "loss": 3.3737,
+      "step": 2244608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3244293788567988e-05,
+      "loss": 3.3585,
+      "step": 2245120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3235907841057468e-05,
+      "loss": 3.3627,
+      "step": 2245632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3227538272350682e-05,
+      "loss": 3.3619,
+      "step": 2246144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3219152324840158e-05,
+      "loss": 3.3763,
+      "step": 2246656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3210782756133372e-05,
+      "loss": 3.358,
+      "step": 2247168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3202396808622852e-05,
+      "loss": 3.3703,
+      "step": 2247680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3194010861112329e-05,
+      "loss": 3.3555,
+      "step": 2248192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.318562491360181e-05,
+      "loss": 3.3672,
+      "step": 2248704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.317723896609129e-05,
+      "loss": 3.3743,
+      "step": 2249216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.316885301858077e-05,
+      "loss": 3.3694,
+      "step": 2249728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.316046707107025e-05,
+      "loss": 3.3657,
+      "step": 2250240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3152097502363461e-05,
+      "loss": 3.3687,
+      "step": 2250752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3143711554852941e-05,
+      "loss": 3.3879,
+      "step": 2251264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3135325607342421e-05,
+      "loss": 3.3574,
+      "step": 2251776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3126939659831903e-05,
+      "loss": 3.3831,
+      "step": 2252288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3118570091125112e-05,
+      "loss": 3.3669,
+      "step": 2252800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3110184143614594e-05,
+      "loss": 3.3612,
+      "step": 2253312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3101814574907803e-05,
+      "loss": 3.3681,
+      "step": 2253824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3093428627397283e-05,
+      "loss": 3.3716,
+      "step": 2254336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3085042679886764e-05,
+      "loss": 3.3777,
+      "step": 2254848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3076656732376244e-05,
+      "loss": 3.3796,
+      "step": 2255360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3068270784865724e-05,
+      "loss": 3.3527,
+      "step": 2255872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3059884837355204e-05,
+      "loss": 3.3533,
+      "step": 2256384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3051498889844684e-05,
+      "loss": 3.3609,
+      "step": 2256896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3043112942334162e-05,
+      "loss": 3.3733,
+      "step": 2257408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3034743373627375e-05,
+      "loss": 3.3663,
+      "step": 2257920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3026357426116857e-05,
+      "loss": 3.3628,
+      "step": 2258432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3017971478606336e-05,
+      "loss": 3.3715,
+      "step": 2258944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3009601909899547e-05,
+      "loss": 3.3649,
+      "step": 2259456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3001215962389027e-05,
+      "loss": 3.3715,
+      "step": 2259968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2992830014878507e-05,
+      "loss": 3.3615,
+      "step": 2260480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2984444067367984e-05,
+      "loss": 3.3731,
+      "step": 2260992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2976074498661198e-05,
+      "loss": 3.3717,
+      "step": 2261504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2967688551150678e-05,
+      "loss": 3.3813,
+      "step": 2262016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2959302603640158e-05,
+      "loss": 3.3566,
+      "step": 2262528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2950916656129636e-05,
+      "loss": 3.361,
+      "step": 2263040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2942547087422849e-05,
+      "loss": 3.3732,
+      "step": 2263552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2934161139912329e-05,
+      "loss": 3.369,
+      "step": 2264064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2925775192401807e-05,
+      "loss": 3.3773,
+      "step": 2264576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2917389244891287e-05,
+      "loss": 3.3704,
+      "step": 2265088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.29090196761845e-05,
+      "loss": 3.3642,
+      "step": 2265600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2900633728673981e-05,
+      "loss": 3.3639,
+      "step": 2266112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2892247781163457e-05,
+      "loss": 3.3648,
+      "step": 2266624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2883861833652937e-05,
+      "loss": 3.3648,
+      "step": 2267136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2875492264946152e-05,
+      "loss": 3.364,
+      "step": 2267648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2867106317435632e-05,
+      "loss": 3.3693,
+      "step": 2268160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.285872036992511e-05,
+      "loss": 3.363,
+      "step": 2268672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.285033442241459e-05,
+      "loss": 3.3626,
+      "step": 2269184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2841964853707802e-05,
+      "loss": 3.3648,
+      "step": 2269696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.283357890619728e-05,
+      "loss": 3.3529,
+      "step": 2270208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2825209337490493e-05,
+      "loss": 3.3768,
+      "step": 2270720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2816823389979973e-05,
+      "loss": 3.3707,
+      "step": 2271232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2808437442469453e-05,
+      "loss": 3.3668,
+      "step": 2271744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2800051494958931e-05,
+      "loss": 3.3789,
+      "step": 2272256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2791665547448411e-05,
+      "loss": 3.3781,
+      "step": 2272768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2783279599937891e-05,
+      "loss": 3.3624,
+      "step": 2273280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2774893652427373e-05,
+      "loss": 3.3632,
+      "step": 2273792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2766507704916853e-05,
+      "loss": 3.3747,
+      "step": 2274304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2758138136210063e-05,
+      "loss": 3.3649,
+      "step": 2274816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2749752188699543e-05,
+      "loss": 3.3603,
+      "step": 2275328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2741366241189023e-05,
+      "loss": 3.3709,
+      "step": 2275840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2732980293678503e-05,
+      "loss": 3.3651,
+      "step": 2276352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2724610724971714e-05,
+      "loss": 3.3669,
+      "step": 2276864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2716224777461194e-05,
+      "loss": 3.363,
+      "step": 2277376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2707838829950674e-05,
+      "loss": 3.3656,
+      "step": 2277888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2699452882440156e-05,
+      "loss": 3.3697,
+      "step": 2278400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2691083313733365e-05,
+      "loss": 3.3611,
+      "step": 2278912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2682697366222845e-05,
+      "loss": 3.3769,
+      "step": 2279424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2674311418712326e-05,
+      "loss": 3.3771,
+      "step": 2279936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2665941850005535e-05,
+      "loss": 3.3729,
+      "step": 2280448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2657555902495017e-05,
+      "loss": 3.3637,
+      "step": 2280960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2649169954984497e-05,
+      "loss": 3.3713,
+      "step": 2281472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2640784007473977e-05,
+      "loss": 3.3677,
+      "step": 2281984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2632414438767188e-05,
+      "loss": 3.3692,
+      "step": 2282496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2624028491256668e-05,
+      "loss": 3.3645,
+      "step": 2283008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2615642543746148e-05,
+      "loss": 3.3595,
+      "step": 2283520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2607256596235628e-05,
+      "loss": 3.3753,
+      "step": 2284032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2598887027528838e-05,
+      "loss": 3.3864,
+      "step": 2284544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2590501080018318e-05,
+      "loss": 3.3634,
+      "step": 2285056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2582115132507798e-05,
+      "loss": 3.3739,
+      "step": 2285568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.257372918499728e-05,
+      "loss": 3.377,
+      "step": 2286080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2565359616290489e-05,
+      "loss": 3.3644,
+      "step": 2286592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2556973668779969e-05,
+      "loss": 3.3727,
+      "step": 2287104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.254858772126945e-05,
+      "loss": 3.3686,
+      "step": 2287616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.254020177375893e-05,
+      "loss": 3.3675,
+      "step": 2288128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2531832205052142e-05,
+      "loss": 3.377,
+      "step": 2288640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2523446257541621e-05,
+      "loss": 3.3586,
+      "step": 2289152
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.869394540786743,
+      "eval_runtime": 306.6439,
+      "eval_samples_per_second": 1244.411,
+      "eval_steps_per_second": 38.889,
+      "step": 2289600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2515076688834832e-05,
+      "loss": 3.368,
+      "step": 2289664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2506690741324312e-05,
+      "loss": 3.3559,
+      "step": 2290176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2498304793813792e-05,
+      "loss": 3.3654,
+      "step": 2290688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2489918846303272e-05,
+      "loss": 3.3809,
+      "step": 2291200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2481532898792752e-05,
+      "loss": 3.379,
+      "step": 2291712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2473146951282232e-05,
+      "loss": 3.3657,
+      "step": 2292224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2464761003771712e-05,
+      "loss": 3.3708,
+      "step": 2292736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.245637505626119e-05,
+      "loss": 3.3608,
+      "step": 2293248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2448005487554403e-05,
+      "loss": 3.3597,
+      "step": 2293760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2439619540043883e-05,
+      "loss": 3.3725,
+      "step": 2294272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2431233592533363e-05,
+      "loss": 3.3699,
+      "step": 2294784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2422847645022843e-05,
+      "loss": 3.3697,
+      "step": 2295296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2414494455119784e-05,
+      "loss": 3.3723,
+      "step": 2295808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2406108507609266e-05,
+      "loss": 3.357,
+      "step": 2296320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2397722560098746e-05,
+      "loss": 3.3563,
+      "step": 2296832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2389336612588224e-05,
+      "loss": 3.3501,
+      "step": 2297344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2380950665077706e-05,
+      "loss": 3.369,
+      "step": 2297856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2372564717567186e-05,
+      "loss": 3.3606,
+      "step": 2298368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2364195148860396e-05,
+      "loss": 3.3635,
+      "step": 2298880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2355809201349876e-05,
+      "loss": 3.3655,
+      "step": 2299392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2347423253839356e-05,
+      "loss": 3.3819,
+      "step": 2299904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2339037306328836e-05,
+      "loss": 3.3685,
+      "step": 2300416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2330651358818315e-05,
+      "loss": 3.3657,
+      "step": 2300928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2322265411307796e-05,
+      "loss": 3.3683,
+      "step": 2301440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2313879463797276e-05,
+      "loss": 3.3765,
+      "step": 2301952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2305493516286754e-05,
+      "loss": 3.3625,
+      "step": 2302464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2297123947579967e-05,
+      "loss": 3.3571,
+      "step": 2302976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2288738000069447e-05,
+      "loss": 3.3666,
+      "step": 2303488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2280352052558927e-05,
+      "loss": 3.3588,
+      "step": 2304000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2271966105048407e-05,
+      "loss": 3.3626,
+      "step": 2304512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2263596536341618e-05,
+      "loss": 3.3655,
+      "step": 2305024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2255210588831098e-05,
+      "loss": 3.3646,
+      "step": 2305536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2246824641320578e-05,
+      "loss": 3.3701,
+      "step": 2306048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2238455072613788e-05,
+      "loss": 3.3706,
+      "step": 2306560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2230069125103268e-05,
+      "loss": 3.3632,
+      "step": 2307072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.222168317759275e-05,
+      "loss": 3.368,
+      "step": 2307584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2213297230082228e-05,
+      "loss": 3.3694,
+      "step": 2308096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.220492766137544e-05,
+      "loss": 3.3497,
+      "step": 2308608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.219654171386492e-05,
+      "loss": 3.3718,
+      "step": 2309120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2188172145158131e-05,
+      "loss": 3.3535,
+      "step": 2309632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2179786197647611e-05,
+      "loss": 3.3562,
+      "step": 2310144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2171400250137091e-05,
+      "loss": 3.3614,
+      "step": 2310656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2163014302626571e-05,
+      "loss": 3.3779,
+      "step": 2311168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.215462835511605e-05,
+      "loss": 3.3637,
+      "step": 2311680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2146242407605531e-05,
+      "loss": 3.3633,
+      "step": 2312192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2137856460095011e-05,
+      "loss": 3.3741,
+      "step": 2312704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2129470512584491e-05,
+      "loss": 3.3673,
+      "step": 2313216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2121100943877702e-05,
+      "loss": 3.3664,
+      "step": 2313728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2112714996367182e-05,
+      "loss": 3.34,
+      "step": 2314240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2104329048856662e-05,
+      "loss": 3.3663,
+      "step": 2314752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2095943101346142e-05,
+      "loss": 3.3717,
+      "step": 2315264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2087573532639353e-05,
+      "loss": 3.3645,
+      "step": 2315776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2079187585128833e-05,
+      "loss": 3.3635,
+      "step": 2316288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2070801637618314e-05,
+      "loss": 3.3547,
+      "step": 2316800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2062415690107792e-05,
+      "loss": 3.3588,
+      "step": 2317312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2054046121401003e-05,
+      "loss": 3.3456,
+      "step": 2317824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2045660173890485e-05,
+      "loss": 3.3664,
+      "step": 2318336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2037274226379965e-05,
+      "loss": 3.3481,
+      "step": 2318848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2028888278869443e-05,
+      "loss": 3.3671,
+      "step": 2319360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2020518710162656e-05,
+      "loss": 3.3641,
+      "step": 2319872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2012132762652136e-05,
+      "loss": 3.3525,
+      "step": 2320384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2003746815141614e-05,
+      "loss": 3.3649,
+      "step": 2320896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1995360867631095e-05,
+      "loss": 3.3495,
+      "step": 2321408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1986991298924306e-05,
+      "loss": 3.353,
+      "step": 2321920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1978605351413786e-05,
+      "loss": 3.3532,
+      "step": 2322432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1970219403903266e-05,
+      "loss": 3.3687,
+      "step": 2322944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1961833456392746e-05,
+      "loss": 3.3494,
+      "step": 2323456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1953463887685957e-05,
+      "loss": 3.3587,
+      "step": 2323968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1945077940175439e-05,
+      "loss": 3.3486,
+      "step": 2324480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1936691992664917e-05,
+      "loss": 3.3559,
+      "step": 2324992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1928322423958128e-05,
+      "loss": 3.3686,
+      "step": 2325504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.191993647644761e-05,
+      "loss": 3.3634,
+      "step": 2326016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1911550528937087e-05,
+      "loss": 3.3545,
+      "step": 2326528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1903164581426567e-05,
+      "loss": 3.3611,
+      "step": 2327040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.189479501271978e-05,
+      "loss": 3.3763,
+      "step": 2327552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.188640906520926e-05,
+      "loss": 3.3512,
+      "step": 2328064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1878023117698738e-05,
+      "loss": 3.3721,
+      "step": 2328576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.186963717018822e-05,
+      "loss": 3.3563,
+      "step": 2329088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.186126760148143e-05,
+      "loss": 3.3556,
+      "step": 2329600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1852898032774643e-05,
+      "loss": 3.3574,
+      "step": 2330112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1844512085264121e-05,
+      "loss": 3.3608,
+      "step": 2330624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1836126137753601e-05,
+      "loss": 3.3704,
+      "step": 2331136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1827740190243081e-05,
+      "loss": 3.3662,
+      "step": 2331648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1819354242732561e-05,
+      "loss": 3.3522,
+      "step": 2332160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1810968295222041e-05,
+      "loss": 3.3376,
+      "step": 2332672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1802582347711521e-05,
+      "loss": 3.3499,
+      "step": 2333184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1794212779004734e-05,
+      "loss": 3.3645,
+      "step": 2333696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1785826831494212e-05,
+      "loss": 3.3565,
+      "step": 2334208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1777440883983692e-05,
+      "loss": 3.3517,
+      "step": 2334720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1769054936473173e-05,
+      "loss": 3.3623,
+      "step": 2335232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1760668988962652e-05,
+      "loss": 3.3593,
+      "step": 2335744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1752283041452132e-05,
+      "loss": 3.3587,
+      "step": 2336256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1743897093941612e-05,
+      "loss": 3.3511,
+      "step": 2336768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1735511146431093e-05,
+      "loss": 3.366,
+      "step": 2337280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1727141577724302e-05,
+      "loss": 3.3593,
+      "step": 2337792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1718755630213784e-05,
+      "loss": 3.3708,
+      "step": 2338304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1710369682703264e-05,
+      "loss": 3.3502,
+      "step": 2338816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1701983735192742e-05,
+      "loss": 3.3506,
+      "step": 2339328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1693614166485955e-05,
+      "loss": 3.3654,
+      "step": 2339840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1685228218975435e-05,
+      "loss": 3.3564,
+      "step": 2340352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1676842271464915e-05,
+      "loss": 3.3653,
+      "step": 2340864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1668456323954393e-05,
+      "loss": 3.3616,
+      "step": 2341376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1660086755247605e-05,
+      "loss": 3.3531,
+      "step": 2341888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1651700807737085e-05,
+      "loss": 3.3536,
+      "step": 2342400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1643314860226565e-05,
+      "loss": 3.3567,
+      "step": 2342912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1634928912716045e-05,
+      "loss": 3.3503,
+      "step": 2343424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1626559344009256e-05,
+      "loss": 3.3569,
+      "step": 2343936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1618173396498738e-05,
+      "loss": 3.3606,
+      "step": 2344448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1609787448988216e-05,
+      "loss": 3.3552,
+      "step": 2344960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1601401501477696e-05,
+      "loss": 3.3495,
+      "step": 2345472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1593031932770908e-05,
+      "loss": 3.3606,
+      "step": 2345984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1584645985260388e-05,
+      "loss": 3.3449,
+      "step": 2346496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.15762764165536e-05,
+      "loss": 3.3657,
+      "step": 2347008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1567890469043079e-05,
+      "loss": 3.3595,
+      "step": 2347520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1559504521532559e-05,
+      "loss": 3.3615,
+      "step": 2348032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1551118574022037e-05,
+      "loss": 3.3661,
+      "step": 2348544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1542732626511519e-05,
+      "loss": 3.3695,
+      "step": 2349056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1534346679000999e-05,
+      "loss": 3.3538,
+      "step": 2349568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1525960731490479e-05,
+      "loss": 3.3575,
+      "step": 2350080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1517574783979957e-05,
+      "loss": 3.3617,
+      "step": 2350592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.150920521527317e-05,
+      "loss": 3.3551,
+      "step": 2351104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.150081926776265e-05,
+      "loss": 3.3485,
+      "step": 2351616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1492433320252128e-05,
+      "loss": 3.3665,
+      "step": 2352128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.148404737274161e-05,
+      "loss": 3.3541,
+      "step": 2352640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.147567780403482e-05,
+      "loss": 3.36,
+      "step": 2353152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.14672918565243e-05,
+      "loss": 3.3507,
+      "step": 2353664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.145890590901378e-05,
+      "loss": 3.3534,
+      "step": 2354176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.145051996150326e-05,
+      "loss": 3.3629,
+      "step": 2354688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1442150392796473e-05,
+      "loss": 3.3543,
+      "step": 2355200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1433764445285953e-05,
+      "loss": 3.3611,
+      "step": 2355712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1425378497775431e-05,
+      "loss": 3.3688,
+      "step": 2356224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.141699255026491e-05,
+      "loss": 3.3602,
+      "step": 2356736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1408622981558123e-05,
+      "loss": 3.3548,
+      "step": 2357248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1400237034047602e-05,
+      "loss": 3.3603,
+      "step": 2357760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1391851086537082e-05,
+      "loss": 3.3603,
+      "step": 2358272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1383481517830294e-05,
+      "loss": 3.3567,
+      "step": 2358784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1375095570319774e-05,
+      "loss": 3.3506,
+      "step": 2359296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1366709622809254e-05,
+      "loss": 3.3538,
+      "step": 2359808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1358323675298734e-05,
+      "loss": 3.3707,
+      "step": 2360320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1349954106591945e-05,
+      "loss": 3.3739,
+      "step": 2360832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1341568159081425e-05,
+      "loss": 3.3588,
+      "step": 2361344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1333182211570905e-05,
+      "loss": 3.3608,
+      "step": 2361856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1324796264060385e-05,
+      "loss": 3.3667,
+      "step": 2362368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1316426695353597e-05,
+      "loss": 3.3564,
+      "step": 2362880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1308040747843075e-05,
+      "loss": 3.3624,
+      "step": 2363392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1299654800332555e-05,
+      "loss": 3.3582,
+      "step": 2363904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1291268852822035e-05,
+      "loss": 3.3558,
+      "step": 2364416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1282899284115248e-05,
+      "loss": 3.368,
+      "step": 2364928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1274513336604726e-05,
+      "loss": 3.3473,
+      "step": 2365440
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.871316909790039,
+      "eval_runtime": 304.2974,
+      "eval_samples_per_second": 1254.007,
+      "eval_steps_per_second": 39.189,
+      "step": 2365920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1266143767897938e-05,
+      "loss": 3.3431,
+      "step": 2365952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1257757820387418e-05,
+      "loss": 3.3502,
+      "step": 2366464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1249371872876898e-05,
+      "loss": 3.3534,
+      "step": 2366976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1240985925366378e-05,
+      "loss": 3.3714,
+      "step": 2367488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1232616356659589e-05,
+      "loss": 3.3709,
+      "step": 2368000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1224230409149069e-05,
+      "loss": 3.3572,
+      "step": 2368512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1215844461638549e-05,
+      "loss": 3.3602,
+      "step": 2369024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1207458514128029e-05,
+      "loss": 3.3543,
+      "step": 2369536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.119908894542124e-05,
+      "loss": 3.3531,
+      "step": 2370048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1190702997910721e-05,
+      "loss": 3.3624,
+      "step": 2370560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11823170504002e-05,
+      "loss": 3.3554,
+      "step": 2371072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.117393110288968e-05,
+      "loss": 3.36,
+      "step": 2371584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1165561534182892e-05,
+      "loss": 3.3627,
+      "step": 2372096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1157191965476103e-05,
+      "loss": 3.3514,
+      "step": 2372608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1148806017965583e-05,
+      "loss": 3.3452,
+      "step": 2373120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1140420070455063e-05,
+      "loss": 3.3437,
+      "step": 2373632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1132034122944543e-05,
+      "loss": 3.3563,
+      "step": 2374144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1123648175434023e-05,
+      "loss": 3.354,
+      "step": 2374656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1115262227923503e-05,
+      "loss": 3.3558,
+      "step": 2375168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1106876280412983e-05,
+      "loss": 3.3557,
+      "step": 2375680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1098506711706193e-05,
+      "loss": 3.3712,
+      "step": 2376192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1090120764195673e-05,
+      "loss": 3.3572,
+      "step": 2376704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1081734816685153e-05,
+      "loss": 3.3593,
+      "step": 2377216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1073348869174633e-05,
+      "loss": 3.3573,
+      "step": 2377728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1064995679271577e-05,
+      "loss": 3.3687,
+      "step": 2378240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1056609731761055e-05,
+      "loss": 3.3503,
+      "step": 2378752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1048223784250537e-05,
+      "loss": 3.3559,
+      "step": 2379264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1039837836740016e-05,
+      "loss": 3.354,
+      "step": 2379776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1031451889229495e-05,
+      "loss": 3.3485,
+      "step": 2380288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1023065941718975e-05,
+      "loss": 3.354,
+      "step": 2380800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1014679994208456e-05,
+      "loss": 3.3529,
+      "step": 2381312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1006294046697935e-05,
+      "loss": 3.3538,
+      "step": 2381824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0997924477991147e-05,
+      "loss": 3.361,
+      "step": 2382336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0989538530480627e-05,
+      "loss": 3.3605,
+      "step": 2382848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0981152582970107e-05,
+      "loss": 3.3578,
+      "step": 2383360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0972766635459585e-05,
+      "loss": 3.3563,
+      "step": 2383872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0964397066752798e-05,
+      "loss": 3.361,
+      "step": 2384384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0956011119242278e-05,
+      "loss": 3.3406,
+      "step": 2384896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0947625171731758e-05,
+      "loss": 3.3624,
+      "step": 2385408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0939255603024968e-05,
+      "loss": 3.3401,
+      "step": 2385920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0930869655514448e-05,
+      "loss": 3.3519,
+      "step": 2386432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0922483708003928e-05,
+      "loss": 3.3503,
+      "step": 2386944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0914097760493408e-05,
+      "loss": 3.3653,
+      "step": 2387456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0905711812982888e-05,
+      "loss": 3.3518,
+      "step": 2387968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0897325865472368e-05,
+      "loss": 3.3585,
+      "step": 2388480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0888939917961848e-05,
+      "loss": 3.3626,
+      "step": 2388992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0880553970451328e-05,
+      "loss": 3.3595,
+      "step": 2389504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0872184401744539e-05,
+      "loss": 3.3557,
+      "step": 2390016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.086379845423402e-05,
+      "loss": 3.3318,
+      "step": 2390528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0855412506723499e-05,
+      "loss": 3.3539,
+      "step": 2391040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0847026559212979e-05,
+      "loss": 3.3623,
+      "step": 2391552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0838656990506191e-05,
+      "loss": 3.3562,
+      "step": 2392064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0830271042995671e-05,
+      "loss": 3.3521,
+      "step": 2392576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.082188509548515e-05,
+      "loss": 3.3444,
+      "step": 2393088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0813499147974631e-05,
+      "loss": 3.3484,
+      "step": 2393600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0805129579267842e-05,
+      "loss": 3.3356,
+      "step": 2394112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0796743631757322e-05,
+      "loss": 3.3593,
+      "step": 2394624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0788357684246802e-05,
+      "loss": 3.3377,
+      "step": 2395136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0779971736736282e-05,
+      "loss": 3.3553,
+      "step": 2395648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0771602168029493e-05,
+      "loss": 3.354,
+      "step": 2396160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0763216220518973e-05,
+      "loss": 3.3469,
+      "step": 2396672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0754830273008452e-05,
+      "loss": 3.3501,
+      "step": 2397184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0746444325497932e-05,
+      "loss": 3.3396,
+      "step": 2397696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0738074756791145e-05,
+      "loss": 3.3482,
+      "step": 2398208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0729688809280623e-05,
+      "loss": 3.3363,
+      "step": 2398720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0721302861770103e-05,
+      "loss": 3.3628,
+      "step": 2399232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0712916914259583e-05,
+      "loss": 3.341,
+      "step": 2399744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0704547345552796e-05,
+      "loss": 3.3428,
+      "step": 2400256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0696161398042274e-05,
+      "loss": 3.3416,
+      "step": 2400768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0687775450531755e-05,
+      "loss": 3.3486,
+      "step": 2401280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0679389503021235e-05,
+      "loss": 3.3599,
+      "step": 2401792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0671019934314446e-05,
+      "loss": 3.3499,
+      "step": 2402304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0662633986803926e-05,
+      "loss": 3.3419,
+      "step": 2402816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0654248039293406e-05,
+      "loss": 3.3542,
+      "step": 2403328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0645878470586617e-05,
+      "loss": 3.3639,
+      "step": 2403840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0637492523076097e-05,
+      "loss": 3.3404,
+      "step": 2404352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0629106575565577e-05,
+      "loss": 3.3623,
+      "step": 2404864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0620720628055057e-05,
+      "loss": 3.3489,
+      "step": 2405376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.061235105934827e-05,
+      "loss": 3.3468,
+      "step": 2405888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0603965111837748e-05,
+      "loss": 3.3459,
+      "step": 2406400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0595579164327228e-05,
+      "loss": 3.3503,
+      "step": 2406912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.058719321681671e-05,
+      "loss": 3.3608,
+      "step": 2407424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0578823648109918e-05,
+      "loss": 3.3568,
+      "step": 2407936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0570437700599398e-05,
+      "loss": 3.3429,
+      "step": 2408448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.056205175308888e-05,
+      "loss": 3.3261,
+      "step": 2408960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.055366580557836e-05,
+      "loss": 3.3461,
+      "step": 2409472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0545312615675301e-05,
+      "loss": 3.3493,
+      "step": 2409984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0536926668164781e-05,
+      "loss": 3.3531,
+      "step": 2410496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0528540720654261e-05,
+      "loss": 3.3382,
+      "step": 2411008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0520154773143743e-05,
+      "loss": 3.355,
+      "step": 2411520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0511768825633221e-05,
+      "loss": 3.3492,
+      "step": 2412032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0503382878122701e-05,
+      "loss": 3.35,
+      "step": 2412544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0494996930612181e-05,
+      "loss": 3.3403,
+      "step": 2413056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0486610983101661e-05,
+      "loss": 3.3554,
+      "step": 2413568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0478241414394872e-05,
+      "loss": 3.3473,
+      "step": 2414080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0469855466884352e-05,
+      "loss": 3.3614,
+      "step": 2414592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0461469519373834e-05,
+      "loss": 3.3443,
+      "step": 2415104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0453083571863312e-05,
+      "loss": 3.3383,
+      "step": 2415616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0444714003156524e-05,
+      "loss": 3.3528,
+      "step": 2416128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0436328055646004e-05,
+      "loss": 3.3507,
+      "step": 2416640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0427942108135482e-05,
+      "loss": 3.3537,
+      "step": 2417152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0419556160624962e-05,
+      "loss": 3.3524,
+      "step": 2417664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0411186591918175e-05,
+      "loss": 3.3442,
+      "step": 2418176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0402800644407655e-05,
+      "loss": 3.3434,
+      "step": 2418688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0394414696897133e-05,
+      "loss": 3.3488,
+      "step": 2419200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0386028749386615e-05,
+      "loss": 3.3369,
+      "step": 2419712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0377659180679826e-05,
+      "loss": 3.3498,
+      "step": 2420224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0369273233169306e-05,
+      "loss": 3.3494,
+      "step": 2420736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0360887285658786e-05,
+      "loss": 3.3487,
+      "step": 2421248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0352501338148265e-05,
+      "loss": 3.3353,
+      "step": 2421760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0344131769441478e-05,
+      "loss": 3.3509,
+      "step": 2422272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0335745821930956e-05,
+      "loss": 3.3384,
+      "step": 2422784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0327359874420436e-05,
+      "loss": 3.3519,
+      "step": 2423296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0318973926909916e-05,
+      "loss": 3.352,
+      "step": 2423808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0310604358203129e-05,
+      "loss": 3.3479,
+      "step": 2424320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0302218410692607e-05,
+      "loss": 3.3583,
+      "step": 2424832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0293832463182087e-05,
+      "loss": 3.3607,
+      "step": 2425344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0285446515671568e-05,
+      "loss": 3.3489,
+      "step": 2425856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0277076946964778e-05,
+      "loss": 3.3441,
+      "step": 2426368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.026869099945426e-05,
+      "loss": 3.3564,
+      "step": 2426880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.026030505194374e-05,
+      "loss": 3.3473,
+      "step": 2427392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0251919104433219e-05,
+      "loss": 3.3405,
+      "step": 2427904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.024354953572643e-05,
+      "loss": 3.3552,
+      "step": 2428416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.023516358821591e-05,
+      "loss": 3.3446,
+      "step": 2428928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.022677764070539e-05,
+      "loss": 3.3511,
+      "step": 2429440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.021839169319487e-05,
+      "loss": 3.3404,
+      "step": 2429952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.021002212448808e-05,
+      "loss": 3.3455,
+      "step": 2430464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.020163617697756e-05,
+      "loss": 3.3487,
+      "step": 2430976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.019325022946704e-05,
+      "loss": 3.3439,
+      "step": 2431488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0184880660760251e-05,
+      "loss": 3.3493,
+      "step": 2432000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0176494713249731e-05,
+      "loss": 3.3624,
+      "step": 2432512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0168108765739213e-05,
+      "loss": 3.3548,
+      "step": 2433024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0159722818228693e-05,
+      "loss": 3.3467,
+      "step": 2433536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0151353249521902e-05,
+      "loss": 3.35,
+      "step": 2434048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0142967302011384e-05,
+      "loss": 3.3517,
+      "step": 2434560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0134581354500864e-05,
+      "loss": 3.3475,
+      "step": 2435072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0126195406990342e-05,
+      "loss": 3.3437,
+      "step": 2435584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0117825838283554e-05,
+      "loss": 3.3403,
+      "step": 2436096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0109456269576765e-05,
+      "loss": 3.359,
+      "step": 2436608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0101070322066245e-05,
+      "loss": 3.3627,
+      "step": 2437120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0092684374555725e-05,
+      "loss": 3.3511,
+      "step": 2437632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0084298427045205e-05,
+      "loss": 3.3503,
+      "step": 2438144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0075912479534685e-05,
+      "loss": 3.3589,
+      "step": 2438656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0067526532024167e-05,
+      "loss": 3.3452,
+      "step": 2439168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0059140584513645e-05,
+      "loss": 3.3531,
+      "step": 2439680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0050754637003125e-05,
+      "loss": 3.3502,
+      "step": 2440192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0042385068296337e-05,
+      "loss": 3.3488,
+      "step": 2440704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0033999120785816e-05,
+      "loss": 3.3588,
+      "step": 2441216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0025613173275295e-05,
+      "loss": 3.3389,
+      "step": 2441728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0017243604568508e-05,
+      "loss": 3.3472,
+      "step": 2442240
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.870638370513916,
+      "eval_runtime": 342.4062,
+      "eval_samples_per_second": 1114.439,
+      "eval_steps_per_second": 34.827,
+      "step": 2442240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0008857657057988e-05,
+      "loss": 3.3384,
+      "step": 2442752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0000471709547466e-05,
+      "loss": 3.3433,
+      "step": 2443264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.992085762036948e-06,
+      "loss": 3.36,
+      "step": 2443776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.983699814526428e-06,
+      "loss": 3.3618,
+      "step": 2444288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.975313867015906e-06,
+      "loss": 3.3504,
+      "step": 2444800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.966927919505386e-06,
+      "loss": 3.3507,
+      "step": 2445312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.958541971994868e-06,
+      "loss": 3.3446,
+      "step": 2445824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.950172403288078e-06,
+      "loss": 3.3377,
+      "step": 2446336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.941786455777557e-06,
+      "loss": 3.3558,
+      "step": 2446848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.933400508267038e-06,
+      "loss": 3.3471,
+      "step": 2447360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.925014560756518e-06,
+      "loss": 3.3522,
+      "step": 2447872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.916644992049729e-06,
+      "loss": 3.3519,
+      "step": 2448384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.90827542334294e-06,
+      "loss": 3.3446,
+      "step": 2448896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.89988947583242e-06,
+      "loss": 3.3371,
+      "step": 2449408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.891503528321901e-06,
+      "loss": 3.3303,
+      "step": 2449920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.88311758081138e-06,
+      "loss": 3.3429,
+      "step": 2450432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.87473163330086e-06,
+      "loss": 3.3452,
+      "step": 2450944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.86634568579034e-06,
+      "loss": 3.347,
+      "step": 2451456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.857959738279821e-06,
+      "loss": 3.3419,
+      "step": 2451968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.84959016957303e-06,
+      "loss": 3.3653,
+      "step": 2452480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.84120422206251e-06,
+      "loss": 3.3477,
+      "step": 2452992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.832818274551992e-06,
+      "loss": 3.3487,
+      "step": 2453504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.82443232704147e-06,
+      "loss": 3.3485,
+      "step": 2454016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.816079137138414e-06,
+      "loss": 3.3622,
+      "step": 2454528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.807693189627894e-06,
+      "loss": 3.3359,
+      "step": 2455040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799307242117374e-06,
+      "loss": 3.3463,
+      "step": 2455552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.790921294606853e-06,
+      "loss": 3.3442,
+      "step": 2456064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.782535347096333e-06,
+      "loss": 3.3445,
+      "step": 2456576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.774149399585813e-06,
+      "loss": 3.3429,
+      "step": 2457088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.765763452075293e-06,
+      "loss": 3.3412,
+      "step": 2457600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.757377504564773e-06,
+      "loss": 3.3448,
+      "step": 2458112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.749007935857984e-06,
+      "loss": 3.3522,
+      "step": 2458624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.740621988347464e-06,
+      "loss": 3.3464,
+      "step": 2459136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.732236040836944e-06,
+      "loss": 3.353,
+      "step": 2459648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.723850093326424e-06,
+      "loss": 3.3458,
+      "step": 2460160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.715480524619636e-06,
+      "loss": 3.3533,
+      "step": 2460672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.707094577109116e-06,
+      "loss": 3.3318,
+      "step": 2461184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.698708629598595e-06,
+      "loss": 3.3536,
+      "step": 2461696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.690339060891807e-06,
+      "loss": 3.331,
+      "step": 2462208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.681953113381287e-06,
+      "loss": 3.3448,
+      "step": 2462720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.673567165870765e-06,
+      "loss": 3.3402,
+      "step": 2463232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.665181218360245e-06,
+      "loss": 3.355,
+      "step": 2463744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.656795270849727e-06,
+      "loss": 3.3402,
+      "step": 2464256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.648409323339207e-06,
+      "loss": 3.3507,
+      "step": 2464768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.640023375828685e-06,
+      "loss": 3.3505,
+      "step": 2465280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.631637428318165e-06,
+      "loss": 3.3489,
+      "step": 2465792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.623267859611378e-06,
+      "loss": 3.3473,
+      "step": 2466304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.614881912100856e-06,
+      "loss": 3.3236,
+      "step": 2466816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.606495964590338e-06,
+      "loss": 3.3482,
+      "step": 2467328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.598110017079817e-06,
+      "loss": 3.3521,
+      "step": 2467840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.589740448373028e-06,
+      "loss": 3.3441,
+      "step": 2468352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.581354500862508e-06,
+      "loss": 3.3452,
+      "step": 2468864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.572968553351988e-06,
+      "loss": 3.334,
+      "step": 2469376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.564582605841468e-06,
+      "loss": 3.341,
+      "step": 2469888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.55621303713468e-06,
+      "loss": 3.3253,
+      "step": 2470400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.547827089624159e-06,
+      "loss": 3.3515,
+      "step": 2470912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.539441142113639e-06,
+      "loss": 3.3265,
+      "step": 2471424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.531055194603119e-06,
+      "loss": 3.3477,
+      "step": 2471936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.52268562589633e-06,
+      "loss": 3.3421,
+      "step": 2472448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.51429967838581e-06,
+      "loss": 3.3351,
+      "step": 2472960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.505913730875291e-06,
+      "loss": 3.3388,
+      "step": 2473472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.497527783364771e-06,
+      "loss": 3.3356,
+      "step": 2473984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.48915821465798e-06,
+      "loss": 3.3355,
+      "step": 2474496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.480772267147462e-06,
+      "loss": 3.3272,
+      "step": 2475008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.472386319636942e-06,
+      "loss": 3.3517,
+      "step": 2475520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.46400037212642e-06,
+      "loss": 3.3352,
+      "step": 2476032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.455630803419633e-06,
+      "loss": 3.3324,
+      "step": 2476544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.447244855909113e-06,
+      "loss": 3.3317,
+      "step": 2477056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.438858908398593e-06,
+      "loss": 3.3389,
+      "step": 2477568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.430472960888072e-06,
+      "loss": 3.3469,
+      "step": 2478080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.422103392181283e-06,
+      "loss": 3.3446,
+      "step": 2478592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.413717444670763e-06,
+      "loss": 3.332,
+      "step": 2479104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.405331497160245e-06,
+      "loss": 3.3461,
+      "step": 2479616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.396961928453454e-06,
+      "loss": 3.3513,
+      "step": 2480128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.388575980942934e-06,
+      "loss": 3.3319,
+      "step": 2480640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.380190033432416e-06,
+      "loss": 3.35,
+      "step": 2481152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.371804085921894e-06,
+      "loss": 3.3423,
+      "step": 2481664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.363434517215106e-06,
+      "loss": 3.3394,
+      "step": 2482176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.355064948508317e-06,
+      "loss": 3.3364,
+      "step": 2482688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.346679000997797e-06,
+      "loss": 3.3413,
+      "step": 2483200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.338293053487277e-06,
+      "loss": 3.3479,
+      "step": 2483712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.329907105976757e-06,
+      "loss": 3.3528,
+      "step": 2484224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.321521158466237e-06,
+      "loss": 3.3319,
+      "step": 2484736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.313135210955717e-06,
+      "loss": 3.3149,
+      "step": 2485248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.304749263445197e-06,
+      "loss": 3.3368,
+      "step": 2485760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.296379694738408e-06,
+      "loss": 3.3406,
+      "step": 2486272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.287993747227888e-06,
+      "loss": 3.3435,
+      "step": 2486784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.279607799717368e-06,
+      "loss": 3.3339,
+      "step": 2487296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.271221852206847e-06,
+      "loss": 3.3433,
+      "step": 2487808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.262835904696327e-06,
+      "loss": 3.339,
+      "step": 2488320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.254449957185807e-06,
+      "loss": 3.3385,
+      "step": 2488832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.246064009675287e-06,
+      "loss": 3.3301,
+      "step": 2489344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.237678062164767e-06,
+      "loss": 3.3472,
+      "step": 2489856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.22930849345798e-06,
+      "loss": 3.3385,
+      "step": 2490368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.220922545947458e-06,
+      "loss": 3.3493,
+      "step": 2490880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.212536598436938e-06,
+      "loss": 3.3416,
+      "step": 2491392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.204150650926418e-06,
+      "loss": 3.3265,
+      "step": 2491904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.19578108221963e-06,
+      "loss": 3.3436,
+      "step": 2492416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.187395134709109e-06,
+      "loss": 3.3432,
+      "step": 2492928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.179009187198589e-06,
+      "loss": 3.3467,
+      "step": 2493440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.17062323968807e-06,
+      "loss": 3.3429,
+      "step": 2493952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.162253670981281e-06,
+      "loss": 3.3365,
+      "step": 2494464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.153867723470761e-06,
+      "loss": 3.3321,
+      "step": 2494976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.145498154763972e-06,
+      "loss": 3.3413,
+      "step": 2495488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.137112207253452e-06,
+      "loss": 3.3286,
+      "step": 2496000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.128726259742932e-06,
+      "loss": 3.337,
+      "step": 2496512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.120340312232412e-06,
+      "loss": 3.3418,
+      "step": 2497024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.111954364721892e-06,
+      "loss": 3.3398,
+      "step": 2497536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.103568417211372e-06,
+      "loss": 3.3278,
+      "step": 2498048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.095182469700852e-06,
+      "loss": 3.3392,
+      "step": 2498560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.086812900994062e-06,
+      "loss": 3.3342,
+      "step": 2499072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.078426953483542e-06,
+      "loss": 3.3445,
+      "step": 2499584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.070041005973022e-06,
+      "loss": 3.3386,
+      "step": 2500096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.061655058462502e-06,
+      "loss": 3.3357,
+      "step": 2500608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.053285489755715e-06,
+      "loss": 3.3517,
+      "step": 2501120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.044899542245195e-06,
+      "loss": 3.3514,
+      "step": 2501632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.036513594734673e-06,
+      "loss": 3.3384,
+      "step": 2502144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.028127647224153e-06,
+      "loss": 3.334,
+      "step": 2502656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.019758078517365e-06,
+      "loss": 3.3441,
+      "step": 2503168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.011372131006844e-06,
+      "loss": 3.3339,
+      "step": 2503680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.002986183496324e-06,
+      "loss": 3.3334,
+      "step": 2504192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.994600235985805e-06,
+      "loss": 3.3467,
+      "step": 2504704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.986230667279016e-06,
+      "loss": 3.3326,
+      "step": 2505216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.977844719768496e-06,
+      "loss": 3.34,
+      "step": 2505728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.969458772257976e-06,
+      "loss": 3.3287,
+      "step": 2506240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.961072824747456e-06,
+      "loss": 3.3372,
+      "step": 2506752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.952703256040668e-06,
+      "loss": 3.3416,
+      "step": 2507264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.944317308530147e-06,
+      "loss": 3.335,
+      "step": 2507776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.935931361019627e-06,
+      "loss": 3.3353,
+      "step": 2508288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.927545413509107e-06,
+      "loss": 3.3556,
+      "step": 2508800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.919175844802317e-06,
+      "loss": 3.3447,
+      "step": 2509312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.910789897291797e-06,
+      "loss": 3.3394,
+      "step": 2509824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.902403949781277e-06,
+      "loss": 3.3365,
+      "step": 2510336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.894018002270759e-06,
+      "loss": 3.339,
+      "step": 2510848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.8856648123677e-06,
+      "loss": 3.3408,
+      "step": 2511360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.87727886485718e-06,
+      "loss": 3.3388,
+      "step": 2511872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.86889291734666e-06,
+      "loss": 3.3354,
+      "step": 2512384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.86050696983614e-06,
+      "loss": 3.3494,
+      "step": 2512896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.85212102232562e-06,
+      "loss": 3.3452,
+      "step": 2513408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.8437350748151e-06,
+      "loss": 3.3461,
+      "step": 2513920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.83534912730458e-06,
+      "loss": 3.342,
+      "step": 2514432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.82696317979406e-06,
+      "loss": 3.3438,
+      "step": 2514944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.818593611087271e-06,
+      "loss": 3.3386,
+      "step": 2515456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.810207663576751e-06,
+      "loss": 3.3428,
+      "step": 2515968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.801821716066231e-06,
+      "loss": 3.3418,
+      "step": 2516480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.793435768555711e-06,
+      "loss": 3.3432,
+      "step": 2516992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.785066199848922e-06,
+      "loss": 3.3481,
+      "step": 2517504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.776680252338403e-06,
+      "loss": 3.3298,
+      "step": 2518016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.768310683631614e-06,
+      "loss": 3.3381,
+      "step": 2518528
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.872457265853882,
+      "eval_runtime": 340.2032,
+      "eval_samples_per_second": 1121.656,
+      "eval_steps_per_second": 35.053,
+      "step": 2518560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.759924736121092e-06,
+      "loss": 3.3298,
+      "step": 2519040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.751538788610574e-06,
+      "loss": 3.3355,
+      "step": 2519552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.743152841100054e-06,
+      "loss": 3.3538,
+      "step": 2520064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.734783272393265e-06,
+      "loss": 3.3464,
+      "step": 2520576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.726397324882745e-06,
+      "loss": 3.3451,
+      "step": 2521088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.718011377372225e-06,
+      "loss": 3.3389,
+      "step": 2521600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.709625429861705e-06,
+      "loss": 3.3366,
+      "step": 2522112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.701255861154915e-06,
+      "loss": 3.3272,
+      "step": 2522624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.692869913644395e-06,
+      "loss": 3.3447,
+      "step": 2523136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.684483966133875e-06,
+      "loss": 3.3416,
+      "step": 2523648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.676098018623355e-06,
+      "loss": 3.3403,
+      "step": 2524160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.667728449916566e-06,
+      "loss": 3.3445,
+      "step": 2524672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.659358881209779e-06,
+      "loss": 3.333,
+      "step": 2525184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.650972933699259e-06,
+      "loss": 3.3329,
+      "step": 2525696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.642586986188737e-06,
+      "loss": 3.3239,
+      "step": 2526208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.634201038678218e-06,
+      "loss": 3.3295,
+      "step": 2526720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.625815091167698e-06,
+      "loss": 3.3373,
+      "step": 2527232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.61744552246091e-06,
+      "loss": 3.334,
+      "step": 2527744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.609059574950389e-06,
+      "loss": 3.3373,
+      "step": 2528256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.600673627439869e-06,
+      "loss": 3.358,
+      "step": 2528768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.592287679929349e-06,
+      "loss": 3.3394,
+      "step": 2529280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.583918111222562e-06,
+      "loss": 3.3432,
+      "step": 2529792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.57553216371204e-06,
+      "loss": 3.3391,
+      "step": 2530304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.56714621620152e-06,
+      "loss": 3.3481,
+      "step": 2530816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.558760268691e-06,
+      "loss": 3.3301,
+      "step": 2531328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.55037432118048e-06,
+      "loss": 3.3381,
+      "step": 2531840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.54198837366996e-06,
+      "loss": 3.334,
+      "step": 2532352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.53360242615944e-06,
+      "loss": 3.3361,
+      "step": 2532864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.52521647864892e-06,
+      "loss": 3.3311,
+      "step": 2533376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.51684690994213e-06,
+      "loss": 3.3316,
+      "step": 2533888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.50846096243161e-06,
+      "loss": 3.3353,
+      "step": 2534400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.500075014921092e-06,
+      "loss": 3.3446,
+      "step": 2534912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.49168906741057e-06,
+      "loss": 3.3379,
+      "step": 2535424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.483319498703781e-06,
+      "loss": 3.342,
+      "step": 2535936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.474933551193263e-06,
+      "loss": 3.3328,
+      "step": 2536448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.466547603682743e-06,
+      "loss": 3.3446,
+      "step": 2536960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.458161656172221e-06,
+      "loss": 3.3251,
+      "step": 2537472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.449792087465433e-06,
+      "loss": 3.3397,
+      "step": 2537984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.441422518758644e-06,
+      "loss": 3.324,
+      "step": 2538496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.433036571248124e-06,
+      "loss": 3.338,
+      "step": 2539008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.424650623737604e-06,
+      "loss": 3.33,
+      "step": 2539520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.416264676227084e-06,
+      "loss": 3.3465,
+      "step": 2540032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.407878728716564e-06,
+      "loss": 3.3316,
+      "step": 2540544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.399492781206044e-06,
+      "loss": 3.3409,
+      "step": 2541056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.391106833695524e-06,
+      "loss": 3.3446,
+      "step": 2541568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.382720886185004e-06,
+      "loss": 3.3407,
+      "step": 2542080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.374351317478215e-06,
+      "loss": 3.3417,
+      "step": 2542592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.365965369967695e-06,
+      "loss": 3.3165,
+      "step": 2543104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.357579422457175e-06,
+      "loss": 3.3338,
+      "step": 2543616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.349193474946654e-06,
+      "loss": 3.3414,
+      "step": 2544128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.340823906239865e-06,
+      "loss": 3.3388,
+      "step": 2544640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.332437958729345e-06,
+      "loss": 3.3327,
+      "step": 2545152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.324052011218827e-06,
+      "loss": 3.3299,
+      "step": 2545664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.315666063708305e-06,
+      "loss": 3.3298,
+      "step": 2546176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.307296495001516e-06,
+      "loss": 3.3217,
+      "step": 2546688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.298910547490998e-06,
+      "loss": 3.3338,
+      "step": 2547200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.290524599980478e-06,
+      "loss": 3.3214,
+      "step": 2547712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.282138652469956e-06,
+      "loss": 3.3342,
+      "step": 2548224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.273769083763168e-06,
+      "loss": 3.3398,
+      "step": 2548736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.265383136252648e-06,
+      "loss": 3.3276,
+      "step": 2549248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.256997188742128e-06,
+      "loss": 3.3271,
+      "step": 2549760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.248611241231608e-06,
+      "loss": 3.3272,
+      "step": 2550272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.240241672524819e-06,
+      "loss": 3.3218,
+      "step": 2550784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.231855725014299e-06,
+      "loss": 3.3202,
+      "step": 2551296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.223486156307511e-06,
+      "loss": 3.3429,
+      "step": 2551808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.21510020879699e-06,
+      "loss": 3.3279,
+      "step": 2552320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.20671426128647e-06,
+      "loss": 3.3228,
+      "step": 2552832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.198328313775951e-06,
+      "loss": 3.3251,
+      "step": 2553344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.18994236626543e-06,
+      "loss": 3.3254,
+      "step": 2553856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.18155641875491e-06,
+      "loss": 3.3355,
+      "step": 2554368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.17317047124439e-06,
+      "loss": 3.3334,
+      "step": 2554880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.16478452373387e-06,
+      "loss": 3.3253,
+      "step": 2555392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.15641495502708e-06,
+      "loss": 3.3376,
+      "step": 2555904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.148029007516562e-06,
+      "loss": 3.3402,
+      "step": 2556416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.139643060006042e-06,
+      "loss": 3.3267,
+      "step": 2556928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.13125711249552e-06,
+      "loss": 3.339,
+      "step": 2557440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.122887543788733e-06,
+      "loss": 3.3337,
+      "step": 2557952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.114501596278212e-06,
+      "loss": 3.3306,
+      "step": 2558464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.106132027571423e-06,
+      "loss": 3.3276,
+      "step": 2558976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.097746080060903e-06,
+      "loss": 3.3305,
+      "step": 2559488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.089360132550383e-06,
+      "loss": 3.3395,
+      "step": 2560000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.080974185039863e-06,
+      "loss": 3.3367,
+      "step": 2560512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.072588237529343e-06,
+      "loss": 3.3278,
+      "step": 2561024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.064202290018823e-06,
+      "loss": 3.3071,
+      "step": 2561536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.055816342508303e-06,
+      "loss": 3.3243,
+      "step": 2562048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.047446773801515e-06,
+      "loss": 3.3353,
+      "step": 2562560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.039060826290994e-06,
+      "loss": 3.3339,
+      "step": 2563072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.030674878780474e-06,
+      "loss": 3.3293,
+      "step": 2563584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.022288931269954e-06,
+      "loss": 3.329,
+      "step": 2564096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.013902983759434e-06,
+      "loss": 3.3316,
+      "step": 2564608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.005517036248914e-06,
+      "loss": 3.3254,
+      "step": 2565120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.997131088738394e-06,
+      "loss": 3.3195,
+      "step": 2565632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.988745141227873e-06,
+      "loss": 3.3422,
+      "step": 2566144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.980375572521084e-06,
+      "loss": 3.3268,
+      "step": 2566656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.971989625010564e-06,
+      "loss": 3.3419,
+      "step": 2567168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.963603677500044e-06,
+      "loss": 3.3355,
+      "step": 2567680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.955217729989524e-06,
+      "loss": 3.3169,
+      "step": 2568192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.946848161282735e-06,
+      "loss": 3.3341,
+      "step": 2568704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.938462213772217e-06,
+      "loss": 3.3283,
+      "step": 2569216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.930076266261697e-06,
+      "loss": 3.3378,
+      "step": 2569728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.921690318751175e-06,
+      "loss": 3.3344,
+      "step": 2570240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.913320750044387e-06,
+      "loss": 3.3271,
+      "step": 2570752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.904934802533867e-06,
+      "loss": 3.322,
+      "step": 2571264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.896548855023347e-06,
+      "loss": 3.3357,
+      "step": 2571776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.888162907512827e-06,
+      "loss": 3.3183,
+      "step": 2572288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.879793338806038e-06,
+      "loss": 3.3273,
+      "step": 2572800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.871407391295518e-06,
+      "loss": 3.3344,
+      "step": 2573312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.863021443784998e-06,
+      "loss": 3.3278,
+      "step": 2573824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.854635496274478e-06,
+      "loss": 3.3169,
+      "step": 2574336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.846265927567689e-06,
+      "loss": 3.3332,
+      "step": 2574848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.83787998005717e-06,
+      "loss": 3.3244,
+      "step": 2575360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.829494032546649e-06,
+      "loss": 3.331,
+      "step": 2575872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.821108085036128e-06,
+      "loss": 3.3315,
+      "step": 2576384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.812738516329341e-06,
+      "loss": 3.3241,
+      "step": 2576896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.80435256881882e-06,
+      "loss": 3.3486,
+      "step": 2577408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.7959666213083e-06,
+      "loss": 3.3427,
+      "step": 2577920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.787580673797779e-06,
+      "loss": 3.3294,
+      "step": 2578432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.779211105090992e-06,
+      "loss": 3.3244,
+      "step": 2578944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.77082515758047e-06,
+      "loss": 3.3329,
+      "step": 2579456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.762439210069952e-06,
+      "loss": 3.3281,
+      "step": 2579968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.754053262559431e-06,
+      "loss": 3.3216,
+      "step": 2580480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.745683693852642e-06,
+      "loss": 3.3372,
+      "step": 2580992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.737297746342122e-06,
+      "loss": 3.3296,
+      "step": 2581504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.728911798831602e-06,
+      "loss": 3.3275,
+      "step": 2582016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.720525851321082e-06,
+      "loss": 3.3242,
+      "step": 2582528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.712156282614293e-06,
+      "loss": 3.3295,
+      "step": 2583040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.703770335103773e-06,
+      "loss": 3.3316,
+      "step": 2583552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.695384387593253e-06,
+      "loss": 3.3269,
+      "step": 2584064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.686998440082733e-06,
+      "loss": 3.3236,
+      "step": 2584576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.678628871375944e-06,
+      "loss": 3.3465,
+      "step": 2585088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.670259302669156e-06,
+      "loss": 3.3401,
+      "step": 2585600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.661873355158636e-06,
+      "loss": 3.33,
+      "step": 2586112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.653487407648116e-06,
+      "loss": 3.3271,
+      "step": 2586624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.645101460137594e-06,
+      "loss": 3.3246,
+      "step": 2587136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.636715512627076e-06,
+      "loss": 3.3358,
+      "step": 2587648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.628329565116556e-06,
+      "loss": 3.3293,
+      "step": 2588160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.619943617606035e-06,
+      "loss": 3.3225,
+      "step": 2588672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.611557670095515e-06,
+      "loss": 3.3435,
+      "step": 2589184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6031881013887265e-06,
+      "loss": 3.3329,
+      "step": 2589696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5948021538782065e-06,
+      "loss": 3.3401,
+      "step": 2590208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.586416206367686e-06,
+      "loss": 3.3299,
+      "step": 2590720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.578030258857166e-06,
+      "loss": 3.3386,
+      "step": 2591232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.569660690150378e-06,
+      "loss": 3.3319,
+      "step": 2591744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.561274742639856e-06,
+      "loss": 3.3314,
+      "step": 2592256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.552888795129337e-06,
+      "loss": 3.3343,
+      "step": 2592768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.544502847618817e-06,
+      "loss": 3.3331,
+      "step": 2593280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.536133278912029e-06,
+      "loss": 3.3376,
+      "step": 2593792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.527747331401508e-06,
+      "loss": 3.3217,
+      "step": 2594304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5193777626947194e-06,
+      "loss": 3.3276,
+      "step": 2594816
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.871983528137207,
+      "eval_runtime": 305.9241,
+      "eval_samples_per_second": 1247.339,
+      "eval_steps_per_second": 38.98,
+      "step": 2594880
+    }
+  ],
+  "logging_steps": 512,
+  "max_steps": 3052726,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 10,
+  "total_flos": 1.8062312068078387e+18,
+  "trial_name": null,
+  "trial_params": null
+}