diff --git "a/checkpoint-2213280/trainer_state.json" "b/checkpoint-2213280/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2213280/trainer_state.json"
@@ -0,0 +1,26189 @@
+{
+  "best_metric": 3.8418779373168945,
+  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-c-command/transformer/1/checkpoints/checkpoint-915840",
+  "epoch": 0.025000606015738065,
+  "eval_steps": 10,
+  "global_step": 2213280,
+  "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": 10.8637,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999161405248948e-05,
+      "loss": 6.8164,
+      "step": 512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998322810497896e-05,
+      "loss": 6.1844,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997484215746844e-05,
+      "loss": 5.9773,
+      "step": 1536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996645620995792e-05,
+      "loss": 5.8147,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99580702624474e-05,
+      "loss": 5.7076,
+      "step": 2560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994968431493688e-05,
+      "loss": 5.6115,
+      "step": 3072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994129836742636e-05,
+      "loss": 5.5486,
+      "step": 3584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993291241991584e-05,
+      "loss": 5.4577,
+      "step": 4096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992452647240532e-05,
+      "loss": 5.4271,
+      "step": 4608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99161405248948e-05,
+      "loss": 5.3684,
+      "step": 5120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.990775457738428e-05,
+      "loss": 5.329,
+      "step": 5632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989938500867749e-05,
+      "loss": 5.2875,
+      "step": 6144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989099906116697e-05,
+      "loss": 5.2217,
+      "step": 6656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.988261311365645e-05,
+      "loss": 5.2023,
+      "step": 7168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.987422716614593e-05,
+      "loss": 5.1555,
+      "step": 7680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986584121863541e-05,
+      "loss": 5.1313,
+      "step": 8192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.985745527112489e-05,
+      "loss": 5.1012,
+      "step": 8704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984906932361437e-05,
+      "loss": 5.0655,
+      "step": 9216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984068337610385e-05,
+      "loss": 5.0388,
+      "step": 9728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.983229742859333e-05,
+      "loss": 5.0252,
+      "step": 10240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9823927859886547e-05,
+      "loss": 4.9964,
+      "step": 10752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9815541912376026e-05,
+      "loss": 4.9649,
+      "step": 11264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9807155964865506e-05,
+      "loss": 4.9448,
+      "step": 11776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9798770017354986e-05,
+      "loss": 4.9356,
+      "step": 12288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9790400448648195e-05,
+      "loss": 4.9078,
+      "step": 12800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9782014501137675e-05,
+      "loss": 4.8793,
+      "step": 13312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9773628553627155e-05,
+      "loss": 4.8693,
+      "step": 13824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9765242606116635e-05,
+      "loss": 4.8488,
+      "step": 14336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9756873037409844e-05,
+      "loss": 4.8338,
+      "step": 14848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9748487089899324e-05,
+      "loss": 4.8189,
+      "step": 15360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9740101142388804e-05,
+      "loss": 4.8043,
+      "step": 15872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9731715194878284e-05,
+      "loss": 4.7923,
+      "step": 16384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97233456261715e-05,
+      "loss": 4.7735,
+      "step": 16896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.971495967866098e-05,
+      "loss": 4.7692,
+      "step": 17408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.970657373115046e-05,
+      "loss": 4.7678,
+      "step": 17920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.969820416244367e-05,
+      "loss": 4.7425,
+      "step": 18432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.968981821493315e-05,
+      "loss": 4.7043,
+      "step": 18944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.968143226742263e-05,
+      "loss": 4.71,
+      "step": 19456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.967304631991211e-05,
+      "loss": 4.6885,
+      "step": 19968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.966466037240159e-05,
+      "loss": 4.6937,
+      "step": 20480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965627442489107e-05,
+      "loss": 4.6801,
+      "step": 20992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964788847738054e-05,
+      "loss": 4.6734,
+      "step": 21504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963950252987002e-05,
+      "loss": 4.6523,
+      "step": 22016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963113296116324e-05,
+      "loss": 4.6601,
+      "step": 22528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962274701365272e-05,
+      "loss": 4.6428,
+      "step": 23040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96143610661422e-05,
+      "loss": 4.642,
+      "step": 23552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960597511863168e-05,
+      "loss": 4.6292,
+      "step": 24064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9597605549924894e-05,
+      "loss": 4.6076,
+      "step": 24576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95892359812181e-05,
+      "loss": 4.6082,
+      "step": 25088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958085003370758e-05,
+      "loss": 4.612,
+      "step": 25600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.957246408619706e-05,
+      "loss": 4.5888,
+      "step": 26112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.956407813868654e-05,
+      "loss": 4.5814,
+      "step": 26624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9555692191176016e-05,
+      "loss": 4.5729,
+      "step": 27136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.954732262246923e-05,
+      "loss": 4.57,
+      "step": 27648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953893667495871e-05,
+      "loss": 4.5537,
+      "step": 28160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953055072744819e-05,
+      "loss": 4.5734,
+      "step": 28672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.952216477993767e-05,
+      "loss": 4.5397,
+      "step": 29184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951377883242715e-05,
+      "loss": 4.5497,
+      "step": 29696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.950539288491663e-05,
+      "loss": 4.5409,
+      "step": 30208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.949700693740611e-05,
+      "loss": 4.5139,
+      "step": 30720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948862098989559e-05,
+      "loss": 4.5223,
+      "step": 31232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.94802514211888e-05,
+      "loss": 4.5149,
+      "step": 31744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947186547367828e-05,
+      "loss": 4.4991,
+      "step": 32256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.946347952616776e-05,
+      "loss": 4.4913,
+      "step": 32768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.945509357865724e-05,
+      "loss": 4.5083,
+      "step": 33280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.944672400995045e-05,
+      "loss": 4.4864,
+      "step": 33792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.943833806243993e-05,
+      "loss": 4.4805,
+      "step": 34304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.942995211492941e-05,
+      "loss": 4.4663,
+      "step": 34816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9421582546222625e-05,
+      "loss": 4.4811,
+      "step": 35328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9413196598712105e-05,
+      "loss": 4.4862,
+      "step": 35840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9404810651201585e-05,
+      "loss": 4.4702,
+      "step": 36352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9396424703691065e-05,
+      "loss": 4.4638,
+      "step": 36864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9388038756180545e-05,
+      "loss": 4.4686,
+      "step": 37376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9379652808670025e-05,
+      "loss": 4.4665,
+      "step": 37888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9371266861159505e-05,
+      "loss": 4.456,
+      "step": 38400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9362897292452714e-05,
+      "loss": 4.4387,
+      "step": 38912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9354511344942194e-05,
+      "loss": 4.4451,
+      "step": 39424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9346125397431674e-05,
+      "loss": 4.4327,
+      "step": 39936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9337739449921154e-05,
+      "loss": 4.4316,
+      "step": 40448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9329353502410634e-05,
+      "loss": 4.4213,
+      "step": 40960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9320967554900114e-05,
+      "loss": 4.4269,
+      "step": 41472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9312581607389594e-05,
+      "loss": 4.4305,
+      "step": 41984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9304195659879074e-05,
+      "loss": 4.4217,
+      "step": 42496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.929582609117229e-05,
+      "loss": 4.3963,
+      "step": 43008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.928744014366177e-05,
+      "loss": 4.4029,
+      "step": 43520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.927905419615125e-05,
+      "loss": 4.4073,
+      "step": 44032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.927066824864073e-05,
+      "loss": 4.4011,
+      "step": 44544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.926229867993394e-05,
+      "loss": 4.3969,
+      "step": 45056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.925392911122715e-05,
+      "loss": 4.3928,
+      "step": 45568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.924554316371663e-05,
+      "loss": 4.3881,
+      "step": 46080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.923715721620611e-05,
+      "loss": 4.377,
+      "step": 46592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922877126869559e-05,
+      "loss": 4.3804,
+      "step": 47104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922038532118507e-05,
+      "loss": 4.3807,
+      "step": 47616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921201575247828e-05,
+      "loss": 4.3718,
+      "step": 48128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.920362980496776e-05,
+      "loss": 4.3815,
+      "step": 48640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919524385745724e-05,
+      "loss": 4.3615,
+      "step": 49152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918685790994672e-05,
+      "loss": 4.3608,
+      "step": 49664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91784719624362e-05,
+      "loss": 4.3584,
+      "step": 50176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9170086014925676e-05,
+      "loss": 4.357,
+      "step": 50688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9161700067415156e-05,
+      "loss": 4.3616,
+      "step": 51200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915333049870837e-05,
+      "loss": 4.3448,
+      "step": 51712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914494455119785e-05,
+      "loss": 4.3419,
+      "step": 52224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9136558603687325e-05,
+      "loss": 4.3392,
+      "step": 52736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9128172656176805e-05,
+      "loss": 4.3431,
+      "step": 53248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911980308747002e-05,
+      "loss": 4.3245,
+      "step": 53760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911143351876323e-05,
+      "loss": 4.3162,
+      "step": 54272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.910304757125272e-05,
+      "loss": 4.3272,
+      "step": 54784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90946616237422e-05,
+      "loss": 4.3313,
+      "step": 55296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.908627567623168e-05,
+      "loss": 4.3235,
+      "step": 55808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.907788972872115e-05,
+      "loss": 4.3215,
+      "step": 56320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906950378121063e-05,
+      "loss": 4.3105,
+      "step": 56832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906111783370011e-05,
+      "loss": 4.3201,
+      "step": 57344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9052748264993326e-05,
+      "loss": 4.3046,
+      "step": 57856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90443623174828e-05,
+      "loss": 4.3122,
+      "step": 58368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.903597636997228e-05,
+      "loss": 4.3138,
+      "step": 58880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.902759042246176e-05,
+      "loss": 4.312,
+      "step": 59392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9019220853754975e-05,
+      "loss": 4.3003,
+      "step": 59904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9010834906244455e-05,
+      "loss": 4.2983,
+      "step": 60416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9002448958733935e-05,
+      "loss": 4.2997,
+      "step": 60928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8994063011223415e-05,
+      "loss": 4.2998,
+      "step": 61440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8985693442516624e-05,
+      "loss": 4.2894,
+      "step": 61952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8977307495006104e-05,
+      "loss": 4.2924,
+      "step": 62464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8968921547495584e-05,
+      "loss": 4.2883,
+      "step": 62976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8960535599985064e-05,
+      "loss": 4.2908,
+      "step": 63488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8952149652474544e-05,
+      "loss": 4.2743,
+      "step": 64000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8943763704964024e-05,
+      "loss": 4.272,
+      "step": 64512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.893539413625723e-05,
+      "loss": 4.2784,
+      "step": 65024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.892700818874671e-05,
+      "loss": 4.2784,
+      "step": 65536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891862224123619e-05,
+      "loss": 4.2694,
+      "step": 66048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891023629372567e-05,
+      "loss": 4.2811,
+      "step": 66560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.890185034621515e-05,
+      "loss": 4.2733,
+      "step": 67072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.889346439870464e-05,
+      "loss": 4.2594,
+      "step": 67584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.888507845119412e-05,
+      "loss": 4.2655,
+      "step": 68096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88766925036836e-05,
+      "loss": 4.2748,
+      "step": 68608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.886832293497681e-05,
+      "loss": 4.2643,
+      "step": 69120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.885993698746629e-05,
+      "loss": 4.2624,
+      "step": 69632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.885155103995577e-05,
+      "loss": 4.2525,
+      "step": 70144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.884316509244525e-05,
+      "loss": 4.26,
+      "step": 70656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.883479552373846e-05,
+      "loss": 4.2632,
+      "step": 71168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.882640957622794e-05,
+      "loss": 4.2492,
+      "step": 71680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.881802362871742e-05,
+      "loss": 4.2431,
+      "step": 72192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88096376812069e-05,
+      "loss": 4.2465,
+      "step": 72704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8801268112500106e-05,
+      "loss": 4.2398,
+      "step": 73216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879289854379332e-05,
+      "loss": 4.2382,
+      "step": 73728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.87845125962828e-05,
+      "loss": 4.246,
+      "step": 74240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.877612664877228e-05,
+      "loss": 4.2357,
+      "step": 74752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.876774070126176e-05,
+      "loss": 4.2377,
+      "step": 75264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875935475375124e-05,
+      "loss": 4.2326,
+      "step": 75776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875096880624072e-05,
+      "loss": 4.2293,
+      "step": 76288
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.198398113250732,
+      "eval_runtime": 305.2453,
+      "eval_samples_per_second": 1250.113,
+      "eval_steps_per_second": 39.067,
+      "step": 76320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.87425828587302e-05,
+      "loss": 4.218,
+      "step": 76800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.873419691121968e-05,
+      "loss": 4.2266,
+      "step": 77312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.872582734251289e-05,
+      "loss": 4.2332,
+      "step": 77824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.87174577738061e-05,
+      "loss": 4.2216,
+      "step": 78336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.870907182629558e-05,
+      "loss": 4.2224,
+      "step": 78848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.870068587878506e-05,
+      "loss": 4.2182,
+      "step": 79360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.869229993127454e-05,
+      "loss": 4.2157,
+      "step": 79872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8683913983764027e-05,
+      "loss": 4.2016,
+      "step": 80384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8675544415057236e-05,
+      "loss": 4.2262,
+      "step": 80896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8667158467546716e-05,
+      "loss": 4.2108,
+      "step": 81408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8658772520036196e-05,
+      "loss": 4.2184,
+      "step": 81920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8650386572525676e-05,
+      "loss": 4.2161,
+      "step": 82432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8642017003818885e-05,
+      "loss": 4.1978,
+      "step": 82944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8633631056308365e-05,
+      "loss": 4.1982,
+      "step": 83456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8625245108797845e-05,
+      "loss": 4.1945,
+      "step": 83968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8616859161287324e-05,
+      "loss": 4.1888,
+      "step": 84480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8608473213776804e-05,
+      "loss": 4.1935,
+      "step": 84992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.860008726626628e-05,
+      "loss": 4.1875,
+      "step": 85504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8591701318755764e-05,
+      "loss": 4.1823,
+      "step": 86016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8583315371245244e-05,
+      "loss": 4.2047,
+      "step": 86528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8574929423734724e-05,
+      "loss": 4.1899,
+      "step": 87040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8566543476224204e-05,
+      "loss": 4.1807,
+      "step": 87552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8558157528713684e-05,
+      "loss": 4.1777,
+      "step": 88064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.854980433881063e-05,
+      "loss": 4.1926,
+      "step": 88576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.854141839130011e-05,
+      "loss": 4.1686,
+      "step": 89088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.853303244378958e-05,
+      "loss": 4.167,
+      "step": 89600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.852464649627906e-05,
+      "loss": 4.1708,
+      "step": 90112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.851626054876854e-05,
+      "loss": 4.1704,
+      "step": 90624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.850787460125802e-05,
+      "loss": 4.1647,
+      "step": 91136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.84994886537475e-05,
+      "loss": 4.1648,
+      "step": 91648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.849110270623698e-05,
+      "loss": 4.1649,
+      "step": 92160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.84827331375302e-05,
+      "loss": 4.1709,
+      "step": 92672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.847434719001968e-05,
+      "loss": 4.1564,
+      "step": 93184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.846596124250916e-05,
+      "loss": 4.1656,
+      "step": 93696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.845757529499864e-05,
+      "loss": 4.1711,
+      "step": 94208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.844920572629185e-05,
+      "loss": 4.1655,
+      "step": 94720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.844081977878133e-05,
+      "loss": 4.1385,
+      "step": 95232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.843243383127081e-05,
+      "loss": 4.1453,
+      "step": 95744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.842404788376029e-05,
+      "loss": 4.1428,
+      "step": 96256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.841566193624977e-05,
+      "loss": 4.1501,
+      "step": 96768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.840727598873925e-05,
+      "loss": 4.1424,
+      "step": 97280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8398906420032456e-05,
+      "loss": 4.1491,
+      "step": 97792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8390520472521936e-05,
+      "loss": 4.1373,
+      "step": 98304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8382134525011416e-05,
+      "loss": 4.1463,
+      "step": 98816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.83737485775009e-05,
+      "loss": 4.1459,
+      "step": 99328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.836537900879411e-05,
+      "loss": 4.1451,
+      "step": 99840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.835699306128359e-05,
+      "loss": 4.1426,
+      "step": 100352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834860711377307e-05,
+      "loss": 4.1223,
+      "step": 100864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834022116626255e-05,
+      "loss": 4.1358,
+      "step": 101376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.833185159755576e-05,
+      "loss": 4.1407,
+      "step": 101888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832346565004524e-05,
+      "loss": 4.1328,
+      "step": 102400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831507970253472e-05,
+      "loss": 4.1292,
+      "step": 102912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.83066937550242e-05,
+      "loss": 4.1191,
+      "step": 103424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.829830780751368e-05,
+      "loss": 4.1276,
+      "step": 103936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.828993823880689e-05,
+      "loss": 4.1116,
+      "step": 104448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.828155229129637e-05,
+      "loss": 4.1394,
+      "step": 104960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8273166343785856e-05,
+      "loss": 4.1178,
+      "step": 105472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8264780396275336e-05,
+      "loss": 4.1294,
+      "step": 105984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8256410827568545e-05,
+      "loss": 4.1284,
+      "step": 106496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8248024880058025e-05,
+      "loss": 4.1023,
+      "step": 107008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8239638932547505e-05,
+      "loss": 4.1178,
+      "step": 107520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8231252985036985e-05,
+      "loss": 4.1141,
+      "step": 108032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8222883416330194e-05,
+      "loss": 4.1003,
+      "step": 108544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8214497468819674e-05,
+      "loss": 4.0972,
+      "step": 109056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8206111521309154e-05,
+      "loss": 4.1165,
+      "step": 109568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8197725573798634e-05,
+      "loss": 4.1041,
+      "step": 110080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.818935600509184e-05,
+      "loss": 4.1001,
+      "step": 110592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.818097005758132e-05,
+      "loss": 4.092,
+      "step": 111104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.817258411007081e-05,
+      "loss": 4.1108,
+      "step": 111616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.816419816256029e-05,
+      "loss": 4.1131,
+      "step": 112128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.81558285938535e-05,
+      "loss": 4.1088,
+      "step": 112640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.814744264634298e-05,
+      "loss": 4.1023,
+      "step": 113152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.813905669883246e-05,
+      "loss": 4.1129,
+      "step": 113664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.813067075132194e-05,
+      "loss": 4.109,
+      "step": 114176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.812230118261515e-05,
+      "loss": 4.1073,
+      "step": 114688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.811391523510463e-05,
+      "loss": 4.0909,
+      "step": 115200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.810552928759411e-05,
+      "loss": 4.103,
+      "step": 115712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.809714334008359e-05,
+      "loss": 4.0945,
+      "step": 116224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.80887737713768e-05,
+      "loss": 4.0946,
+      "step": 116736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.808038782386628e-05,
+      "loss": 4.09,
+      "step": 117248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.807200187635576e-05,
+      "loss": 4.0938,
+      "step": 117760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8063615928845243e-05,
+      "loss": 4.1048,
+      "step": 118272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.805524636013845e-05,
+      "loss": 4.0977,
+      "step": 118784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.804686041262793e-05,
+      "loss": 4.0757,
+      "step": 119296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.803847446511741e-05,
+      "loss": 4.082,
+      "step": 119808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8030088517606886e-05,
+      "loss": 4.0885,
+      "step": 120320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.80217189489001e-05,
+      "loss": 4.0862,
+      "step": 120832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.801333300138958e-05,
+      "loss": 4.0857,
+      "step": 121344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.800494705387906e-05,
+      "loss": 4.0825,
+      "step": 121856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7996561106368535e-05,
+      "loss": 4.0824,
+      "step": 122368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.798819153766175e-05,
+      "loss": 4.0711,
+      "step": 122880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.797980559015123e-05,
+      "loss": 4.0822,
+      "step": 123392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.797141964264071e-05,
+      "loss": 4.0791,
+      "step": 123904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.796303369513019e-05,
+      "loss": 4.0776,
+      "step": 124416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7954664126423406e-05,
+      "loss": 4.0862,
+      "step": 124928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7946278178912886e-05,
+      "loss": 4.0754,
+      "step": 125440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.793789223140236e-05,
+      "loss": 4.0687,
+      "step": 125952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.792950628389184e-05,
+      "loss": 4.0731,
+      "step": 126464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7921136715185055e-05,
+      "loss": 4.073,
+      "step": 126976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7912767146478264e-05,
+      "loss": 4.0778,
+      "step": 127488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7904381198967744e-05,
+      "loss": 4.067,
+      "step": 128000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7895995251457224e-05,
+      "loss": 4.0637,
+      "step": 128512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7887609303946704e-05,
+      "loss": 4.0627,
+      "step": 129024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7879223356436184e-05,
+      "loss": 4.0706,
+      "step": 129536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7870837408925664e-05,
+      "loss": 4.0539,
+      "step": 130048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.786246784021888e-05,
+      "loss": 4.0449,
+      "step": 130560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.785408189270836e-05,
+      "loss": 4.0614,
+      "step": 131072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.784569594519783e-05,
+      "loss": 4.0635,
+      "step": 131584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.783730999768731e-05,
+      "loss": 4.0554,
+      "step": 132096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.782892405017679e-05,
+      "loss": 4.0589,
+      "step": 132608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.782053810266627e-05,
+      "loss": 4.0529,
+      "step": 133120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.781215215515575e-05,
+      "loss": 4.0611,
+      "step": 133632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.780376620764523e-05,
+      "loss": 4.0475,
+      "step": 134144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.779539663893844e-05,
+      "loss": 4.0561,
+      "step": 134656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.778701069142792e-05,
+      "loss": 4.0623,
+      "step": 135168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.77786247439174e-05,
+      "loss": 4.0626,
+      "step": 135680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.777025517521062e-05,
+      "loss": 4.0499,
+      "step": 136192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.77618692277001e-05,
+      "loss": 4.0499,
+      "step": 136704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.775348328018958e-05,
+      "loss": 4.0524,
+      "step": 137216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.774509733267906e-05,
+      "loss": 4.0548,
+      "step": 137728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.773672776397227e-05,
+      "loss": 4.0425,
+      "step": 138240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7728341816461747e-05,
+      "loss": 4.0523,
+      "step": 138752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7719955868951227e-05,
+      "loss": 4.0434,
+      "step": 139264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7711569921440707e-05,
+      "loss": 4.0556,
+      "step": 139776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7703183973930186e-05,
+      "loss": 4.0358,
+      "step": 140288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7694798026419666e-05,
+      "loss": 4.0375,
+      "step": 140800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7686412078909146e-05,
+      "loss": 4.0416,
+      "step": 141312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7678026131398626e-05,
+      "loss": 4.0427,
+      "step": 141824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.766965656269184e-05,
+      "loss": 4.0355,
+      "step": 142336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.766127061518132e-05,
+      "loss": 4.0559,
+      "step": 142848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.76528846676708e-05,
+      "loss": 4.0433,
+      "step": 143360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.764449872016028e-05,
+      "loss": 4.0337,
+      "step": 143872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.763612915145349e-05,
+      "loss": 4.0362,
+      "step": 144384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.762774320394297e-05,
+      "loss": 4.0509,
+      "step": 144896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.761935725643245e-05,
+      "loss": 4.042,
+      "step": 145408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.761097130892193e-05,
+      "loss": 4.0385,
+      "step": 145920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.760260174021514e-05,
+      "loss": 4.0329,
+      "step": 146432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.759421579270462e-05,
+      "loss": 4.0403,
+      "step": 146944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.75858298451941e-05,
+      "loss": 4.0432,
+      "step": 147456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.757744389768358e-05,
+      "loss": 4.0332,
+      "step": 147968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7569074328976796e-05,
+      "loss": 4.0295,
+      "step": 148480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7560688381466276e-05,
+      "loss": 4.0319,
+      "step": 148992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7552302433955756e-05,
+      "loss": 4.0273,
+      "step": 149504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7543932865248965e-05,
+      "loss": 4.0249,
+      "step": 150016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7535546917738445e-05,
+      "loss": 4.0392,
+      "step": 150528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7527160970227925e-05,
+      "loss": 4.0263,
+      "step": 151040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7518775022717405e-05,
+      "loss": 4.0269,
+      "step": 151552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7510389075206885e-05,
+      "loss": 4.0257,
+      "step": 152064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7502003127696365e-05,
+      "loss": 4.0213,
+      "step": 152576
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.026689529418945,
+      "eval_runtime": 304.4444,
+      "eval_samples_per_second": 1253.401,
+      "eval_steps_per_second": 39.17,
+      "step": 152640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7493617180185845e-05,
+      "loss": 4.0118,
+      "step": 153088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.748523123267532e-05,
+      "loss": 4.0229,
+      "step": 153600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.74768452851648e-05,
+      "loss": 4.0333,
+      "step": 154112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.746845933765428e-05,
+      "loss": 4.0214,
+      "step": 154624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7460073390143764e-05,
+      "loss": 4.0257,
+      "step": 155136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7451687442633244e-05,
+      "loss": 4.0183,
+      "step": 155648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7443301495122724e-05,
+      "loss": 4.0197,
+      "step": 156160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7434915547612204e-05,
+      "loss": 4.0052,
+      "step": 156672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7426529600101684e-05,
+      "loss": 4.0283,
+      "step": 157184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7418143652591164e-05,
+      "loss": 4.0224,
+      "step": 157696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7409757705080644e-05,
+      "loss": 4.0187,
+      "step": 158208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.740138813637385e-05,
+      "loss": 4.0267,
+      "step": 158720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.739301856766706e-05,
+      "loss": 4.0057,
+      "step": 159232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.738463262015654e-05,
+      "loss": 4.0082,
+      "step": 159744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.737624667264602e-05,
+      "loss": 4.0129,
+      "step": 160256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.73678607251355e-05,
+      "loss": 3.9972,
+      "step": 160768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.735947477762498e-05,
+      "loss": 4.0057,
+      "step": 161280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.73511052089182e-05,
+      "loss": 4.0014,
+      "step": 161792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.734271926140768e-05,
+      "loss": 3.9991,
+      "step": 162304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.733433331389716e-05,
+      "loss": 4.0245,
+      "step": 162816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.732594736638664e-05,
+      "loss": 4.0045,
+      "step": 163328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.731756141887612e-05,
+      "loss": 4.0052,
+      "step": 163840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.730919185016933e-05,
+      "loss": 3.9961,
+      "step": 164352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.730080590265881e-05,
+      "loss": 4.0128,
+      "step": 164864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.729241995514829e-05,
+      "loss": 3.9865,
+      "step": 165376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.728403400763777e-05,
+      "loss": 3.9922,
+      "step": 165888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.727564806012725e-05,
+      "loss": 3.9918,
+      "step": 166400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.726726211261673e-05,
+      "loss": 3.9997,
+      "step": 166912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.725887616510621e-05,
+      "loss": 3.99,
+      "step": 167424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.725049021759568e-05,
+      "loss": 3.9915,
+      "step": 167936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7242120648888896e-05,
+      "loss": 3.9955,
+      "step": 168448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.723375108018211e-05,
+      "loss": 4.0019,
+      "step": 168960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.722536513267159e-05,
+      "loss": 3.982,
+      "step": 169472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.721697918516107e-05,
+      "loss": 3.9989,
+      "step": 169984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.720859323765055e-05,
+      "loss": 3.9977,
+      "step": 170496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.720020729014003e-05,
+      "loss": 3.9957,
+      "step": 171008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7191821342629505e-05,
+      "loss": 3.9736,
+      "step": 171520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7183435395118985e-05,
+      "loss": 3.9829,
+      "step": 172032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.71750658264122e-05,
+      "loss": 3.9725,
+      "step": 172544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.716667987890168e-05,
+      "loss": 3.9875,
+      "step": 173056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7158293931391154e-05,
+      "loss": 3.98,
+      "step": 173568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7149907983880634e-05,
+      "loss": 3.9889,
+      "step": 174080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.714153841517385e-05,
+      "loss": 3.9777,
+      "step": 174592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.713315246766333e-05,
+      "loss": 3.9804,
+      "step": 175104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.712476652015281e-05,
+      "loss": 3.9857,
+      "step": 175616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.711638057264229e-05,
+      "loss": 3.9881,
+      "step": 176128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.710799462513177e-05,
+      "loss": 3.9856,
+      "step": 176640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.709960867762125e-05,
+      "loss": 3.9631,
+      "step": 177152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.709122273011073e-05,
+      "loss": 3.9756,
+      "step": 177664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.708283678260021e-05,
+      "loss": 3.988,
+      "step": 178176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.707446721389342e-05,
+      "loss": 3.9717,
+      "step": 178688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.70660812663829e-05,
+      "loss": 3.9728,
+      "step": 179200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705769531887238e-05,
+      "loss": 3.9691,
+      "step": 179712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.704930937136186e-05,
+      "loss": 3.9713,
+      "step": 180224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.704092342385134e-05,
+      "loss": 3.9603,
+      "step": 180736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7032553855144554e-05,
+      "loss": 3.981,
+      "step": 181248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7024167907634034e-05,
+      "loss": 3.9698,
+      "step": 181760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7015781960123514e-05,
+      "loss": 3.9745,
+      "step": 182272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7007396012612994e-05,
+      "loss": 3.9841,
+      "step": 182784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6999010065102474e-05,
+      "loss": 3.9518,
+      "step": 183296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.699064049639568e-05,
+      "loss": 3.9692,
+      "step": 183808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.698225454888516e-05,
+      "loss": 3.9623,
+      "step": 184320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.697386860137464e-05,
+      "loss": 3.9571,
+      "step": 184832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.696548265386412e-05,
+      "loss": 3.9524,
+      "step": 185344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.695711308515733e-05,
+      "loss": 3.9662,
+      "step": 185856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.694872713764681e-05,
+      "loss": 3.9631,
+      "step": 186368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.694034119013629e-05,
+      "loss": 3.9532,
+      "step": 186880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.693195524262577e-05,
+      "loss": 3.9468,
+      "step": 187392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.692356929511526e-05,
+      "loss": 3.9669,
+      "step": 187904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.691518334760474e-05,
+      "loss": 3.9744,
+      "step": 188416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.690679740009422e-05,
+      "loss": 3.9636,
+      "step": 188928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.689842783138743e-05,
+      "loss": 3.9632,
+      "step": 189440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.689004188387691e-05,
+      "loss": 3.9717,
+      "step": 189952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.688165593636639e-05,
+      "loss": 3.967,
+      "step": 190464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.687326998885587e-05,
+      "loss": 3.9694,
+      "step": 190976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.686488404134534e-05,
+      "loss": 3.9475,
+      "step": 191488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6856514472638556e-05,
+      "loss": 3.9653,
+      "step": 192000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6848128525128036e-05,
+      "loss": 3.9607,
+      "step": 192512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.683974257761751e-05,
+      "loss": 3.9538,
+      "step": 193024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6831356630106996e-05,
+      "loss": 3.9583,
+      "step": 193536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6822970682596476e-05,
+      "loss": 3.9576,
+      "step": 194048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.681460111388969e-05,
+      "loss": 3.9703,
+      "step": 194560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6806215166379165e-05,
+      "loss": 3.9591,
+      "step": 195072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6797829218868645e-05,
+      "loss": 3.9404,
+      "step": 195584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6789443271358125e-05,
+      "loss": 3.9456,
+      "step": 196096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.678107370265134e-05,
+      "loss": 3.9527,
+      "step": 196608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6772687755140814e-05,
+      "loss": 3.9554,
+      "step": 197120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6764301807630294e-05,
+      "loss": 3.954,
+      "step": 197632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6755915860119774e-05,
+      "loss": 3.9474,
+      "step": 198144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.674754629141298e-05,
+      "loss": 3.9507,
+      "step": 198656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.673916034390246e-05,
+      "loss": 3.9396,
+      "step": 199168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.673077439639194e-05,
+      "loss": 3.953,
+      "step": 199680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.672238844888143e-05,
+      "loss": 3.9449,
+      "step": 200192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.671401888017464e-05,
+      "loss": 3.9489,
+      "step": 200704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.670563293266412e-05,
+      "loss": 3.9583,
+      "step": 201216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.66972469851536e-05,
+      "loss": 3.9486,
+      "step": 201728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.668886103764308e-05,
+      "loss": 3.9414,
+      "step": 202240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.668049146893629e-05,
+      "loss": 3.9468,
+      "step": 202752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.667210552142577e-05,
+      "loss": 3.9463,
+      "step": 203264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.666371957391525e-05,
+      "loss": 3.9509,
+      "step": 203776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.665533362640473e-05,
+      "loss": 3.9422,
+      "step": 204288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.664694767889421e-05,
+      "loss": 3.9397,
+      "step": 204800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.663857811018742e-05,
+      "loss": 3.9335,
+      "step": 205312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.66301921626769e-05,
+      "loss": 3.9463,
+      "step": 205824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6621806215166383e-05,
+      "loss": 3.933,
+      "step": 206336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6613420267655863e-05,
+      "loss": 3.9223,
+      "step": 206848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.660505069894907e-05,
+      "loss": 3.9355,
+      "step": 207360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.659666475143855e-05,
+      "loss": 3.9425,
+      "step": 207872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.658827880392803e-05,
+      "loss": 3.9292,
+      "step": 208384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657989285641751e-05,
+      "loss": 3.9407,
+      "step": 208896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657152328771072e-05,
+      "loss": 3.9287,
+      "step": 209408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.65631373402002e-05,
+      "loss": 3.9424,
+      "step": 209920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.655475139268968e-05,
+      "loss": 3.9278,
+      "step": 210432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.654636544517916e-05,
+      "loss": 3.9318,
+      "step": 210944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.653797949766864e-05,
+      "loss": 3.9452,
+      "step": 211456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.652960992896185e-05,
+      "loss": 3.9415,
+      "step": 211968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.652122398145134e-05,
+      "loss": 3.9323,
+      "step": 212480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.651283803394082e-05,
+      "loss": 3.9332,
+      "step": 212992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.65044520864303e-05,
+      "loss": 3.9299,
+      "step": 213504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6496082517723506e-05,
+      "loss": 3.9361,
+      "step": 214016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6487696570212986e-05,
+      "loss": 3.9252,
+      "step": 214528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6479310622702466e-05,
+      "loss": 3.9377,
+      "step": 215040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6470924675191946e-05,
+      "loss": 3.9271,
+      "step": 215552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6462538727681426e-05,
+      "loss": 3.9382,
+      "step": 216064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6454169158974635e-05,
+      "loss": 3.9179,
+      "step": 216576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6445783211464115e-05,
+      "loss": 3.9267,
+      "step": 217088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6437397263953595e-05,
+      "loss": 3.923,
+      "step": 217600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6429011316443075e-05,
+      "loss": 3.9274,
+      "step": 218112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642064174773629e-05,
+      "loss": 3.9217,
+      "step": 218624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.641225580022577e-05,
+      "loss": 3.9438,
+      "step": 219136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.640386985271525e-05,
+      "loss": 3.9293,
+      "step": 219648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.639548390520473e-05,
+      "loss": 3.9235,
+      "step": 220160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.638709795769421e-05,
+      "loss": 3.9167,
+      "step": 220672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.637872838898742e-05,
+      "loss": 3.9365,
+      "step": 221184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.63703424414769e-05,
+      "loss": 3.9306,
+      "step": 221696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.636195649396638e-05,
+      "loss": 3.9296,
+      "step": 222208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.635357054645586e-05,
+      "loss": 3.9221,
+      "step": 222720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.634520097774907e-05,
+      "loss": 3.9258,
+      "step": 223232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.633681503023855e-05,
+      "loss": 3.9321,
+      "step": 223744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.632842908272803e-05,
+      "loss": 3.9228,
+      "step": 224256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.632004313521751e-05,
+      "loss": 3.9198,
+      "step": 224768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.631165718770699e-05,
+      "loss": 3.925,
+      "step": 225280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6303287619000204e-05,
+      "loss": 3.9191,
+      "step": 225792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6294901671489684e-05,
+      "loss": 3.9131,
+      "step": 226304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6286515723979164e-05,
+      "loss": 3.9263,
+      "step": 226816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6278129776468644e-05,
+      "loss": 3.9247,
+      "step": 227328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626976020776185e-05,
+      "loss": 3.9173,
+      "step": 227840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626137426025133e-05,
+      "loss": 3.9149,
+      "step": 228352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.625298831274081e-05,
+      "loss": 3.9115,
+      "step": 228864
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.951958656311035,
+      "eval_runtime": 301.9713,
+      "eval_samples_per_second": 1263.667,
+      "eval_steps_per_second": 39.491,
+      "step": 228960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.624460236523029e-05,
+      "loss": 3.9139,
+      "step": 229376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.62362327965235e-05,
+      "loss": 3.9154,
+      "step": 229888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.622784684901298e-05,
+      "loss": 3.9272,
+      "step": 230400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.621946090150246e-05,
+      "loss": 3.9137,
+      "step": 230912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.621107495399194e-05,
+      "loss": 3.9232,
+      "step": 231424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.620270538528516e-05,
+      "loss": 3.9088,
+      "step": 231936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.619431943777464e-05,
+      "loss": 3.9156,
+      "step": 232448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.618593349026412e-05,
+      "loss": 3.9005,
+      "step": 232960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.617754754275359e-05,
+      "loss": 3.9243,
+      "step": 233472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.616917797404681e-05,
+      "loss": 3.9183,
+      "step": 233984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.616079202653629e-05,
+      "loss": 3.9164,
+      "step": 234496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.615240607902577e-05,
+      "loss": 3.9233,
+      "step": 235008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6144036510318976e-05,
+      "loss": 3.8999,
+      "step": 235520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6135650562808456e-05,
+      "loss": 3.9068,
+      "step": 236032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6127264615297936e-05,
+      "loss": 3.912,
+      "step": 236544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6118878667787416e-05,
+      "loss": 3.9005,
+      "step": 237056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6110492720276896e-05,
+      "loss": 3.9008,
+      "step": 237568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6102106772766376e-05,
+      "loss": 3.8993,
+      "step": 238080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6093720825255856e-05,
+      "loss": 3.8998,
+      "step": 238592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6085334877745336e-05,
+      "loss": 3.9255,
+      "step": 239104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6076965309038545e-05,
+      "loss": 3.903,
+      "step": 239616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6068579361528025e-05,
+      "loss": 3.9043,
+      "step": 240128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6060193414017505e-05,
+      "loss": 3.8992,
+      "step": 240640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6051807466506985e-05,
+      "loss": 3.9089,
+      "step": 241152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6043437897800194e-05,
+      "loss": 3.8909,
+      "step": 241664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6035051950289674e-05,
+      "loss": 3.8929,
+      "step": 242176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6026666002779154e-05,
+      "loss": 3.8952,
+      "step": 242688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6018280055268634e-05,
+      "loss": 3.9025,
+      "step": 243200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.600991048656185e-05,
+      "loss": 3.8889,
+      "step": 243712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.600152453905133e-05,
+      "loss": 3.8969,
+      "step": 244224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.599313859154081e-05,
+      "loss": 3.8949,
+      "step": 244736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.598475264403029e-05,
+      "loss": 3.9053,
+      "step": 245248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.59763830753235e-05,
+      "loss": 3.8923,
+      "step": 245760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.596799712781298e-05,
+      "loss": 3.8976,
+      "step": 246272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.595961118030246e-05,
+      "loss": 3.8993,
+      "step": 246784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.595122523279194e-05,
+      "loss": 3.8989,
+      "step": 247296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.594285566408515e-05,
+      "loss": 3.8809,
+      "step": 247808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.593446971657463e-05,
+      "loss": 3.8912,
+      "step": 248320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.592608376906411e-05,
+      "loss": 3.8772,
+      "step": 248832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.591769782155359e-05,
+      "loss": 3.8946,
+      "step": 249344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.59093282528468e-05,
+      "loss": 3.8821,
+      "step": 249856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.590094230533628e-05,
+      "loss": 3.896,
+      "step": 250368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.589255635782576e-05,
+      "loss": 3.8884,
+      "step": 250880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.588417041031524e-05,
+      "loss": 3.8833,
+      "step": 251392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.587580084160845e-05,
+      "loss": 3.8952,
+      "step": 251904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.586741489409793e-05,
+      "loss": 3.8929,
+      "step": 252416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.585902894658741e-05,
+      "loss": 3.8925,
+      "step": 252928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.585064299907689e-05,
+      "loss": 3.8763,
+      "step": 253440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.58422734303701e-05,
+      "loss": 3.8798,
+      "step": 253952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.583388748285958e-05,
+      "loss": 3.8968,
+      "step": 254464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.582550153534906e-05,
+      "loss": 3.8827,
+      "step": 254976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.581713196664228e-05,
+      "loss": 3.8782,
+      "step": 255488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.580874601913176e-05,
+      "loss": 3.8807,
+      "step": 256000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.580036007162124e-05,
+      "loss": 3.885,
+      "step": 256512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.579197412411072e-05,
+      "loss": 3.8687,
+      "step": 257024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.57835881766002e-05,
+      "loss": 3.889,
+      "step": 257536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.577520222908968e-05,
+      "loss": 3.8809,
+      "step": 258048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.576681628157916e-05,
+      "loss": 3.8853,
+      "step": 258560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5758430334068637e-05,
+      "loss": 3.8918,
+      "step": 259072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5750060765361846e-05,
+      "loss": 3.8696,
+      "step": 259584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5741674817851326e-05,
+      "loss": 3.8744,
+      "step": 260096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5733288870340806e-05,
+      "loss": 3.8794,
+      "step": 260608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5724902922830286e-05,
+      "loss": 3.8666,
+      "step": 261120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5716533354123495e-05,
+      "loss": 3.8657,
+      "step": 261632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5708147406612975e-05,
+      "loss": 3.8766,
+      "step": 262144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.569976145910246e-05,
+      "loss": 3.8778,
+      "step": 262656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.569137551159194e-05,
+      "loss": 3.8645,
+      "step": 263168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.568300594288515e-05,
+      "loss": 3.8614,
+      "step": 263680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.567461999537463e-05,
+      "loss": 3.8784,
+      "step": 264192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.566623404786411e-05,
+      "loss": 3.8859,
+      "step": 264704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.565784810035359e-05,
+      "loss": 3.8745,
+      "step": 265216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.56494785316468e-05,
+      "loss": 3.8808,
+      "step": 265728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.564109258413628e-05,
+      "loss": 3.8847,
+      "step": 266240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.563270663662576e-05,
+      "loss": 3.8823,
+      "step": 266752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.562432068911524e-05,
+      "loss": 3.8864,
+      "step": 267264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.561595112040845e-05,
+      "loss": 3.861,
+      "step": 267776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.560756517289793e-05,
+      "loss": 3.8802,
+      "step": 268288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5599195604191144e-05,
+      "loss": 3.8778,
+      "step": 268800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5590809656680624e-05,
+      "loss": 3.867,
+      "step": 269312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5582423709170104e-05,
+      "loss": 3.8785,
+      "step": 269824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5574037761659584e-05,
+      "loss": 3.8715,
+      "step": 270336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5565651814149064e-05,
+      "loss": 3.8834,
+      "step": 270848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5557265866638544e-05,
+      "loss": 3.8769,
+      "step": 271360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5548879919128024e-05,
+      "loss": 3.8553,
+      "step": 271872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.55404939716175e-05,
+      "loss": 3.8649,
+      "step": 272384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.553212440291071e-05,
+      "loss": 3.8657,
+      "step": 272896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.552373845540019e-05,
+      "loss": 3.8705,
+      "step": 273408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5515352507889666e-05,
+      "loss": 3.8722,
+      "step": 273920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.550696656037915e-05,
+      "loss": 3.8643,
+      "step": 274432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.549859699167237e-05,
+      "loss": 3.8672,
+      "step": 274944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.549021104416185e-05,
+      "loss": 3.8595,
+      "step": 275456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.548182509665132e-05,
+      "loss": 3.869,
+      "step": 275968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.54734391491408e-05,
+      "loss": 3.861,
+      "step": 276480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.546506958043402e-05,
+      "loss": 3.869,
+      "step": 276992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.54566836329235e-05,
+      "loss": 3.8785,
+      "step": 277504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.544829768541297e-05,
+      "loss": 3.8674,
+      "step": 278016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.543991173790245e-05,
+      "loss": 3.8599,
+      "step": 278528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5431542169195667e-05,
+      "loss": 3.8663,
+      "step": 279040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.542315622168514e-05,
+      "loss": 3.8643,
+      "step": 279552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.541477027417462e-05,
+      "loss": 3.8713,
+      "step": 280064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5406384326664106e-05,
+      "loss": 3.8639,
+      "step": 280576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.539801475795732e-05,
+      "loss": 3.8611,
+      "step": 281088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5389628810446796e-05,
+      "loss": 3.8512,
+      "step": 281600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5381242862936275e-05,
+      "loss": 3.8698,
+      "step": 282112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5372856915425755e-05,
+      "loss": 3.849,
+      "step": 282624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.536448734671897e-05,
+      "loss": 3.8463,
+      "step": 283136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5356101399208444e-05,
+      "loss": 3.8543,
+      "step": 283648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5347715451697924e-05,
+      "loss": 3.8618,
+      "step": 284160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5339329504187404e-05,
+      "loss": 3.8533,
+      "step": 284672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5330959935480613e-05,
+      "loss": 3.8672,
+      "step": 285184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5322573987970093e-05,
+      "loss": 3.8518,
+      "step": 285696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5314188040459573e-05,
+      "loss": 3.8558,
+      "step": 286208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.530580209294906e-05,
+      "loss": 3.8561,
+      "step": 286720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.529743252424227e-05,
+      "loss": 3.8491,
+      "step": 287232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.528904657673175e-05,
+      "loss": 3.8699,
+      "step": 287744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.528066062922123e-05,
+      "loss": 3.8627,
+      "step": 288256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.527227468171071e-05,
+      "loss": 3.8574,
+      "step": 288768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.526390511300392e-05,
+      "loss": 3.8575,
+      "step": 289280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.52555191654934e-05,
+      "loss": 3.8475,
+      "step": 289792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.524713321798288e-05,
+      "loss": 3.8605,
+      "step": 290304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.523874727047236e-05,
+      "loss": 3.8498,
+      "step": 290816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.523037770176557e-05,
+      "loss": 3.8644,
+      "step": 291328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.522199175425505e-05,
+      "loss": 3.8496,
+      "step": 291840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.521360580674453e-05,
+      "loss": 3.8566,
+      "step": 292352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.520521985923401e-05,
+      "loss": 3.8464,
+      "step": 292864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.519685029052722e-05,
+      "loss": 3.8509,
+      "step": 293376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.51884643430167e-05,
+      "loss": 3.8463,
+      "step": 293888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.518007839550618e-05,
+      "loss": 3.8513,
+      "step": 294400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.517169244799566e-05,
+      "loss": 3.8454,
+      "step": 294912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.516332287928887e-05,
+      "loss": 3.8685,
+      "step": 295424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.515493693177835e-05,
+      "loss": 3.8569,
+      "step": 295936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.514655098426783e-05,
+      "loss": 3.8509,
+      "step": 296448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.513816503675731e-05,
+      "loss": 3.8387,
+      "step": 296960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.512979546805052e-05,
+      "loss": 3.8613,
+      "step": 297472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.512140952054e-05,
+      "loss": 3.855,
+      "step": 297984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.511302357302948e-05,
+      "loss": 3.8583,
+      "step": 298496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.510463762551896e-05,
+      "loss": 3.8476,
+      "step": 299008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5096268056812177e-05,
+      "loss": 3.8492,
+      "step": 299520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5087882109301657e-05,
+      "loss": 3.8569,
+      "step": 300032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5079496161791136e-05,
+      "loss": 3.8529,
+      "step": 300544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5071110214280616e-05,
+      "loss": 3.8428,
+      "step": 301056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5062740645573826e-05,
+      "loss": 3.8515,
+      "step": 301568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5054354698063305e-05,
+      "loss": 3.8449,
+      "step": 302080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5045968750552785e-05,
+      "loss": 3.8398,
+      "step": 302592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5037599181845995e-05,
+      "loss": 3.854,
+      "step": 303104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5029213234335474e-05,
+      "loss": 3.8541,
+      "step": 303616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5020827286824954e-05,
+      "loss": 3.8435,
+      "step": 304128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5012441339314434e-05,
+      "loss": 3.8389,
+      "step": 304640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5004055391803914e-05,
+      "loss": 3.8476,
+      "step": 305152
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9105594158172607,
+      "eval_runtime": 306.2905,
+      "eval_samples_per_second": 1245.847,
+      "eval_steps_per_second": 38.934,
+      "step": 305280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.49956694442934e-05,
+      "loss": 3.8364,
+      "step": 305664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.498728349678288e-05,
+      "loss": 3.8427,
+      "step": 306176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.497889754927236e-05,
+      "loss": 3.8544,
+      "step": 306688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.497051160176184e-05,
+      "loss": 3.8422,
+      "step": 307200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.496212565425132e-05,
+      "loss": 3.8551,
+      "step": 307712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.49537397067408e-05,
+      "loss": 3.8387,
+      "step": 308224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4945353759230274e-05,
+      "loss": 3.8431,
+      "step": 308736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4936967811719754e-05,
+      "loss": 3.8303,
+      "step": 309248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.492859824301297e-05,
+      "loss": 3.8562,
+      "step": 309760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.492021229550245e-05,
+      "loss": 3.8459,
+      "step": 310272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.491182634799192e-05,
+      "loss": 3.8417,
+      "step": 310784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.49034404004814e-05,
+      "loss": 3.8546,
+      "step": 311296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4895087210578355e-05,
+      "loss": 3.8279,
+      "step": 311808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4886701263067835e-05,
+      "loss": 3.8415,
+      "step": 312320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4878315315557315e-05,
+      "loss": 3.8412,
+      "step": 312832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4869929368046795e-05,
+      "loss": 3.8308,
+      "step": 313344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4861543420536275e-05,
+      "loss": 3.832,
+      "step": 313856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.485315747302575e-05,
+      "loss": 3.8316,
+      "step": 314368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.484477152551523e-05,
+      "loss": 3.8341,
+      "step": 314880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4836401956808444e-05,
+      "loss": 3.852,
+      "step": 315392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4828016009297924e-05,
+      "loss": 3.8357,
+      "step": 315904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.48196300617874e-05,
+      "loss": 3.8365,
+      "step": 316416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.481124411427688e-05,
+      "loss": 3.8319,
+      "step": 316928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.480285816676636e-05,
+      "loss": 3.8404,
+      "step": 317440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4794472219255837e-05,
+      "loss": 3.8245,
+      "step": 317952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.478608627174532e-05,
+      "loss": 3.8257,
+      "step": 318464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.47777003242348e-05,
+      "loss": 3.8262,
+      "step": 318976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.476933075552801e-05,
+      "loss": 3.8355,
+      "step": 319488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.476094480801749e-05,
+      "loss": 3.8209,
+      "step": 320000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.475255886050697e-05,
+      "loss": 3.8298,
+      "step": 320512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.474417291299645e-05,
+      "loss": 3.8305,
+      "step": 321024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.473580334428966e-05,
+      "loss": 3.8375,
+      "step": 321536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.472741739677914e-05,
+      "loss": 3.8249,
+      "step": 322048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.471903144926862e-05,
+      "loss": 3.832,
+      "step": 322560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.471066188056183e-05,
+      "loss": 3.8323,
+      "step": 323072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.470227593305131e-05,
+      "loss": 3.833,
+      "step": 323584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.469388998554079e-05,
+      "loss": 3.8196,
+      "step": 324096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.468550403803028e-05,
+      "loss": 3.8215,
+      "step": 324608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4677134469323486e-05,
+      "loss": 3.8131,
+      "step": 325120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4668748521812966e-05,
+      "loss": 3.8264,
+      "step": 325632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4660362574302446e-05,
+      "loss": 3.8169,
+      "step": 326144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4651976626791926e-05,
+      "loss": 3.8304,
+      "step": 326656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4643607058085135e-05,
+      "loss": 3.8214,
+      "step": 327168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4635221110574615e-05,
+      "loss": 3.8223,
+      "step": 327680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4626835163064095e-05,
+      "loss": 3.8281,
+      "step": 328192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4618449215553575e-05,
+      "loss": 3.8261,
+      "step": 328704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4610079646846784e-05,
+      "loss": 3.8289,
+      "step": 329216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4601693699336264e-05,
+      "loss": 3.8162,
+      "step": 329728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4593307751825744e-05,
+      "loss": 3.81,
+      "step": 330240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4584921804315224e-05,
+      "loss": 3.833,
+      "step": 330752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.457655223560844e-05,
+      "loss": 3.8184,
+      "step": 331264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.456818266690165e-05,
+      "loss": 3.8163,
+      "step": 331776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455979671939113e-05,
+      "loss": 3.8161,
+      "step": 332288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455141077188061e-05,
+      "loss": 3.8227,
+      "step": 332800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.454302482437009e-05,
+      "loss": 3.8036,
+      "step": 333312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.453463887685957e-05,
+      "loss": 3.8264,
+      "step": 333824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.452625292934905e-05,
+      "loss": 3.8189,
+      "step": 334336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.451786698183853e-05,
+      "loss": 3.8202,
+      "step": 334848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.450948103432801e-05,
+      "loss": 3.8301,
+      "step": 335360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.450111146562122e-05,
+      "loss": 3.8113,
+      "step": 335872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.44927255181107e-05,
+      "loss": 3.8089,
+      "step": 336384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.448433957060018e-05,
+      "loss": 3.8212,
+      "step": 336896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4475953623089664e-05,
+      "loss": 3.801,
+      "step": 337408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.446758405438287e-05,
+      "loss": 3.8026,
+      "step": 337920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.445919810687235e-05,
+      "loss": 3.8152,
+      "step": 338432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.445081215936183e-05,
+      "loss": 3.8161,
+      "step": 338944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444242621185131e-05,
+      "loss": 3.8026,
+      "step": 339456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.443405664314452e-05,
+      "loss": 3.801,
+      "step": 339968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4425670695634e-05,
+      "loss": 3.8137,
+      "step": 340480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.441728474812348e-05,
+      "loss": 3.822,
+      "step": 340992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.440889880061296e-05,
+      "loss": 3.8166,
+      "step": 341504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.440052923190617e-05,
+      "loss": 3.8158,
+      "step": 342016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.439214328439565e-05,
+      "loss": 3.8246,
+      "step": 342528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.438375733688513e-05,
+      "loss": 3.8181,
+      "step": 343040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.437537138937462e-05,
+      "loss": 3.8292,
+      "step": 343552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.436700182066783e-05,
+      "loss": 3.7989,
+      "step": 344064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.435861587315731e-05,
+      "loss": 3.8186,
+      "step": 344576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4350246304450516e-05,
+      "loss": 3.8174,
+      "step": 345088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4341860356939996e-05,
+      "loss": 3.8048,
+      "step": 345600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4333474409429476e-05,
+      "loss": 3.8214,
+      "step": 346112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4325088461918956e-05,
+      "loss": 3.8112,
+      "step": 346624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4316702514408436e-05,
+      "loss": 3.8204,
+      "step": 347136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4308316566897916e-05,
+      "loss": 3.8197,
+      "step": 347648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4299930619387396e-05,
+      "loss": 3.7965,
+      "step": 348160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4291544671876876e-05,
+      "loss": 3.8057,
+      "step": 348672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4283175103170085e-05,
+      "loss": 3.8053,
+      "step": 349184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.427478915565957e-05,
+      "loss": 3.8139,
+      "step": 349696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.426640320814905e-05,
+      "loss": 3.8108,
+      "step": 350208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.425801726063853e-05,
+      "loss": 3.8034,
+      "step": 350720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.424966407073547e-05,
+      "loss": 3.8109,
+      "step": 351232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.424127812322495e-05,
+      "loss": 3.8015,
+      "step": 351744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.423289217571443e-05,
+      "loss": 3.806,
+      "step": 352256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.422450622820391e-05,
+      "loss": 3.8035,
+      "step": 352768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.421612028069339e-05,
+      "loss": 3.8124,
+      "step": 353280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.420773433318287e-05,
+      "loss": 3.8178,
+      "step": 353792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.419934838567235e-05,
+      "loss": 3.8082,
+      "step": 354304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.419096243816182e-05,
+      "loss": 3.7994,
+      "step": 354816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.418259286945504e-05,
+      "loss": 3.806,
+      "step": 355328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4174206921944525e-05,
+      "loss": 3.808,
+      "step": 355840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4165820974434005e-05,
+      "loss": 3.8128,
+      "step": 356352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.415743502692348e-05,
+      "loss": 3.8082,
+      "step": 356864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4149065458216694e-05,
+      "loss": 3.8014,
+      "step": 357376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4140679510706174e-05,
+      "loss": 3.7951,
+      "step": 357888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4132293563195654e-05,
+      "loss": 3.8083,
+      "step": 358400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.412390761568513e-05,
+      "loss": 3.7932,
+      "step": 358912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.411553804697834e-05,
+      "loss": 3.7955,
+      "step": 359424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.410715209946782e-05,
+      "loss": 3.791,
+      "step": 359936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4098766151957296e-05,
+      "loss": 3.8047,
+      "step": 360448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.409039658325051e-05,
+      "loss": 3.7906,
+      "step": 360960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.408201063573999e-05,
+      "loss": 3.8134,
+      "step": 361472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.407362468822948e-05,
+      "loss": 3.7923,
+      "step": 361984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.406523874071895e-05,
+      "loss": 3.8032,
+      "step": 362496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.405685279320843e-05,
+      "loss": 3.7993,
+      "step": 363008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.404846684569791e-05,
+      "loss": 3.7916,
+      "step": 363520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.404009727699113e-05,
+      "loss": 3.8127,
+      "step": 364032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.40317113294806e-05,
+      "loss": 3.8072,
+      "step": 364544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.402332538197008e-05,
+      "loss": 3.7988,
+      "step": 365056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.401493943445956e-05,
+      "loss": 3.8004,
+      "step": 365568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.400655348694904e-05,
+      "loss": 3.7924,
+      "step": 366080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.399816753943852e-05,
+      "loss": 3.8035,
+      "step": 366592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.398979797073173e-05,
+      "loss": 3.786,
+      "step": 367104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.398141202322122e-05,
+      "loss": 3.8095,
+      "step": 367616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.39730260757107e-05,
+      "loss": 3.7983,
+      "step": 368128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.396464012820018e-05,
+      "loss": 3.7984,
+      "step": 368640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3956254180689657e-05,
+      "loss": 3.7919,
+      "step": 369152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3947868233179137e-05,
+      "loss": 3.7952,
+      "step": 369664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3939498664472346e-05,
+      "loss": 3.7877,
+      "step": 370176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3931112716961826e-05,
+      "loss": 3.7969,
+      "step": 370688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3922726769451306e-05,
+      "loss": 3.7909,
+      "step": 371200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3914340821940786e-05,
+      "loss": 3.8098,
+      "step": 371712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3905954874430265e-05,
+      "loss": 3.8004,
+      "step": 372224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3897568926919745e-05,
+      "loss": 3.7981,
+      "step": 372736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3889182979409225e-05,
+      "loss": 3.7826,
+      "step": 373248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3880797031898705e-05,
+      "loss": 3.8036,
+      "step": 373760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3872427463191914e-05,
+      "loss": 3.7997,
+      "step": 374272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3864041515681394e-05,
+      "loss": 3.8006,
+      "step": 374784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.385565556817088e-05,
+      "loss": 3.7915,
+      "step": 375296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.384726962066036e-05,
+      "loss": 3.7958,
+      "step": 375808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.383888367314984e-05,
+      "loss": 3.801,
+      "step": 376320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3830497725639314e-05,
+      "loss": 3.8046,
+      "step": 376832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3822111778128794e-05,
+      "loss": 3.7881,
+      "step": 377344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3813725830618274e-05,
+      "loss": 3.7997,
+      "step": 377856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.380535626191148e-05,
+      "loss": 3.7873,
+      "step": 378368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.379697031440096e-05,
+      "loss": 3.7869,
+      "step": 378880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.378860074569418e-05,
+      "loss": 3.8035,
+      "step": 379392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.378021479818366e-05,
+      "loss": 3.7936,
+      "step": 379904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.377182885067313e-05,
+      "loss": 3.7912,
+      "step": 380416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.376344290316262e-05,
+      "loss": 3.7854,
+      "step": 380928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.37550569556521e-05,
+      "loss": 3.7931,
+      "step": 381440
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.885331869125366,
+      "eval_runtime": 313.3388,
+      "eval_samples_per_second": 1217.822,
+      "eval_steps_per_second": 38.058,
+      "step": 381600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.374667100814158e-05,
+      "loss": 3.7881,
+      "step": 381952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.373828506063106e-05,
+      "loss": 3.7857,
+      "step": 382464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.372989911312054e-05,
+      "loss": 3.8027,
+      "step": 382976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.372152954441375e-05,
+      "loss": 3.7918,
+      "step": 383488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.371315997570696e-05,
+      "loss": 3.7972,
+      "step": 384000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.370477402819644e-05,
+      "loss": 3.792,
+      "step": 384512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.369638808068592e-05,
+      "loss": 3.7855,
+      "step": 385024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.36880021331754e-05,
+      "loss": 3.7789,
+      "step": 385536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.367961618566488e-05,
+      "loss": 3.7971,
+      "step": 386048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3671246616958086e-05,
+      "loss": 3.7969,
+      "step": 386560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.366286066944757e-05,
+      "loss": 3.7893,
+      "step": 387072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.365447472193705e-05,
+      "loss": 3.7991,
+      "step": 387584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.364608877442653e-05,
+      "loss": 3.7775,
+      "step": 388096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.363770282691601e-05,
+      "loss": 3.7876,
+      "step": 388608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.362931687940549e-05,
+      "loss": 3.7904,
+      "step": 389120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.36209473106987e-05,
+      "loss": 3.7767,
+      "step": 389632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.361256136318818e-05,
+      "loss": 3.7801,
+      "step": 390144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.360417541567766e-05,
+      "loss": 3.7804,
+      "step": 390656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.359578946816714e-05,
+      "loss": 3.7803,
+      "step": 391168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.358740352065662e-05,
+      "loss": 3.7994,
+      "step": 391680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.35790175731461e-05,
+      "loss": 3.784,
+      "step": 392192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.357063162563558e-05,
+      "loss": 3.7874,
+      "step": 392704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.356224567812506e-05,
+      "loss": 3.7779,
+      "step": 393216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.355387610941827e-05,
+      "loss": 3.7889,
+      "step": 393728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3545506540711486e-05,
+      "loss": 3.7724,
+      "step": 394240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3537120593200966e-05,
+      "loss": 3.7758,
+      "step": 394752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3528734645690446e-05,
+      "loss": 3.7782,
+      "step": 395264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3520348698179926e-05,
+      "loss": 3.7829,
+      "step": 395776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3511962750669406e-05,
+      "loss": 3.7698,
+      "step": 396288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3503576803158886e-05,
+      "loss": 3.7777,
+      "step": 396800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3495207234452095e-05,
+      "loss": 3.784,
+      "step": 397312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3486821286941575e-05,
+      "loss": 3.7864,
+      "step": 397824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3478435339431055e-05,
+      "loss": 3.7771,
+      "step": 398336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3470049391920535e-05,
+      "loss": 3.779,
+      "step": 398848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3461663444410015e-05,
+      "loss": 3.7813,
+      "step": 399360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3453293875703224e-05,
+      "loss": 3.78,
+      "step": 399872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.344490792819271e-05,
+      "loss": 3.7735,
+      "step": 400384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.343652198068219e-05,
+      "loss": 3.7682,
+      "step": 400896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.342813603317167e-05,
+      "loss": 3.7656,
+      "step": 401408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.341975008566115e-05,
+      "loss": 3.7747,
+      "step": 401920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3411364138150624e-05,
+      "loss": 3.7701,
+      "step": 402432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.340299456944384e-05,
+      "loss": 3.7796,
+      "step": 402944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.339460862193332e-05,
+      "loss": 3.7719,
+      "step": 403456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.338622267442279e-05,
+      "loss": 3.7715,
+      "step": 403968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.337783672691227e-05,
+      "loss": 3.7804,
+      "step": 404480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.336945077940175e-05,
+      "loss": 3.7733,
+      "step": 404992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.336108121069497e-05,
+      "loss": 3.7803,
+      "step": 405504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.335269526318445e-05,
+      "loss": 3.7646,
+      "step": 406016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.334430931567393e-05,
+      "loss": 3.7609,
+      "step": 406528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.333592336816341e-05,
+      "loss": 3.7864,
+      "step": 407040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.332753742065289e-05,
+      "loss": 3.7696,
+      "step": 407552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.331915147314237e-05,
+      "loss": 3.767,
+      "step": 408064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.331078190443558e-05,
+      "loss": 3.7663,
+      "step": 408576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.330239595692506e-05,
+      "loss": 3.7759,
+      "step": 409088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.329401000941454e-05,
+      "loss": 3.7548,
+      "step": 409600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.328562406190402e-05,
+      "loss": 3.7751,
+      "step": 410112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.32772381143935e-05,
+      "loss": 3.7716,
+      "step": 410624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.326885216688298e-05,
+      "loss": 3.7665,
+      "step": 411136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.326046621937246e-05,
+      "loss": 3.786,
+      "step": 411648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3252096650665666e-05,
+      "loss": 3.7606,
+      "step": 412160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3243710703155146e-05,
+      "loss": 3.7625,
+      "step": 412672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.323534113444836e-05,
+      "loss": 3.7741,
+      "step": 413184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.322695518693784e-05,
+      "loss": 3.7567,
+      "step": 413696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.321856923942732e-05,
+      "loss": 3.7559,
+      "step": 414208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.32101832919168e-05,
+      "loss": 3.7639,
+      "step": 414720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.320179734440628e-05,
+      "loss": 3.7667,
+      "step": 415232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.319341139689576e-05,
+      "loss": 3.7572,
+      "step": 415744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.318504182818897e-05,
+      "loss": 3.755,
+      "step": 416256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.317665588067845e-05,
+      "loss": 3.7605,
+      "step": 416768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.316826993316793e-05,
+      "loss": 3.7707,
+      "step": 417280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.315988398565741e-05,
+      "loss": 3.7699,
+      "step": 417792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.315149803814689e-05,
+      "loss": 3.7686,
+      "step": 418304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.314311209063637e-05,
+      "loss": 3.7739,
+      "step": 418816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.313474252192958e-05,
+      "loss": 3.773,
+      "step": 419328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3126356574419067e-05,
+      "loss": 3.7835,
+      "step": 419840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3117970626908546e-05,
+      "loss": 3.7513,
+      "step": 420352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3109584679398026e-05,
+      "loss": 3.7774,
+      "step": 420864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3101215110691236e-05,
+      "loss": 3.7727,
+      "step": 421376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3092829163180715e-05,
+      "loss": 3.7509,
+      "step": 421888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3084443215670195e-05,
+      "loss": 3.7766,
+      "step": 422400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3076057268159675e-05,
+      "loss": 3.7634,
+      "step": 422912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3067671320649155e-05,
+      "loss": 3.7733,
+      "step": 423424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3059301751942364e-05,
+      "loss": 3.7745,
+      "step": 423936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3050915804431844e-05,
+      "loss": 3.7513,
+      "step": 424448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3042529856921324e-05,
+      "loss": 3.7548,
+      "step": 424960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3034143909410804e-05,
+      "loss": 3.7596,
+      "step": 425472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3025757961900284e-05,
+      "loss": 3.7682,
+      "step": 425984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.30173883931935e-05,
+      "loss": 3.7613,
+      "step": 426496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.300900244568298e-05,
+      "loss": 3.7597,
+      "step": 427008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.300061649817245e-05,
+      "loss": 3.7583,
+      "step": 427520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.299223055066193e-05,
+      "loss": 3.7588,
+      "step": 428032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.298386098195515e-05,
+      "loss": 3.7597,
+      "step": 428544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.297547503444463e-05,
+      "loss": 3.7545,
+      "step": 429056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.29670890869341e-05,
+      "loss": 3.77,
+      "step": 429568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.295870313942358e-05,
+      "loss": 3.7695,
+      "step": 430080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.29503335707168e-05,
+      "loss": 3.7697,
+      "step": 430592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.294194762320628e-05,
+      "loss": 3.7498,
+      "step": 431104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.293356167569576e-05,
+      "loss": 3.7634,
+      "step": 431616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.292517572818524e-05,
+      "loss": 3.7602,
+      "step": 432128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.291678978067472e-05,
+      "loss": 3.7652,
+      "step": 432640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.29084038331642e-05,
+      "loss": 3.765,
+      "step": 433152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.290003426445741e-05,
+      "loss": 3.7585,
+      "step": 433664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.289164831694689e-05,
+      "loss": 3.7478,
+      "step": 434176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.288326236943637e-05,
+      "loss": 3.7614,
+      "step": 434688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.287487642192585e-05,
+      "loss": 3.7511,
+      "step": 435200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.286649047441533e-05,
+      "loss": 3.7474,
+      "step": 435712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.285810452690481e-05,
+      "loss": 3.742,
+      "step": 436224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2849734958198016e-05,
+      "loss": 3.7595,
+      "step": 436736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2841349010687496e-05,
+      "loss": 3.7521,
+      "step": 437248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2832963063176976e-05,
+      "loss": 3.7655,
+      "step": 437760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2824577115666456e-05,
+      "loss": 3.7494,
+      "step": 438272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.281619116815594e-05,
+      "loss": 3.7575,
+      "step": 438784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.280782159944915e-05,
+      "loss": 3.7561,
+      "step": 439296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.279943565193863e-05,
+      "loss": 3.7459,
+      "step": 439808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.279104970442811e-05,
+      "loss": 3.7692,
+      "step": 440320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.278266375691759e-05,
+      "loss": 3.7606,
+      "step": 440832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.27742941882108e-05,
+      "loss": 3.7545,
+      "step": 441344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.276590824070028e-05,
+      "loss": 3.7555,
+      "step": 441856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.275752229318976e-05,
+      "loss": 3.7507,
+      "step": 442368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.274913634567924e-05,
+      "loss": 3.7588,
+      "step": 442880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.274076677697245e-05,
+      "loss": 3.7404,
+      "step": 443392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.273238082946193e-05,
+      "loss": 3.7657,
+      "step": 443904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.272399488195141e-05,
+      "loss": 3.7495,
+      "step": 444416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2715608934440896e-05,
+      "loss": 3.7517,
+      "step": 444928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2707239365734105e-05,
+      "loss": 3.7522,
+      "step": 445440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2698853418223585e-05,
+      "loss": 3.7499,
+      "step": 445952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2690467470713065e-05,
+      "loss": 3.7482,
+      "step": 446464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2682081523202545e-05,
+      "loss": 3.751,
+      "step": 446976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2673695575692025e-05,
+      "loss": 3.7512,
+      "step": 447488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2665309628181505e-05,
+      "loss": 3.7625,
+      "step": 448000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2656923680670985e-05,
+      "loss": 3.7537,
+      "step": 448512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2648554111964194e-05,
+      "loss": 3.7575,
+      "step": 449024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2640168164453674e-05,
+      "loss": 3.7369,
+      "step": 449536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2631782216943154e-05,
+      "loss": 3.7635,
+      "step": 450048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2623396269432634e-05,
+      "loss": 3.7537,
+      "step": 450560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2615010321922114e-05,
+      "loss": 3.7566,
+      "step": 451072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2606624374411594e-05,
+      "loss": 3.7519,
+      "step": 451584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.259825480570481e-05,
+      "loss": 3.7503,
+      "step": 452096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.258986885819429e-05,
+      "loss": 3.7606,
+      "step": 452608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.258148291068376e-05,
+      "loss": 3.7599,
+      "step": 453120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.257309696317324e-05,
+      "loss": 3.7441,
+      "step": 453632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.256471101566272e-05,
+      "loss": 3.7524,
+      "step": 454144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.25563250681522e-05,
+      "loss": 3.7479,
+      "step": 454656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.254793912064168e-05,
+      "loss": 3.7438,
+      "step": 455168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.253958593073863e-05,
+      "loss": 3.7605,
+      "step": 455680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.253119998322811e-05,
+      "loss": 3.7501,
+      "step": 456192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.252281403571758e-05,
+      "loss": 3.7528,
+      "step": 456704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.251442808820707e-05,
+      "loss": 3.7379,
+      "step": 457216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.250604214069655e-05,
+      "loss": 3.7539,
+      "step": 457728
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8691282272338867,
+      "eval_runtime": 305.0865,
+      "eval_samples_per_second": 1250.763,
+      "eval_steps_per_second": 39.087,
+      "step": 457920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.249765619318603e-05,
+      "loss": 3.7454,
+      "step": 458240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.248927024567551e-05,
+      "loss": 3.7399,
+      "step": 458752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.248088429816499e-05,
+      "loss": 3.7582,
+      "step": 459264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.247249835065447e-05,
+      "loss": 3.7529,
+      "step": 459776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.246411240314395e-05,
+      "loss": 3.7516,
+      "step": 460288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2455742834437156e-05,
+      "loss": 3.7496,
+      "step": 460800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2447356886926636e-05,
+      "loss": 3.7367,
+      "step": 461312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2438970939416116e-05,
+      "loss": 3.7428,
+      "step": 461824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2430584991905596e-05,
+      "loss": 3.7521,
+      "step": 462336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2422199044395076e-05,
+      "loss": 3.7583,
+      "step": 462848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2413829475688285e-05,
+      "loss": 3.7457,
+      "step": 463360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2405443528177765e-05,
+      "loss": 3.7578,
+      "step": 463872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.239705758066725e-05,
+      "loss": 3.7393,
+      "step": 464384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.238867163315673e-05,
+      "loss": 3.7399,
+      "step": 464896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.238030206444994e-05,
+      "loss": 3.7516,
+      "step": 465408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.237191611693942e-05,
+      "loss": 3.7342,
+      "step": 465920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.23635301694289e-05,
+      "loss": 3.738,
+      "step": 466432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.235514422191838e-05,
+      "loss": 3.7383,
+      "step": 466944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.234677465321159e-05,
+      "loss": 3.7413,
+      "step": 467456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.233838870570107e-05,
+      "loss": 3.7521,
+      "step": 467968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.233000275819055e-05,
+      "loss": 3.7449,
+      "step": 468480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.232161681068003e-05,
+      "loss": 3.7473,
+      "step": 468992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.231323086316951e-05,
+      "loss": 3.7338,
+      "step": 469504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.230486129446272e-05,
+      "loss": 3.7479,
+      "step": 470016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2296475346952206e-05,
+      "loss": 3.7332,
+      "step": 470528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2288089399441686e-05,
+      "loss": 3.7352,
+      "step": 471040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2279703451931166e-05,
+      "loss": 3.733,
+      "step": 471552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2271333883224375e-05,
+      "loss": 3.7383,
+      "step": 472064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2262947935713855e-05,
+      "loss": 3.7369,
+      "step": 472576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2254561988203335e-05,
+      "loss": 3.7322,
+      "step": 473088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2246176040692814e-05,
+      "loss": 3.745,
+      "step": 473600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2237806471986024e-05,
+      "loss": 3.7442,
+      "step": 474112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2229420524475504e-05,
+      "loss": 3.736,
+      "step": 474624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2221034576964983e-05,
+      "loss": 3.7405,
+      "step": 475136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2212648629454463e-05,
+      "loss": 3.7395,
+      "step": 475648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.220427906074767e-05,
+      "loss": 3.7396,
+      "step": 476160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.219589311323716e-05,
+      "loss": 3.7379,
+      "step": 476672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.218750716572664e-05,
+      "loss": 3.7234,
+      "step": 477184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.217912121821612e-05,
+      "loss": 3.7226,
+      "step": 477696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.21707352707056e-05,
+      "loss": 3.7361,
+      "step": 478208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.216234932319507e-05,
+      "loss": 3.7311,
+      "step": 478720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.215397975448829e-05,
+      "loss": 3.7357,
+      "step": 479232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.214559380697777e-05,
+      "loss": 3.733,
+      "step": 479744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.213720785946725e-05,
+      "loss": 3.7317,
+      "step": 480256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.212882191195672e-05,
+      "loss": 3.741,
+      "step": 480768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.21204359644462e-05,
+      "loss": 3.7328,
+      "step": 481280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.211206639573942e-05,
+      "loss": 3.7419,
+      "step": 481792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.21036804482289e-05,
+      "loss": 3.725,
+      "step": 482304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.209529450071838e-05,
+      "loss": 3.7164,
+      "step": 482816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.208690855320786e-05,
+      "loss": 3.7459,
+      "step": 483328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.207853898450107e-05,
+      "loss": 3.7341,
+      "step": 483840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2070153036990546e-05,
+      "loss": 3.7257,
+      "step": 484352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2061767089480026e-05,
+      "loss": 3.7286,
+      "step": 484864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2053381141969506e-05,
+      "loss": 3.7319,
+      "step": 485376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.204501157326272e-05,
+      "loss": 3.7189,
+      "step": 485888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2036625625752195e-05,
+      "loss": 3.7317,
+      "step": 486400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2028239678241675e-05,
+      "loss": 3.7332,
+      "step": 486912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2019853730731155e-05,
+      "loss": 3.7306,
+      "step": 487424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2011467783220635e-05,
+      "loss": 3.7448,
+      "step": 487936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.200309821451385e-05,
+      "loss": 3.7221,
+      "step": 488448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.199471226700333e-05,
+      "loss": 3.7245,
+      "step": 488960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.198632631949281e-05,
+      "loss": 3.7382,
+      "step": 489472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.197794037198229e-05,
+      "loss": 3.7167,
+      "step": 489984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.196955442447177e-05,
+      "loss": 3.7142,
+      "step": 490496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.196118485576498e-05,
+      "loss": 3.7268,
+      "step": 491008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.195279890825446e-05,
+      "loss": 3.7275,
+      "step": 491520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.194441296074394e-05,
+      "loss": 3.7175,
+      "step": 492032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.193602701323342e-05,
+      "loss": 3.7179,
+      "step": 492544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.19276410657229e-05,
+      "loss": 3.7173,
+      "step": 493056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.191925511821238e-05,
+      "loss": 3.737,
+      "step": 493568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.191086917070186e-05,
+      "loss": 3.7316,
+      "step": 494080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.190249960199507e-05,
+      "loss": 3.7294,
+      "step": 494592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.189411365448455e-05,
+      "loss": 3.7361,
+      "step": 495104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.188572770697403e-05,
+      "loss": 3.7326,
+      "step": 495616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1877341759463515e-05,
+      "loss": 3.7454,
+      "step": 496128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1868955811952995e-05,
+      "loss": 3.7158,
+      "step": 496640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1860569864442475e-05,
+      "loss": 3.7357,
+      "step": 497152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1852183916931955e-05,
+      "loss": 3.7322,
+      "step": 497664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1843814348225164e-05,
+      "loss": 3.7151,
+      "step": 498176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1835428400714644e-05,
+      "loss": 3.733,
+      "step": 498688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1827042453204124e-05,
+      "loss": 3.7276,
+      "step": 499200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1818656505693604e-05,
+      "loss": 3.7347,
+      "step": 499712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.181028693698681e-05,
+      "loss": 3.7366,
+      "step": 500224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.180190098947629e-05,
+      "loss": 3.7179,
+      "step": 500736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.179351504196577e-05,
+      "loss": 3.7111,
+      "step": 501248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.178512909445525e-05,
+      "loss": 3.7198,
+      "step": 501760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.177675952574847e-05,
+      "loss": 3.7343,
+      "step": 502272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.176837357823795e-05,
+      "loss": 3.7221,
+      "step": 502784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.175998763072743e-05,
+      "loss": 3.7226,
+      "step": 503296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.175160168321691e-05,
+      "loss": 3.7234,
+      "step": 503808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.174323211451012e-05,
+      "loss": 3.7215,
+      "step": 504320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.17348461669996e-05,
+      "loss": 3.7242,
+      "step": 504832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.172646021948908e-05,
+      "loss": 3.7178,
+      "step": 505344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.171807427197855e-05,
+      "loss": 3.7311,
+      "step": 505856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.170970470327177e-05,
+      "loss": 3.7313,
+      "step": 506368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.170131875576125e-05,
+      "loss": 3.7314,
+      "step": 506880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.169293280825073e-05,
+      "loss": 3.7131,
+      "step": 507392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1684546860740207e-05,
+      "loss": 3.7236,
+      "step": 507904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1676160913229687e-05,
+      "loss": 3.7236,
+      "step": 508416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16677913445229e-05,
+      "loss": 3.727,
+      "step": 508928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.165940539701238e-05,
+      "loss": 3.7266,
+      "step": 509440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1651019449501856e-05,
+      "loss": 3.7163,
+      "step": 509952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1642633501991336e-05,
+      "loss": 3.7149,
+      "step": 510464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.163426393328455e-05,
+      "loss": 3.722,
+      "step": 510976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1625877985774025e-05,
+      "loss": 3.7113,
+      "step": 511488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1617492038263505e-05,
+      "loss": 3.7185,
+      "step": 512000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1609106090752984e-05,
+      "loss": 3.6988,
+      "step": 512512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16007365220462e-05,
+      "loss": 3.7252,
+      "step": 513024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1592350574535674e-05,
+      "loss": 3.716,
+      "step": 513536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.158396462702516e-05,
+      "loss": 3.7256,
+      "step": 514048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.157557867951464e-05,
+      "loss": 3.7137,
+      "step": 514560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.156719273200412e-05,
+      "loss": 3.7228,
+      "step": 515072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.155882316329733e-05,
+      "loss": 3.7221,
+      "step": 515584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.155043721578681e-05,
+      "loss": 3.7062,
+      "step": 516096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.154205126827629e-05,
+      "loss": 3.7337,
+      "step": 516608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.153366532076577e-05,
+      "loss": 3.7232,
+      "step": 517120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.152529575205898e-05,
+      "loss": 3.718,
+      "step": 517632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.151690980454846e-05,
+      "loss": 3.7226,
+      "step": 518144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.150852385703794e-05,
+      "loss": 3.7074,
+      "step": 518656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.150013790952742e-05,
+      "loss": 3.7265,
+      "step": 519168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.149176834082063e-05,
+      "loss": 3.707,
+      "step": 519680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1483382393310114e-05,
+      "loss": 3.7263,
+      "step": 520192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1474996445799594e-05,
+      "loss": 3.7114,
+      "step": 520704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1466610498289074e-05,
+      "loss": 3.7177,
+      "step": 521216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.145824092958228e-05,
+      "loss": 3.7126,
+      "step": 521728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.144985498207176e-05,
+      "loss": 3.715,
+      "step": 522240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.144146903456124e-05,
+      "loss": 3.7146,
+      "step": 522752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.143308308705072e-05,
+      "loss": 3.7111,
+      "step": 523264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14246971395402e-05,
+      "loss": 3.7153,
+      "step": 523776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.141631119202968e-05,
+      "loss": 3.7269,
+      "step": 524288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.140794162332289e-05,
+      "loss": 3.715,
+      "step": 524800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.139955567581237e-05,
+      "loss": 3.7241,
+      "step": 525312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.139116972830185e-05,
+      "loss": 3.7013,
+      "step": 525824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.138278378079133e-05,
+      "loss": 3.7262,
+      "step": 526336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.137441421208455e-05,
+      "loss": 3.7216,
+      "step": 526848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.136602826457403e-05,
+      "loss": 3.7158,
+      "step": 527360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.135764231706351e-05,
+      "loss": 3.7148,
+      "step": 527872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.134925636955299e-05,
+      "loss": 3.716,
+      "step": 528384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1340886800846197e-05,
+      "loss": 3.7217,
+      "step": 528896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1332500853335676e-05,
+      "loss": 3.7275,
+      "step": 529408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1324114905825156e-05,
+      "loss": 3.7061,
+      "step": 529920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1315728958314636e-05,
+      "loss": 3.7203,
+      "step": 530432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1307343010804116e-05,
+      "loss": 3.7114,
+      "step": 530944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1298973442097325e-05,
+      "loss": 3.7085,
+      "step": 531456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1290587494586805e-05,
+      "loss": 3.7199,
+      "step": 531968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1282201547076285e-05,
+      "loss": 3.7147,
+      "step": 532480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1273815599565765e-05,
+      "loss": 3.7158,
+      "step": 532992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1265429652055245e-05,
+      "loss": 3.7024,
+      "step": 533504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.125706008334846e-05,
+      "loss": 3.7232,
+      "step": 534016
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8591363430023193,
+      "eval_runtime": 304.426,
+      "eval_samples_per_second": 1253.477,
+      "eval_steps_per_second": 39.172,
+      "step": 534240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.124867413583794e-05,
+      "loss": 3.7092,
+      "step": 534528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.124028818832742e-05,
+      "loss": 3.7051,
+      "step": 535040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.12319022408169e-05,
+      "loss": 3.7204,
+      "step": 535552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.122351629330638e-05,
+      "loss": 3.7187,
+      "step": 536064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.121514672459959e-05,
+      "loss": 3.715,
+      "step": 536576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.120676077708907e-05,
+      "loss": 3.7163,
+      "step": 537088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.119837482957855e-05,
+      "loss": 3.7037,
+      "step": 537600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.118998888206803e-05,
+      "loss": 3.7064,
+      "step": 538112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.118161931336124e-05,
+      "loss": 3.7169,
+      "step": 538624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.117323336585072e-05,
+      "loss": 3.7199,
+      "step": 539136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.11648474183402e-05,
+      "loss": 3.7117,
+      "step": 539648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1156461470829686e-05,
+      "loss": 3.7209,
+      "step": 540160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1148091902122895e-05,
+      "loss": 3.708,
+      "step": 540672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1139705954612375e-05,
+      "loss": 3.7062,
+      "step": 541184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1131320007101855e-05,
+      "loss": 3.7136,
+      "step": 541696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1122934059591335e-05,
+      "loss": 3.6981,
+      "step": 542208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.111454811208081e-05,
+      "loss": 3.7049,
+      "step": 542720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1106178543374024e-05,
+      "loss": 3.702,
+      "step": 543232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1097792595863504e-05,
+      "loss": 3.7072,
+      "step": 543744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1089406648352984e-05,
+      "loss": 3.7147,
+      "step": 544256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.108102070084246e-05,
+      "loss": 3.7181,
+      "step": 544768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.107263475333194e-05,
+      "loss": 3.7095,
+      "step": 545280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.106426518462515e-05,
+      "loss": 3.7002,
+      "step": 545792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.105587923711463e-05,
+      "loss": 3.7132,
+      "step": 546304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.104749328960411e-05,
+      "loss": 3.6954,
+      "step": 546816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.103910734209359e-05,
+      "loss": 3.7033,
+      "step": 547328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.103073777338681e-05,
+      "loss": 3.6971,
+      "step": 547840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.102235182587628e-05,
+      "loss": 3.7073,
+      "step": 548352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.101396587836576e-05,
+      "loss": 3.6984,
+      "step": 548864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.100557993085524e-05,
+      "loss": 3.7025,
+      "step": 549376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.099721036214846e-05,
+      "loss": 3.7063,
+      "step": 549888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.098882441463793e-05,
+      "loss": 3.7146,
+      "step": 550400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.098043846712741e-05,
+      "loss": 3.7029,
+      "step": 550912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.097205251961689e-05,
+      "loss": 3.7006,
+      "step": 551424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0963682950910106e-05,
+      "loss": 3.7063,
+      "step": 551936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0955297003399586e-05,
+      "loss": 3.7053,
+      "step": 552448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0946911055889066e-05,
+      "loss": 3.7067,
+      "step": 552960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0938525108378546e-05,
+      "loss": 3.6868,
+      "step": 553472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0930155539671755e-05,
+      "loss": 3.6927,
+      "step": 553984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0921769592161235e-05,
+      "loss": 3.699,
+      "step": 554496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0913383644650715e-05,
+      "loss": 3.7017,
+      "step": 555008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0904997697140195e-05,
+      "loss": 3.7016,
+      "step": 555520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0896628128433404e-05,
+      "loss": 3.7006,
+      "step": 556032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0888242180922884e-05,
+      "loss": 3.6971,
+      "step": 556544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0879856233412364e-05,
+      "loss": 3.7072,
+      "step": 557056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0871470285901844e-05,
+      "loss": 3.6963,
+      "step": 557568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.086310071719506e-05,
+      "loss": 3.7159,
+      "step": 558080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.085471476968454e-05,
+      "loss": 3.6894,
+      "step": 558592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.084632882217402e-05,
+      "loss": 3.6771,
+      "step": 559104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.08379428746635e-05,
+      "loss": 3.7169,
+      "step": 559616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.082957330595671e-05,
+      "loss": 3.6987,
+      "step": 560128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.082118735844619e-05,
+      "loss": 3.6982,
+      "step": 560640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.081280141093567e-05,
+      "loss": 3.6972,
+      "step": 561152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.080441546342515e-05,
+      "loss": 3.6962,
+      "step": 561664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.079602951591463e-05,
+      "loss": 3.6879,
+      "step": 562176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.078765994720784e-05,
+      "loss": 3.6961,
+      "step": 562688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.077927399969732e-05,
+      "loss": 3.6997,
+      "step": 563200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.07708880521868e-05,
+      "loss": 3.6986,
+      "step": 563712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0762502104676284e-05,
+      "loss": 3.7131,
+      "step": 564224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0754132535969494e-05,
+      "loss": 3.6883,
+      "step": 564736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0745746588458974e-05,
+      "loss": 3.6886,
+      "step": 565248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0737360640948453e-05,
+      "loss": 3.708,
+      "step": 565760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0728974693437933e-05,
+      "loss": 3.6828,
+      "step": 566272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.072058874592741e-05,
+      "loss": 3.6808,
+      "step": 566784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.071221917722062e-05,
+      "loss": 3.6937,
+      "step": 567296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.07038332297101e-05,
+      "loss": 3.6957,
+      "step": 567808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.069544728219958e-05,
+      "loss": 3.6812,
+      "step": 568320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.068706133468906e-05,
+      "loss": 3.6901,
+      "step": 568832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.067869176598227e-05,
+      "loss": 3.685,
+      "step": 569344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.067030581847175e-05,
+      "loss": 3.7013,
+      "step": 569856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.066191987096124e-05,
+      "loss": 3.7015,
+      "step": 570368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.065353392345072e-05,
+      "loss": 3.6929,
+      "step": 570880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.06451479759402e-05,
+      "loss": 3.7013,
+      "step": 571392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.063677840723341e-05,
+      "loss": 3.7005,
+      "step": 571904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.062839245972289e-05,
+      "loss": 3.715,
+      "step": 572416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.062000651221237e-05,
+      "loss": 3.682,
+      "step": 572928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.061162056470185e-05,
+      "loss": 3.7084,
+      "step": 573440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0603250995995056e-05,
+      "loss": 3.6954,
+      "step": 573952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0594865048484536e-05,
+      "loss": 3.6854,
+      "step": 574464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0586479100974016e-05,
+      "loss": 3.7005,
+      "step": 574976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0578093153463496e-05,
+      "loss": 3.6973,
+      "step": 575488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0569707205952976e-05,
+      "loss": 3.6994,
+      "step": 576000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.056133763724619e-05,
+      "loss": 3.7073,
+      "step": 576512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.055295168973567e-05,
+      "loss": 3.6839,
+      "step": 577024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.054456574222515e-05,
+      "loss": 3.6777,
+      "step": 577536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.053617979471463e-05,
+      "loss": 3.6909,
+      "step": 578048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.052781022600784e-05,
+      "loss": 3.7015,
+      "step": 578560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.051942427849732e-05,
+      "loss": 3.6914,
+      "step": 579072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.05110383309868e-05,
+      "loss": 3.69,
+      "step": 579584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.050265238347628e-05,
+      "loss": 3.6868,
+      "step": 580096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.049428281476949e-05,
+      "loss": 3.6912,
+      "step": 580608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.048589686725897e-05,
+      "loss": 3.6906,
+      "step": 581120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.047751091974845e-05,
+      "loss": 3.6885,
+      "step": 581632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.046912497223793e-05,
+      "loss": 3.6976,
+      "step": 582144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.046075540353114e-05,
+      "loss": 3.7013,
+      "step": 582656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0452369456020625e-05,
+      "loss": 3.6973,
+      "step": 583168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0443983508510105e-05,
+      "loss": 3.6821,
+      "step": 583680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0435597560999585e-05,
+      "loss": 3.694,
+      "step": 584192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0427211613489065e-05,
+      "loss": 3.687,
+      "step": 584704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0418842044782274e-05,
+      "loss": 3.696,
+      "step": 585216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0410456097271754e-05,
+      "loss": 3.6961,
+      "step": 585728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0402070149761234e-05,
+      "loss": 3.6903,
+      "step": 586240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.039368420225071e-05,
+      "loss": 3.6786,
+      "step": 586752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.038531463354392e-05,
+      "loss": 3.6853,
+      "step": 587264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.03769286860334e-05,
+      "loss": 3.6871,
+      "step": 587776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.036854273852288e-05,
+      "loss": 3.6848,
+      "step": 588288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.036015679101236e-05,
+      "loss": 3.6696,
+      "step": 588800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.035178722230558e-05,
+      "loss": 3.6929,
+      "step": 589312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.034340127479506e-05,
+      "loss": 3.6892,
+      "step": 589824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.033501532728454e-05,
+      "loss": 3.6909,
+      "step": 590336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.032662937977401e-05,
+      "loss": 3.6793,
+      "step": 590848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.031824343226349e-05,
+      "loss": 3.6984,
+      "step": 591360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.030987386355671e-05,
+      "loss": 3.6869,
+      "step": 591872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.030148791604618e-05,
+      "loss": 3.6768,
+      "step": 592384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.029310196853566e-05,
+      "loss": 3.6994,
+      "step": 592896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.028471602102514e-05,
+      "loss": 3.6924,
+      "step": 593408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.027634645231836e-05,
+      "loss": 3.6844,
+      "step": 593920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.026796050480783e-05,
+      "loss": 3.6943,
+      "step": 594432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.025957455729732e-05,
+      "loss": 3.677,
+      "step": 594944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.02511886097868e-05,
+      "loss": 3.6962,
+      "step": 595456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.024281904108001e-05,
+      "loss": 3.6787,
+      "step": 595968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0234433093569486e-05,
+      "loss": 3.6913,
+      "step": 596480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0226047146058966e-05,
+      "loss": 3.681,
+      "step": 596992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0217661198548446e-05,
+      "loss": 3.6853,
+      "step": 597504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0209291629841655e-05,
+      "loss": 3.6821,
+      "step": 598016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0200905682331135e-05,
+      "loss": 3.6838,
+      "step": 598528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0192519734820615e-05,
+      "loss": 3.6811,
+      "step": 599040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0184133787310095e-05,
+      "loss": 3.6802,
+      "step": 599552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0175747839799575e-05,
+      "loss": 3.6884,
+      "step": 600064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0167378271092784e-05,
+      "loss": 3.6963,
+      "step": 600576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.015899232358227e-05,
+      "loss": 3.6791,
+      "step": 601088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.015060637607175e-05,
+      "loss": 3.6974,
+      "step": 601600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.014222042856123e-05,
+      "loss": 3.6687,
+      "step": 602112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.013385085985444e-05,
+      "loss": 3.6959,
+      "step": 602624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.012546491234392e-05,
+      "loss": 3.6897,
+      "step": 603136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.01170789648334e-05,
+      "loss": 3.6861,
+      "step": 603648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.010869301732288e-05,
+      "loss": 3.6825,
+      "step": 604160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.010030706981236e-05,
+      "loss": 3.6891,
+      "step": 604672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.009193750110557e-05,
+      "loss": 3.6911,
+      "step": 605184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.008355155359505e-05,
+      "loss": 3.6984,
+      "step": 605696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.007516560608453e-05,
+      "loss": 3.6742,
+      "step": 606208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.006677965857401e-05,
+      "loss": 3.6864,
+      "step": 606720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0058410089867224e-05,
+      "loss": 3.6797,
+      "step": 607232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0050024142356704e-05,
+      "loss": 3.6839,
+      "step": 607744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0041638194846184e-05,
+      "loss": 3.6838,
+      "step": 608256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0033252247335664e-05,
+      "loss": 3.6838,
+      "step": 608768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.002488267862887e-05,
+      "loss": 3.6891,
+      "step": 609280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.001649673111835e-05,
+      "loss": 3.6716,
+      "step": 609792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.000811078360783e-05,
+      "loss": 3.692,
+      "step": 610304
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8517251014709473,
+      "eval_runtime": 304.9407,
+      "eval_samples_per_second": 1251.361,
+      "eval_steps_per_second": 39.106,
+      "step": 610560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.999972483609731e-05,
+      "loss": 3.6786,
+      "step": 610816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.999133888858679e-05,
+      "loss": 3.6774,
+      "step": 611328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.998295294107627e-05,
+      "loss": 3.6894,
+      "step": 611840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.997456699356575e-05,
+      "loss": 3.6866,
+      "step": 612352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.996618104605523e-05,
+      "loss": 3.6894,
+      "step": 612864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.995781147734844e-05,
+      "loss": 3.6824,
+      "step": 613376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.994942552983792e-05,
+      "loss": 3.6766,
+      "step": 613888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.994103958232741e-05,
+      "loss": 3.6786,
+      "step": 614400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.993265363481689e-05,
+      "loss": 3.6819,
+      "step": 614912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.99242840661101e-05,
+      "loss": 3.6906,
+      "step": 615424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.991589811859958e-05,
+      "loss": 3.6853,
+      "step": 615936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.990751217108906e-05,
+      "loss": 3.6883,
+      "step": 616448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.989914260238227e-05,
+      "loss": 3.6763,
+      "step": 616960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.989075665487175e-05,
+      "loss": 3.6773,
+      "step": 617472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9882370707361227e-05,
+      "loss": 3.6868,
+      "step": 617984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9873984759850707e-05,
+      "loss": 3.6689,
+      "step": 618496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9865598812340187e-05,
+      "loss": 3.6757,
+      "step": 619008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9857212864829666e-05,
+      "loss": 3.6735,
+      "step": 619520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9848826917319146e-05,
+      "loss": 3.6788,
+      "step": 620032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9840440969808626e-05,
+      "loss": 3.6819,
+      "step": 620544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.983207140110184e-05,
+      "loss": 3.6888,
+      "step": 621056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9823685453591315e-05,
+      "loss": 3.6814,
+      "step": 621568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9815299506080795e-05,
+      "loss": 3.6734,
+      "step": 622080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9806913558570275e-05,
+      "loss": 3.6799,
+      "step": 622592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9798527611059755e-05,
+      "loss": 3.6707,
+      "step": 623104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9790158042352964e-05,
+      "loss": 3.6756,
+      "step": 623616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9781772094842444e-05,
+      "loss": 3.668,
+      "step": 624128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9773386147331924e-05,
+      "loss": 3.6794,
+      "step": 624640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9765000199821404e-05,
+      "loss": 3.6752,
+      "step": 625152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.975663063111461e-05,
+      "loss": 3.6681,
+      "step": 625664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974824468360409e-05,
+      "loss": 3.6753,
+      "step": 626176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.973985873609358e-05,
+      "loss": 3.6863,
+      "step": 626688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.973147278858306e-05,
+      "loss": 3.6734,
+      "step": 627200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.972310321987627e-05,
+      "loss": 3.6735,
+      "step": 627712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.971471727236575e-05,
+      "loss": 3.6739,
+      "step": 628224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.970633132485523e-05,
+      "loss": 3.6805,
+      "step": 628736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.969794537734471e-05,
+      "loss": 3.6727,
+      "step": 629248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.968957580863792e-05,
+      "loss": 3.6601,
+      "step": 629760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.96811898611274e-05,
+      "loss": 3.6625,
+      "step": 630272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.967280391361688e-05,
+      "loss": 3.6708,
+      "step": 630784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.966441796610636e-05,
+      "loss": 3.6697,
+      "step": 631296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.965604839739957e-05,
+      "loss": 3.6702,
+      "step": 631808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.964766244988905e-05,
+      "loss": 3.6732,
+      "step": 632320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9639276502378534e-05,
+      "loss": 3.6704,
+      "step": 632832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9630890554868014e-05,
+      "loss": 3.6757,
+      "step": 633344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9622504607357494e-05,
+      "loss": 3.6687,
+      "step": 633856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.96141350386507e-05,
+      "loss": 3.6834,
+      "step": 634368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.960574909114018e-05,
+      "loss": 3.6642,
+      "step": 634880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.959736314362966e-05,
+      "loss": 3.6479,
+      "step": 635392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.958897719611914e-05,
+      "loss": 3.6875,
+      "step": 635904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.958060762741235e-05,
+      "loss": 3.6687,
+      "step": 636416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.957222167990183e-05,
+      "loss": 3.6712,
+      "step": 636928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.956383573239131e-05,
+      "loss": 3.6658,
+      "step": 637440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.955544978488079e-05,
+      "loss": 3.6645,
+      "step": 637952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9547080216174e-05,
+      "loss": 3.659,
+      "step": 638464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.953869426866349e-05,
+      "loss": 3.6657,
+      "step": 638976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.953030832115297e-05,
+      "loss": 3.6733,
+      "step": 639488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.952192237364245e-05,
+      "loss": 3.6706,
+      "step": 640000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9513552804935656e-05,
+      "loss": 3.6857,
+      "step": 640512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9505166857425136e-05,
+      "loss": 3.66,
+      "step": 641024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9496780909914616e-05,
+      "loss": 3.6587,
+      "step": 641536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9488394962404096e-05,
+      "loss": 3.6817,
+      "step": 642048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9480025393697305e-05,
+      "loss": 3.6538,
+      "step": 642560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9471639446186785e-05,
+      "loss": 3.6493,
+      "step": 643072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9463253498676265e-05,
+      "loss": 3.6652,
+      "step": 643584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9454867551165745e-05,
+      "loss": 3.6665,
+      "step": 644096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9446497982458954e-05,
+      "loss": 3.6557,
+      "step": 644608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.943811203494844e-05,
+      "loss": 3.6602,
+      "step": 645120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.942972608743792e-05,
+      "loss": 3.6576,
+      "step": 645632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.94213401399274e-05,
+      "loss": 3.6734,
+      "step": 646144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.941297057122061e-05,
+      "loss": 3.6709,
+      "step": 646656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.940458462371009e-05,
+      "loss": 3.6696,
+      "step": 647168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.939619867619957e-05,
+      "loss": 3.6726,
+      "step": 647680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.938781272868905e-05,
+      "loss": 3.6727,
+      "step": 648192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.937942678117853e-05,
+      "loss": 3.6833,
+      "step": 648704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.937104083366801e-05,
+      "loss": 3.6546,
+      "step": 649216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.936267126496122e-05,
+      "loss": 3.6806,
+      "step": 649728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.93542853174507e-05,
+      "loss": 3.664,
+      "step": 650240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.934589936994018e-05,
+      "loss": 3.6585,
+      "step": 650752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.933751342242966e-05,
+      "loss": 3.674,
+      "step": 651264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.932912747491914e-05,
+      "loss": 3.6703,
+      "step": 651776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9320757906212355e-05,
+      "loss": 3.6725,
+      "step": 652288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9312371958701835e-05,
+      "loss": 3.6778,
+      "step": 652800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9303986011191315e-05,
+      "loss": 3.655,
+      "step": 653312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9295600063680794e-05,
+      "loss": 3.6471,
+      "step": 653824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9287230494974004e-05,
+      "loss": 3.6641,
+      "step": 654336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9278844547463484e-05,
+      "loss": 3.6746,
+      "step": 654848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9270458599952964e-05,
+      "loss": 3.667,
+      "step": 655360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9262072652442443e-05,
+      "loss": 3.6634,
+      "step": 655872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.925370308373565e-05,
+      "loss": 3.6578,
+      "step": 656384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.924531713622513e-05,
+      "loss": 3.6612,
+      "step": 656896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.923693118871461e-05,
+      "loss": 3.666,
+      "step": 657408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.922854524120409e-05,
+      "loss": 3.6582,
+      "step": 657920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.922017567249731e-05,
+      "loss": 3.6668,
+      "step": 658432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.921178972498679e-05,
+      "loss": 3.6723,
+      "step": 658944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.920340377747627e-05,
+      "loss": 3.6704,
+      "step": 659456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.919501782996575e-05,
+      "loss": 3.6569,
+      "step": 659968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.918664826125896e-05,
+      "loss": 3.6627,
+      "step": 660480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.917826231374844e-05,
+      "loss": 3.6616,
+      "step": 660992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.916987636623792e-05,
+      "loss": 3.6662,
+      "step": 661504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.916149041872739e-05,
+      "loss": 3.6702,
+      "step": 662016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9153120850020606e-05,
+      "loss": 3.6643,
+      "step": 662528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9144734902510086e-05,
+      "loss": 3.6568,
+      "step": 663040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9136348954999566e-05,
+      "loss": 3.654,
+      "step": 663552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9127963007489046e-05,
+      "loss": 3.6619,
+      "step": 664064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.911959343878226e-05,
+      "loss": 3.6557,
+      "step": 664576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.911120749127174e-05,
+      "loss": 3.6436,
+      "step": 665088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.910282154376122e-05,
+      "loss": 3.6693,
+      "step": 665600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9094435596250695e-05,
+      "loss": 3.6618,
+      "step": 666112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.908606602754391e-05,
+      "loss": 3.6646,
+      "step": 666624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.907768008003339e-05,
+      "loss": 3.65,
+      "step": 667136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9069294132522864e-05,
+      "loss": 3.6682,
+      "step": 667648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9060908185012344e-05,
+      "loss": 3.6615,
+      "step": 668160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.905253861630556e-05,
+      "loss": 3.6488,
+      "step": 668672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.904415266879504e-05,
+      "loss": 3.6743,
+      "step": 669184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.903576672128452e-05,
+      "loss": 3.666,
+      "step": 669696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9027380773774e-05,
+      "loss": 3.6607,
+      "step": 670208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9019011205067216e-05,
+      "loss": 3.665,
+      "step": 670720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9010625257556696e-05,
+      "loss": 3.6487,
+      "step": 671232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900223931004617e-05,
+      "loss": 3.6665,
+      "step": 671744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.899385336253565e-05,
+      "loss": 3.656,
+      "step": 672256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8985483793828865e-05,
+      "loss": 3.6625,
+      "step": 672768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.897709784631834e-05,
+      "loss": 3.6552,
+      "step": 673280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.896871189880782e-05,
+      "loss": 3.6625,
+      "step": 673792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.89603259512973e-05,
+      "loss": 3.6548,
+      "step": 674304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8951956382590514e-05,
+      "loss": 3.6555,
+      "step": 674816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8943570435079994e-05,
+      "loss": 3.6537,
+      "step": 675328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8935184487569473e-05,
+      "loss": 3.6572,
+      "step": 675840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8926798540058953e-05,
+      "loss": 3.6574,
+      "step": 676352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.891842897135217e-05,
+      "loss": 3.6672,
+      "step": 676864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.891004302384164e-05,
+      "loss": 3.6548,
+      "step": 677376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.890165707633112e-05,
+      "loss": 3.6727,
+      "step": 677888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88932711288206e-05,
+      "loss": 3.6431,
+      "step": 678400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.888490156011381e-05,
+      "loss": 3.6677,
+      "step": 678912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.887651561260329e-05,
+      "loss": 3.6639,
+      "step": 679424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.886812966509277e-05,
+      "loss": 3.6611,
+      "step": 679936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.885974371758225e-05,
+      "loss": 3.656,
+      "step": 680448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.885135777007173e-05,
+      "loss": 3.6576,
+      "step": 680960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.884298820136494e-05,
+      "loss": 3.6604,
+      "step": 681472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.883460225385443e-05,
+      "loss": 3.6759,
+      "step": 681984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.882621630634391e-05,
+      "loss": 3.6468,
+      "step": 682496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.881783035883339e-05,
+      "loss": 3.6581,
+      "step": 683008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8809460790126596e-05,
+      "loss": 3.652,
+      "step": 683520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8801074842616076e-05,
+      "loss": 3.6605,
+      "step": 684032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8792688895105556e-05,
+      "loss": 3.6516,
+      "step": 684544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8784302947595036e-05,
+      "loss": 3.6596,
+      "step": 685056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8775933378888245e-05,
+      "loss": 3.6593,
+      "step": 685568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8767547431377725e-05,
+      "loss": 3.654,
+      "step": 686080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8759161483867205e-05,
+      "loss": 3.661,
+      "step": 686592
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.847093343734741,
+      "eval_runtime": 318.0685,
+      "eval_samples_per_second": 1199.713,
+      "eval_steps_per_second": 37.492,
+      "step": 686880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8750775536356685e-05,
+      "loss": 3.6495,
+      "step": 687104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8742405967649894e-05,
+      "loss": 3.649,
+      "step": 687616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.873402002013938e-05,
+      "loss": 3.6657,
+      "step": 688128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.872563407262886e-05,
+      "loss": 3.6611,
+      "step": 688640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.871724812511834e-05,
+      "loss": 3.6639,
+      "step": 689152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.870887855641155e-05,
+      "loss": 3.657,
+      "step": 689664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.870049260890103e-05,
+      "loss": 3.6498,
+      "step": 690176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.869210666139051e-05,
+      "loss": 3.6543,
+      "step": 690688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.868372071387999e-05,
+      "loss": 3.6527,
+      "step": 691200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.86753511451732e-05,
+      "loss": 3.665,
+      "step": 691712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.866696519766268e-05,
+      "loss": 3.659,
+      "step": 692224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.865857925015216e-05,
+      "loss": 3.6585,
+      "step": 692736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.865020968144537e-05,
+      "loss": 3.6599,
+      "step": 693248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.864182373393485e-05,
+      "loss": 3.6437,
+      "step": 693760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8633437786424334e-05,
+      "loss": 3.6596,
+      "step": 694272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8625051838913814e-05,
+      "loss": 3.6438,
+      "step": 694784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8616665891403294e-05,
+      "loss": 3.647,
+      "step": 695296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8608279943892774e-05,
+      "loss": 3.6519,
+      "step": 695808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8599893996382254e-05,
+      "loss": 3.648,
+      "step": 696320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8591508048871734e-05,
+      "loss": 3.6565,
+      "step": 696832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.858313848016494e-05,
+      "loss": 3.6675,
+      "step": 697344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.857475253265442e-05,
+      "loss": 3.658,
+      "step": 697856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.85663665851439e-05,
+      "loss": 3.6473,
+      "step": 698368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.855798063763338e-05,
+      "loss": 3.6494,
+      "step": 698880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.854961106892659e-05,
+      "loss": 3.6436,
+      "step": 699392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.854122512141607e-05,
+      "loss": 3.6494,
+      "step": 699904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.853283917390555e-05,
+      "loss": 3.643,
+      "step": 700416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.852445322639503e-05,
+      "loss": 3.6569,
+      "step": 700928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.851608365768825e-05,
+      "loss": 3.6451,
+      "step": 701440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.850769771017773e-05,
+      "loss": 3.6464,
+      "step": 701952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.849931176266721e-05,
+      "loss": 3.6467,
+      "step": 702464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.849092581515669e-05,
+      "loss": 3.6612,
+      "step": 702976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.84825562464499e-05,
+      "loss": 3.6501,
+      "step": 703488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.847417029893938e-05,
+      "loss": 3.6485,
+      "step": 704000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.846578435142886e-05,
+      "loss": 3.6496,
+      "step": 704512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.845739840391834e-05,
+      "loss": 3.6494,
+      "step": 705024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8449028835211546e-05,
+      "loss": 3.6527,
+      "step": 705536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8440642887701026e-05,
+      "loss": 3.6371,
+      "step": 706048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8432256940190506e-05,
+      "loss": 3.6369,
+      "step": 706560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8423870992679986e-05,
+      "loss": 3.6437,
+      "step": 707072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.84155014239732e-05,
+      "loss": 3.6412,
+      "step": 707584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.840711547646268e-05,
+      "loss": 3.6452,
+      "step": 708096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.839872952895216e-05,
+      "loss": 3.6478,
+      "step": 708608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.839034358144164e-05,
+      "loss": 3.648,
+      "step": 709120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.838195763393112e-05,
+      "loss": 3.6448,
+      "step": 709632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.837358806522433e-05,
+      "loss": 3.6467,
+      "step": 710144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.836520211771381e-05,
+      "loss": 3.6551,
+      "step": 710656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.835681617020329e-05,
+      "loss": 3.6422,
+      "step": 711168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.83484466014965e-05,
+      "loss": 3.6243,
+      "step": 711680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.834006065398598e-05,
+      "loss": 3.6573,
+      "step": 712192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.833167470647546e-05,
+      "loss": 3.6426,
+      "step": 712704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.832328875896494e-05,
+      "loss": 3.6479,
+      "step": 713216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.831490281145442e-05,
+      "loss": 3.6451,
+      "step": 713728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8306533242747635e-05,
+      "loss": 3.6365,
+      "step": 714240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8298147295237115e-05,
+      "loss": 3.6344,
+      "step": 714752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8289761347726595e-05,
+      "loss": 3.6401,
+      "step": 715264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.828137540021607e-05,
+      "loss": 3.6507,
+      "step": 715776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8273005831509284e-05,
+      "loss": 3.6429,
+      "step": 716288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8264619883998764e-05,
+      "loss": 3.6561,
+      "step": 716800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8256233936488244e-05,
+      "loss": 3.6418,
+      "step": 717312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.824784798897772e-05,
+      "loss": 3.63,
+      "step": 717824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.823947842027093e-05,
+      "loss": 3.6569,
+      "step": 718336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.823109247276041e-05,
+      "loss": 3.6294,
+      "step": 718848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.822270652524989e-05,
+      "loss": 3.6267,
+      "step": 719360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.821432057773937e-05,
+      "loss": 3.6421,
+      "step": 719872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.820595100903259e-05,
+      "loss": 3.6389,
+      "step": 720384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.819756506152207e-05,
+      "loss": 3.6313,
+      "step": 720896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.818917911401154e-05,
+      "loss": 3.6342,
+      "step": 721408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.818079316650102e-05,
+      "loss": 3.633,
+      "step": 721920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.817242359779424e-05,
+      "loss": 3.6464,
+      "step": 722432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.816403765028371e-05,
+      "loss": 3.6503,
+      "step": 722944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.815565170277319e-05,
+      "loss": 3.6439,
+      "step": 723456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.814726575526267e-05,
+      "loss": 3.6488,
+      "step": 723968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.813889618655589e-05,
+      "loss": 3.645,
+      "step": 724480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.813051023904537e-05,
+      "loss": 3.6589,
+      "step": 724992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.812212429153485e-05,
+      "loss": 3.634,
+      "step": 725504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.811373834402433e-05,
+      "loss": 3.655,
+      "step": 726016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.810536877531754e-05,
+      "loss": 3.6367,
+      "step": 726528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8096982827807016e-05,
+      "loss": 3.6375,
+      "step": 727040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8088596880296496e-05,
+      "loss": 3.6448,
+      "step": 727552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8080210932785976e-05,
+      "loss": 3.647,
+      "step": 728064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8071841364079185e-05,
+      "loss": 3.6509,
+      "step": 728576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8063455416568665e-05,
+      "loss": 3.6527,
+      "step": 729088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8055069469058145e-05,
+      "loss": 3.629,
+      "step": 729600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8046683521547625e-05,
+      "loss": 3.6259,
+      "step": 730112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8038313952840834e-05,
+      "loss": 3.6381,
+      "step": 730624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.802992800533032e-05,
+      "loss": 3.6476,
+      "step": 731136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.80215420578198e-05,
+      "loss": 3.6387,
+      "step": 731648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.801315611030928e-05,
+      "loss": 3.6383,
+      "step": 732160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.800478654160249e-05,
+      "loss": 3.6373,
+      "step": 732672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.799640059409197e-05,
+      "loss": 3.6385,
+      "step": 733184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.798801464658145e-05,
+      "loss": 3.6392,
+      "step": 733696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.797962869907093e-05,
+      "loss": 3.6367,
+      "step": 734208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.797125913036414e-05,
+      "loss": 3.6443,
+      "step": 734720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.796287318285362e-05,
+      "loss": 3.6472,
+      "step": 735232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.79544872353431e-05,
+      "loss": 3.6501,
+      "step": 735744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.794610128783258e-05,
+      "loss": 3.6328,
+      "step": 736256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.793773171912579e-05,
+      "loss": 3.6373,
+      "step": 736768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7929345771615274e-05,
+      "loss": 3.6377,
+      "step": 737280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7920959824104754e-05,
+      "loss": 3.6408,
+      "step": 737792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7912573876594234e-05,
+      "loss": 3.6448,
+      "step": 738304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.790420430788744e-05,
+      "loss": 3.6423,
+      "step": 738816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.789581836037692e-05,
+      "loss": 3.6328,
+      "step": 739328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.78874324128664e-05,
+      "loss": 3.6282,
+      "step": 739840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.787904646535588e-05,
+      "loss": 3.64,
+      "step": 740352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.787067689664909e-05,
+      "loss": 3.631,
+      "step": 740864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.786229094913857e-05,
+      "loss": 3.6175,
+      "step": 741376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.785390500162805e-05,
+      "loss": 3.6396,
+      "step": 741888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.784551905411753e-05,
+      "loss": 3.6459,
+      "step": 742400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.783714948541074e-05,
+      "loss": 3.6359,
+      "step": 742912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.782876353790023e-05,
+      "loss": 3.6289,
+      "step": 743424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.782037759038971e-05,
+      "loss": 3.6373,
+      "step": 743936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.781199164287919e-05,
+      "loss": 3.6377,
+      "step": 744448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.78036220741724e-05,
+      "loss": 3.6258,
+      "step": 744960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.779523612666188e-05,
+      "loss": 3.6523,
+      "step": 745472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.778685017915136e-05,
+      "loss": 3.6451,
+      "step": 745984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.777846423164084e-05,
+      "loss": 3.6331,
+      "step": 746496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7770094662934046e-05,
+      "loss": 3.6412,
+      "step": 747008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7761708715423526e-05,
+      "loss": 3.6288,
+      "step": 747520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7753322767913006e-05,
+      "loss": 3.6417,
+      "step": 748032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7744936820402486e-05,
+      "loss": 3.6324,
+      "step": 748544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7736567251695695e-05,
+      "loss": 3.6366,
+      "step": 749056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.772818130418518e-05,
+      "loss": 3.6309,
+      "step": 749568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.771979535667466e-05,
+      "loss": 3.6405,
+      "step": 750080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.771140940916414e-05,
+      "loss": 3.6334,
+      "step": 750592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.770303984045735e-05,
+      "loss": 3.6338,
+      "step": 751104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.769465389294683e-05,
+      "loss": 3.6286,
+      "step": 751616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.768626794543631e-05,
+      "loss": 3.6344,
+      "step": 752128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.767788199792579e-05,
+      "loss": 3.6319,
+      "step": 752640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7669512429219e-05,
+      "loss": 3.6432,
+      "step": 753152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.766112648170848e-05,
+      "loss": 3.6317,
+      "step": 753664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.765274053419796e-05,
+      "loss": 3.6463,
+      "step": 754176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.764435458668744e-05,
+      "loss": 3.6181,
+      "step": 754688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.763598501798065e-05,
+      "loss": 3.6457,
+      "step": 755200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7627599070470135e-05,
+      "loss": 3.6378,
+      "step": 755712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7619213122959615e-05,
+      "loss": 3.6378,
+      "step": 756224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7610827175449095e-05,
+      "loss": 3.6345,
+      "step": 756736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7602441227938575e-05,
+      "loss": 3.6335,
+      "step": 757248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7594071659231784e-05,
+      "loss": 3.6375,
+      "step": 757760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7585685711721264e-05,
+      "loss": 3.6563,
+      "step": 758272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7577299764210744e-05,
+      "loss": 3.6212,
+      "step": 758784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7568913816700224e-05,
+      "loss": 3.6362,
+      "step": 759296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.756054424799343e-05,
+      "loss": 3.6314,
+      "step": 759808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.755215830048291e-05,
+      "loss": 3.6348,
+      "step": 760320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.754377235297239e-05,
+      "loss": 3.6282,
+      "step": 760832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.753538640546187e-05,
+      "loss": 3.6384,
+      "step": 761344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.752701683675509e-05,
+      "loss": 3.635,
+      "step": 761856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.751863088924457e-05,
+      "loss": 3.635,
+      "step": 762368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.751024494173405e-05,
+      "loss": 3.6361,
+      "step": 762880
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8443191051483154,
+      "eval_runtime": 312.6916,
+      "eval_samples_per_second": 1220.343,
+      "eval_steps_per_second": 38.137,
+      "step": 763200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.750185899422353e-05,
+      "loss": 3.6257,
+      "step": 763392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.749347304671301e-05,
+      "loss": 3.6259,
+      "step": 763904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.748508709920249e-05,
+      "loss": 3.6394,
+      "step": 764416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.747670115169197e-05,
+      "loss": 3.6413,
+      "step": 764928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.746833158298518e-05,
+      "loss": 3.6397,
+      "step": 765440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.745994563547466e-05,
+      "loss": 3.6337,
+      "step": 765952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.745155968796414e-05,
+      "loss": 3.6284,
+      "step": 766464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.744317374045362e-05,
+      "loss": 3.6293,
+      "step": 766976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.743480417174683e-05,
+      "loss": 3.6308,
+      "step": 767488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.742641822423631e-05,
+      "loss": 3.6418,
+      "step": 768000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7418032276725787e-05,
+      "loss": 3.637,
+      "step": 768512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7409646329215267e-05,
+      "loss": 3.6367,
+      "step": 769024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.740129313931221e-05,
+      "loss": 3.6401,
+      "step": 769536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.739290719180169e-05,
+      "loss": 3.6183,
+      "step": 770048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.738452124429117e-05,
+      "loss": 3.6413,
+      "step": 770560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.737613529678065e-05,
+      "loss": 3.6186,
+      "step": 771072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.736774934927013e-05,
+      "loss": 3.624,
+      "step": 771584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.735936340175961e-05,
+      "loss": 3.6276,
+      "step": 772096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.735097745424909e-05,
+      "loss": 3.6241,
+      "step": 772608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7342591506738564e-05,
+      "loss": 3.633,
+      "step": 773120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.733422193803178e-05,
+      "loss": 3.641,
+      "step": 773632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.732583599052126e-05,
+      "loss": 3.6378,
+      "step": 774144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.731745004301074e-05,
+      "loss": 3.6261,
+      "step": 774656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.730906409550022e-05,
+      "loss": 3.623,
+      "step": 775168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7300694526793436e-05,
+      "loss": 3.6266,
+      "step": 775680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7292308579282916e-05,
+      "loss": 3.6245,
+      "step": 776192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.728392263177239e-05,
+      "loss": 3.62,
+      "step": 776704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.727553668426187e-05,
+      "loss": 3.6357,
+      "step": 777216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7267167115555085e-05,
+      "loss": 3.6189,
+      "step": 777728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7258781168044565e-05,
+      "loss": 3.6282,
+      "step": 778240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.725039522053404e-05,
+      "loss": 3.6204,
+      "step": 778752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.724200927302352e-05,
+      "loss": 3.6368,
+      "step": 779264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7233639704316734e-05,
+      "loss": 3.6313,
+      "step": 779776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7225253756806214e-05,
+      "loss": 3.6225,
+      "step": 780288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7216867809295694e-05,
+      "loss": 3.6312,
+      "step": 780800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7208481861785174e-05,
+      "loss": 3.6205,
+      "step": 781312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.720011229307839e-05,
+      "loss": 3.6337,
+      "step": 781824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.719172634556786e-05,
+      "loss": 3.6203,
+      "step": 782336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.718334039805734e-05,
+      "loss": 3.6099,
+      "step": 782848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.717495445054682e-05,
+      "loss": 3.6194,
+      "step": 783360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.716658488184004e-05,
+      "loss": 3.6246,
+      "step": 783872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.715819893432951e-05,
+      "loss": 3.6202,
+      "step": 784384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.714981298681899e-05,
+      "loss": 3.6234,
+      "step": 784896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.714142703930847e-05,
+      "loss": 3.6277,
+      "step": 785408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.713304109179795e-05,
+      "loss": 3.6224,
+      "step": 785920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.712467152309117e-05,
+      "loss": 3.6274,
+      "step": 786432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.711628557558065e-05,
+      "loss": 3.6363,
+      "step": 786944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.710789962807013e-05,
+      "loss": 3.6175,
+      "step": 787456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.709951368055961e-05,
+      "loss": 3.5986,
+      "step": 787968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7091144111852817e-05,
+      "loss": 3.6342,
+      "step": 788480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7082758164342297e-05,
+      "loss": 3.6196,
+      "step": 788992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7074372216831777e-05,
+      "loss": 3.6249,
+      "step": 789504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7065986269321256e-05,
+      "loss": 3.6212,
+      "step": 790016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7057616700614466e-05,
+      "loss": 3.6178,
+      "step": 790528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7049230753103946e-05,
+      "loss": 3.6116,
+      "step": 791040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7040844805593425e-05,
+      "loss": 3.6144,
+      "step": 791552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7032458858082905e-05,
+      "loss": 3.6321,
+      "step": 792064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.702408928937612e-05,
+      "loss": 3.6164,
+      "step": 792576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.70157033418656e-05,
+      "loss": 3.6375,
+      "step": 793088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.700731739435508e-05,
+      "loss": 3.619,
+      "step": 793600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.699893144684456e-05,
+      "loss": 3.6056,
+      "step": 794112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.699056187813777e-05,
+      "loss": 3.6324,
+      "step": 794624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.698217593062725e-05,
+      "loss": 3.615,
+      "step": 795136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.697378998311673e-05,
+      "loss": 3.6002,
+      "step": 795648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.696540403560621e-05,
+      "loss": 3.6189,
+      "step": 796160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.695703446689942e-05,
+      "loss": 3.6171,
+      "step": 796672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.69486485193889e-05,
+      "loss": 3.61,
+      "step": 797184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.694026257187838e-05,
+      "loss": 3.6106,
+      "step": 797696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.693187662436786e-05,
+      "loss": 3.6115,
+      "step": 798208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6923507055661075e-05,
+      "loss": 3.623,
+      "step": 798720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6915121108150555e-05,
+      "loss": 3.6285,
+      "step": 799232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6906735160640035e-05,
+      "loss": 3.6174,
+      "step": 799744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6898349213129515e-05,
+      "loss": 3.6278,
+      "step": 800256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6889979644422724e-05,
+      "loss": 3.6244,
+      "step": 800768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6881593696912204e-05,
+      "loss": 3.6355,
+      "step": 801280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6873207749401684e-05,
+      "loss": 3.6131,
+      "step": 801792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6864821801891164e-05,
+      "loss": 3.6351,
+      "step": 802304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.685645223318437e-05,
+      "loss": 3.6144,
+      "step": 802816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.684806628567385e-05,
+      "loss": 3.6143,
+      "step": 803328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.683968033816333e-05,
+      "loss": 3.6259,
+      "step": 803840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.683129439065281e-05,
+      "loss": 3.6198,
+      "step": 804352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.682292482194603e-05,
+      "loss": 3.6308,
+      "step": 804864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.681453887443551e-05,
+      "loss": 3.633,
+      "step": 805376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.680615292692499e-05,
+      "loss": 3.6092,
+      "step": 805888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.679776697941447e-05,
+      "loss": 3.6034,
+      "step": 806400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.678938103190395e-05,
+      "loss": 3.6158,
+      "step": 806912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.678101146319716e-05,
+      "loss": 3.6308,
+      "step": 807424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.677262551568664e-05,
+      "loss": 3.6133,
+      "step": 807936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.676423956817612e-05,
+      "loss": 3.6177,
+      "step": 808448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.67558536206656e-05,
+      "loss": 3.6191,
+      "step": 808960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6747484051958807e-05,
+      "loss": 3.6114,
+      "step": 809472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6739098104448286e-05,
+      "loss": 3.6196,
+      "step": 809984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6730712156937766e-05,
+      "loss": 3.6125,
+      "step": 810496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6722326209427246e-05,
+      "loss": 3.6213,
+      "step": 811008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.671395664072046e-05,
+      "loss": 3.6217,
+      "step": 811520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.670557069320994e-05,
+      "loss": 3.6318,
+      "step": 812032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.669718474569942e-05,
+      "loss": 3.6105,
+      "step": 812544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66887987981889e-05,
+      "loss": 3.6197,
+      "step": 813056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.668042922948211e-05,
+      "loss": 3.6134,
+      "step": 813568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.667204328197159e-05,
+      "loss": 3.6208,
+      "step": 814080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.666365733446107e-05,
+      "loss": 3.6251,
+      "step": 814592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.665527138695055e-05,
+      "loss": 3.6169,
+      "step": 815104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.664690181824376e-05,
+      "loss": 3.6147,
+      "step": 815616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.663851587073324e-05,
+      "loss": 3.6079,
+      "step": 816128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.663012992322272e-05,
+      "loss": 3.6155,
+      "step": 816640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66217439757122e-05,
+      "loss": 3.6131,
+      "step": 817152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6613374407005416e-05,
+      "loss": 3.5955,
+      "step": 817664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6604988459494896e-05,
+      "loss": 3.6198,
+      "step": 818176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6596602511984376e-05,
+      "loss": 3.6194,
+      "step": 818688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6588216564473856e-05,
+      "loss": 3.6133,
+      "step": 819200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6579846995767065e-05,
+      "loss": 3.6095,
+      "step": 819712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6571461048256545e-05,
+      "loss": 3.6169,
+      "step": 820224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6563075100746025e-05,
+      "loss": 3.6176,
+      "step": 820736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6554689153235505e-05,
+      "loss": 3.6035,
+      "step": 821248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6546319584528714e-05,
+      "loss": 3.6322,
+      "step": 821760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6537933637018194e-05,
+      "loss": 3.6209,
+      "step": 822272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6529547689507674e-05,
+      "loss": 3.6117,
+      "step": 822784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6521161741997154e-05,
+      "loss": 3.62,
+      "step": 823296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.651279217329037e-05,
+      "loss": 3.6078,
+      "step": 823808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.650440622577985e-05,
+      "loss": 3.6186,
+      "step": 824320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.649602027826933e-05,
+      "loss": 3.6132,
+      "step": 824832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.648763433075881e-05,
+      "loss": 3.6129,
+      "step": 825344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.647926476205202e-05,
+      "loss": 3.6139,
+      "step": 825856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.64708788145415e-05,
+      "loss": 3.6155,
+      "step": 826368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.646249286703098e-05,
+      "loss": 3.6156,
+      "step": 826880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.645410691952046e-05,
+      "loss": 3.6097,
+      "step": 827392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.644573735081367e-05,
+      "loss": 3.6112,
+      "step": 827904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.643735140330315e-05,
+      "loss": 3.6072,
+      "step": 828416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.642896545579263e-05,
+      "loss": 3.6108,
+      "step": 828928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.642057950828211e-05,
+      "loss": 3.6237,
+      "step": 829440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.641219356077159e-05,
+      "loss": 3.6133,
+      "step": 829952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.64038239920648e-05,
+      "loss": 3.6243,
+      "step": 830464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.639543804455428e-05,
+      "loss": 3.5946,
+      "step": 830976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.638705209704376e-05,
+      "loss": 3.6277,
+      "step": 831488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6378666149533236e-05,
+      "loss": 3.6151,
+      "step": 832000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.637029658082645e-05,
+      "loss": 3.615,
+      "step": 832512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.636191063331593e-05,
+      "loss": 3.6137,
+      "step": 833024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.635352468580541e-05,
+      "loss": 3.6076,
+      "step": 833536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.634515511709862e-05,
+      "loss": 3.6169,
+      "step": 834048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.63367691695881e-05,
+      "loss": 3.6318,
+      "step": 834560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.632838322207758e-05,
+      "loss": 3.6034,
+      "step": 835072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.631999727456706e-05,
+      "loss": 3.6162,
+      "step": 835584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.631161132705654e-05,
+      "loss": 3.6094,
+      "step": 836096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.630324175834976e-05,
+      "loss": 3.6116,
+      "step": 836608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.629485581083924e-05,
+      "loss": 3.6088,
+      "step": 837120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.628646986332871e-05,
+      "loss": 3.6163,
+      "step": 837632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.627808391581819e-05,
+      "loss": 3.6134,
+      "step": 838144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6269714347111406e-05,
+      "loss": 3.6173,
+      "step": 838656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6261328399600886e-05,
+      "loss": 3.6111,
+      "step": 839168
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8425068855285645,
+      "eval_runtime": 306.2789,
+      "eval_samples_per_second": 1245.894,
+      "eval_steps_per_second": 38.935,
+      "step": 839520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.625294245209036e-05,
+      "loss": 3.6066,
+      "step": 839680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.624455650457984e-05,
+      "loss": 3.6043,
+      "step": 840192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6236186935873055e-05,
+      "loss": 3.6181,
+      "step": 840704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.622780098836253e-05,
+      "loss": 3.6194,
+      "step": 841216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6219415040852015e-05,
+      "loss": 3.6211,
+      "step": 841728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6211029093341495e-05,
+      "loss": 3.6124,
+      "step": 842240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.620265952463471e-05,
+      "loss": 3.6088,
+      "step": 842752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6194273577124184e-05,
+      "loss": 3.6137,
+      "step": 843264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6185887629613664e-05,
+      "loss": 3.6068,
+      "step": 843776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6177501682103144e-05,
+      "loss": 3.617,
+      "step": 844288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.616913211339636e-05,
+      "loss": 3.617,
+      "step": 844800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.616074616588583e-05,
+      "loss": 3.6187,
+      "step": 845312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.615237659717905e-05,
+      "loss": 3.6177,
+      "step": 845824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.614399064966853e-05,
+      "loss": 3.5988,
+      "step": 846336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6135604702158e-05,
+      "loss": 3.6194,
+      "step": 846848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.612721875464748e-05,
+      "loss": 3.5956,
+      "step": 847360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.611883280713697e-05,
+      "loss": 3.6057,
+      "step": 847872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.611044685962645e-05,
+      "loss": 3.6072,
+      "step": 848384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.610206091211593e-05,
+      "loss": 3.6011,
+      "step": 848896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.609367496460541e-05,
+      "loss": 3.6095,
+      "step": 849408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.608530539589862e-05,
+      "loss": 3.623,
+      "step": 849920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.60769194483881e-05,
+      "loss": 3.6107,
+      "step": 850432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.606853350087758e-05,
+      "loss": 3.6111,
+      "step": 850944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.606014755336706e-05,
+      "loss": 3.6054,
+      "step": 851456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6051777984660266e-05,
+      "loss": 3.6049,
+      "step": 851968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6043392037149746e-05,
+      "loss": 3.6017,
+      "step": 852480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6035006089639226e-05,
+      "loss": 3.5969,
+      "step": 852992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6026620142128706e-05,
+      "loss": 3.6146,
+      "step": 853504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.601825057342192e-05,
+      "loss": 3.6012,
+      "step": 854016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.60098646259114e-05,
+      "loss": 3.6047,
+      "step": 854528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.600147867840088e-05,
+      "loss": 3.6003,
+      "step": 855040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.599309273089036e-05,
+      "loss": 3.6166,
+      "step": 855552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.598472316218357e-05,
+      "loss": 3.6038,
+      "step": 856064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.597633721467305e-05,
+      "loss": 3.6074,
+      "step": 856576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.596795126716253e-05,
+      "loss": 3.6088,
+      "step": 857088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.595956531965201e-05,
+      "loss": 3.6027,
+      "step": 857600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.595119575094522e-05,
+      "loss": 3.6159,
+      "step": 858112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.59428098034347e-05,
+      "loss": 3.5947,
+      "step": 858624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.593442385592418e-05,
+      "loss": 3.5919,
+      "step": 859136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.592603790841366e-05,
+      "loss": 3.5989,
+      "step": 859648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5917668339706876e-05,
+      "loss": 3.6048,
+      "step": 860160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5909282392196356e-05,
+      "loss": 3.5969,
+      "step": 860672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5900896444685836e-05,
+      "loss": 3.6038,
+      "step": 861184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5892510497175316e-05,
+      "loss": 3.6048,
+      "step": 861696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5884124549664796e-05,
+      "loss": 3.6015,
+      "step": 862208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5875754980958005e-05,
+      "loss": 3.6055,
+      "step": 862720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5867369033447485e-05,
+      "loss": 3.6116,
+      "step": 863232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5858983085936965e-05,
+      "loss": 3.6014,
+      "step": 863744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5850597138426445e-05,
+      "loss": 3.5814,
+      "step": 864256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5842227569719654e-05,
+      "loss": 3.6139,
+      "step": 864768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5833841622209134e-05,
+      "loss": 3.6,
+      "step": 865280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5825455674698614e-05,
+      "loss": 3.6036,
+      "step": 865792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5817069727188093e-05,
+      "loss": 3.5996,
+      "step": 866304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.580870015848131e-05,
+      "loss": 3.5976,
+      "step": 866816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.580031421097079e-05,
+      "loss": 3.592,
+      "step": 867328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.579192826346027e-05,
+      "loss": 3.5943,
+      "step": 867840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.578354231594975e-05,
+      "loss": 3.6161,
+      "step": 868352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.577517274724296e-05,
+      "loss": 3.5961,
+      "step": 868864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.576678679973244e-05,
+      "loss": 3.6164,
+      "step": 869376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.575840085222192e-05,
+      "loss": 3.6007,
+      "step": 869888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.57500149047114e-05,
+      "loss": 3.5825,
+      "step": 870400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.574164533600461e-05,
+      "loss": 3.6122,
+      "step": 870912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.573325938849409e-05,
+      "loss": 3.5947,
+      "step": 871424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.572487344098357e-05,
+      "loss": 3.5842,
+      "step": 871936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.571648749347305e-05,
+      "loss": 3.5981,
+      "step": 872448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.570811792476626e-05,
+      "loss": 3.5951,
+      "step": 872960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.569973197725574e-05,
+      "loss": 3.5893,
+      "step": 873472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.569134602974522e-05,
+      "loss": 3.5953,
+      "step": 873984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.56829600822347e-05,
+      "loss": 3.5937,
+      "step": 874496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.567459051352791e-05,
+      "loss": 3.5993,
+      "step": 875008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.566620456601739e-05,
+      "loss": 3.6088,
+      "step": 875520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.565781861850687e-05,
+      "loss": 3.5987,
+      "step": 876032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.564943267099635e-05,
+      "loss": 3.6049,
+      "step": 876544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.564106310228956e-05,
+      "loss": 3.6029,
+      "step": 877056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.563267715477904e-05,
+      "loss": 3.6168,
+      "step": 877568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.562429120726852e-05,
+      "loss": 3.5953,
+      "step": 878080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5615905259758e-05,
+      "loss": 3.6154,
+      "step": 878592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.560753569105122e-05,
+      "loss": 3.594,
+      "step": 879104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.55991497435407e-05,
+      "loss": 3.5947,
+      "step": 879616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5590763796030177e-05,
+      "loss": 3.6025,
+      "step": 880128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5582377848519657e-05,
+      "loss": 3.6004,
+      "step": 880640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5574008279812866e-05,
+      "loss": 3.6101,
+      "step": 881152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5565622332302346e-05,
+      "loss": 3.6143,
+      "step": 881664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5557236384791826e-05,
+      "loss": 3.5935,
+      "step": 882176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5548850437281306e-05,
+      "loss": 3.5823,
+      "step": 882688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5540480868574515e-05,
+      "loss": 3.5913,
+      "step": 883200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5532094921063995e-05,
+      "loss": 3.6128,
+      "step": 883712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5523708973553475e-05,
+      "loss": 3.5921,
+      "step": 884224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5515323026042954e-05,
+      "loss": 3.6012,
+      "step": 884736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.550695345733617e-05,
+      "loss": 3.5995,
+      "step": 885248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.549856750982565e-05,
+      "loss": 3.5906,
+      "step": 885760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.549018156231513e-05,
+      "loss": 3.6002,
+      "step": 886272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.548179561480461e-05,
+      "loss": 3.5952,
+      "step": 886784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.547342604609782e-05,
+      "loss": 3.5989,
+      "step": 887296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.54650400985873e-05,
+      "loss": 3.6075,
+      "step": 887808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.545665415107678e-05,
+      "loss": 3.609,
+      "step": 888320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.544826820356626e-05,
+      "loss": 3.5926,
+      "step": 888832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.543989863485947e-05,
+      "loss": 3.5992,
+      "step": 889344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.543151268734895e-05,
+      "loss": 3.5943,
+      "step": 889856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.542312673983843e-05,
+      "loss": 3.6011,
+      "step": 890368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.541474079232791e-05,
+      "loss": 3.6035,
+      "step": 890880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5406371223621124e-05,
+      "loss": 3.5964,
+      "step": 891392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5397985276110604e-05,
+      "loss": 3.593,
+      "step": 891904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5389599328600084e-05,
+      "loss": 3.5871,
+      "step": 892416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.538121338108956e-05,
+      "loss": 3.5986,
+      "step": 892928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.537284381238277e-05,
+      "loss": 3.5917,
+      "step": 893440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.536445786487225e-05,
+      "loss": 3.5761,
+      "step": 893952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.535607191736173e-05,
+      "loss": 3.6001,
+      "step": 894464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5347685969851206e-05,
+      "loss": 3.599,
+      "step": 894976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.533931640114442e-05,
+      "loss": 3.593,
+      "step": 895488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.53309304536339e-05,
+      "loss": 3.5892,
+      "step": 896000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.532254450612338e-05,
+      "loss": 3.5944,
+      "step": 896512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.531415855861286e-05,
+      "loss": 3.5983,
+      "step": 897024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.530578898990608e-05,
+      "loss": 3.5844,
+      "step": 897536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.529740304239556e-05,
+      "loss": 3.6092,
+      "step": 898048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.528901709488503e-05,
+      "loss": 3.6001,
+      "step": 898560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.528063114737451e-05,
+      "loss": 3.5947,
+      "step": 899072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.527226157866773e-05,
+      "loss": 3.6004,
+      "step": 899584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5263875631157207e-05,
+      "loss": 3.5901,
+      "step": 900096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.525548968364668e-05,
+      "loss": 3.5986,
+      "step": 900608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.524710373613616e-05,
+      "loss": 3.5941,
+      "step": 901120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5238734167429376e-05,
+      "loss": 3.5926,
+      "step": 901632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5230348219918856e-05,
+      "loss": 3.5949,
+      "step": 902144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.522196227240833e-05,
+      "loss": 3.5955,
+      "step": 902656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5213576324897815e-05,
+      "loss": 3.596,
+      "step": 903168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.520520675619103e-05,
+      "loss": 3.5899,
+      "step": 903680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5196820808680505e-05,
+      "loss": 3.5893,
+      "step": 904192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5188434861169984e-05,
+      "loss": 3.5909,
+      "step": 904704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5180048913659464e-05,
+      "loss": 3.5875,
+      "step": 905216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.517167934495268e-05,
+      "loss": 3.6068,
+      "step": 905728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5163293397442153e-05,
+      "loss": 3.5909,
+      "step": 906240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5154907449931633e-05,
+      "loss": 3.6078,
+      "step": 906752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5146521502421113e-05,
+      "loss": 3.5752,
+      "step": 907264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.513815193371433e-05,
+      "loss": 3.6066,
+      "step": 907776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.51297659862038e-05,
+      "loss": 3.5928,
+      "step": 908288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.512138003869328e-05,
+      "loss": 3.6023,
+      "step": 908800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.511299409118277e-05,
+      "loss": 3.5941,
+      "step": 909312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.510462452247598e-05,
+      "loss": 3.5854,
+      "step": 909824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.509623857496546e-05,
+      "loss": 3.6011,
+      "step": 910336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.508785262745494e-05,
+      "loss": 3.6176,
+      "step": 910848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.507946667994442e-05,
+      "loss": 3.5787,
+      "step": 911360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.507109711123763e-05,
+      "loss": 3.5971,
+      "step": 911872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.506271116372711e-05,
+      "loss": 3.591,
+      "step": 912384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.505432521621659e-05,
+      "loss": 3.5917,
+      "step": 912896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.504593926870607e-05,
+      "loss": 3.5892,
+      "step": 913408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5037569699999276e-05,
+      "loss": 3.5979,
+      "step": 913920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5029183752488756e-05,
+      "loss": 3.5904,
+      "step": 914432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5020797804978236e-05,
+      "loss": 3.598,
+      "step": 914944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.501241185746772e-05,
+      "loss": 3.5927,
+      "step": 915456
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8418779373168945,
+      "eval_runtime": 305.3947,
+      "eval_samples_per_second": 1249.501,
+      "eval_steps_per_second": 39.048,
+      "step": 915840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.500404228876093e-05,
+      "loss": 3.5967,
+      "step": 915968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.499565634125041e-05,
+      "loss": 3.5825,
+      "step": 916480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.498727039373989e-05,
+      "loss": 3.5997,
+      "step": 916992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.497888444622937e-05,
+      "loss": 3.6027,
+      "step": 917504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.497051487752258e-05,
+      "loss": 3.5968,
+      "step": 918016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.496212893001206e-05,
+      "loss": 3.5932,
+      "step": 918528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.495374298250154e-05,
+      "loss": 3.5965,
+      "step": 919040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.494535703499102e-05,
+      "loss": 3.591,
+      "step": 919552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.493698746628423e-05,
+      "loss": 3.5887,
+      "step": 920064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.492860151877371e-05,
+      "loss": 3.5941,
+      "step": 920576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.492021557126319e-05,
+      "loss": 3.6014,
+      "step": 921088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4911829623752676e-05,
+      "loss": 3.5981,
+      "step": 921600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.490347643384962e-05,
+      "loss": 3.6008,
+      "step": 922112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.48950904863391e-05,
+      "loss": 3.5799,
+      "step": 922624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4886704538828575e-05,
+      "loss": 3.5985,
+      "step": 923136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4878318591318055e-05,
+      "loss": 3.5777,
+      "step": 923648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4869932643807535e-05,
+      "loss": 3.5895,
+      "step": 924160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4861546696297014e-05,
+      "loss": 3.5852,
+      "step": 924672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4853160748786494e-05,
+      "loss": 3.5823,
+      "step": 925184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4844774801275974e-05,
+      "loss": 3.5895,
+      "step": 925696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4836405232569184e-05,
+      "loss": 3.6053,
+      "step": 926208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4828019285058663e-05,
+      "loss": 3.5939,
+      "step": 926720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4819633337548143e-05,
+      "loss": 3.5912,
+      "step": 927232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.481124739003763e-05,
+      "loss": 3.5861,
+      "step": 927744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.480287782133084e-05,
+      "loss": 3.5916,
+      "step": 928256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.479449187382032e-05,
+      "loss": 3.5839,
+      "step": 928768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.47861059263098e-05,
+      "loss": 3.576,
+      "step": 929280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.477771997879928e-05,
+      "loss": 3.5943,
+      "step": 929792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.476935041009249e-05,
+      "loss": 3.5801,
+      "step": 930304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.476096446258197e-05,
+      "loss": 3.585,
+      "step": 930816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.475257851507145e-05,
+      "loss": 3.5869,
+      "step": 931328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.474419256756093e-05,
+      "loss": 3.5931,
+      "step": 931840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.473582299885414e-05,
+      "loss": 3.5848,
+      "step": 932352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.472743705134362e-05,
+      "loss": 3.5887,
+      "step": 932864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.47190511038331e-05,
+      "loss": 3.5888,
+      "step": 933376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.471066515632258e-05,
+      "loss": 3.5853,
+      "step": 933888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.470229558761579e-05,
+      "loss": 3.5974,
+      "step": 934400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.469390964010527e-05,
+      "loss": 3.5741,
+      "step": 934912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.468552369259475e-05,
+      "loss": 3.5771,
+      "step": 935424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.467713774508423e-05,
+      "loss": 3.5796,
+      "step": 935936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.466876817637744e-05,
+      "loss": 3.5828,
+      "step": 936448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.466038222886692e-05,
+      "loss": 3.5787,
+      "step": 936960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.46519962813564e-05,
+      "loss": 3.5864,
+      "step": 937472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.464361033384588e-05,
+      "loss": 3.5889,
+      "step": 937984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.463524076513909e-05,
+      "loss": 3.5819,
+      "step": 938496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.462685481762857e-05,
+      "loss": 3.5837,
+      "step": 939008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.461846887011805e-05,
+      "loss": 3.5938,
+      "step": 939520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.461008292260753e-05,
+      "loss": 3.5803,
+      "step": 940032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4601713353900747e-05,
+      "loss": 3.5637,
+      "step": 940544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4593327406390227e-05,
+      "loss": 3.5949,
+      "step": 941056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4584941458879706e-05,
+      "loss": 3.5842,
+      "step": 941568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4576555511369186e-05,
+      "loss": 3.5824,
+      "step": 942080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4568202321466125e-05,
+      "loss": 3.5841,
+      "step": 942592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4559816373955605e-05,
+      "loss": 3.5781,
+      "step": 943104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4551430426445085e-05,
+      "loss": 3.5731,
+      "step": 943616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4543044478934565e-05,
+      "loss": 3.5758,
+      "step": 944128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4534658531424045e-05,
+      "loss": 3.5962,
+      "step": 944640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4526272583913524e-05,
+      "loss": 3.5766,
+      "step": 945152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4517886636403004e-05,
+      "loss": 3.6001,
+      "step": 945664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4509500688892484e-05,
+      "loss": 3.5822,
+      "step": 946176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.45011311201857e-05,
+      "loss": 3.563,
+      "step": 946688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.449274517267518e-05,
+      "loss": 3.5914,
+      "step": 947200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.448435922516466e-05,
+      "loss": 3.5768,
+      "step": 947712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.447597327765414e-05,
+      "loss": 3.5689,
+      "step": 948224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.446760370894735e-05,
+      "loss": 3.5763,
+      "step": 948736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.445921776143683e-05,
+      "loss": 3.5786,
+      "step": 949248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.445083181392631e-05,
+      "loss": 3.5722,
+      "step": 949760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.444244586641579e-05,
+      "loss": 3.5785,
+      "step": 950272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4434076297709e-05,
+      "loss": 3.5692,
+      "step": 950784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.442569035019848e-05,
+      "loss": 3.5851,
+      "step": 951296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.441730440268796e-05,
+      "loss": 3.5899,
+      "step": 951808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.440891845517744e-05,
+      "loss": 3.5792,
+      "step": 952320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4400548886470654e-05,
+      "loss": 3.5875,
+      "step": 952832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4392162938960134e-05,
+      "loss": 3.5846,
+      "step": 953344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4383776991449614e-05,
+      "loss": 3.5978,
+      "step": 953856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4375391043939094e-05,
+      "loss": 3.5793,
+      "step": 954368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.43670214752323e-05,
+      "loss": 3.5905,
+      "step": 954880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.435863552772178e-05,
+      "loss": 3.5816,
+      "step": 955392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.435024958021126e-05,
+      "loss": 3.5763,
+      "step": 955904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.434186363270074e-05,
+      "loss": 3.5881,
+      "step": 956416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.433349406399395e-05,
+      "loss": 3.5776,
+      "step": 956928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.432510811648343e-05,
+      "loss": 3.5901,
+      "step": 957440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.431672216897291e-05,
+      "loss": 3.5956,
+      "step": 957952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.430833622146239e-05,
+      "loss": 3.5787,
+      "step": 958464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.429996665275561e-05,
+      "loss": 3.5587,
+      "step": 958976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.429158070524509e-05,
+      "loss": 3.5734,
+      "step": 959488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.428319475773457e-05,
+      "loss": 3.5892,
+      "step": 960000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.427480881022405e-05,
+      "loss": 3.5767,
+      "step": 960512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4266439241517257e-05,
+      "loss": 3.5814,
+      "step": 961024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4258053294006736e-05,
+      "loss": 3.5839,
+      "step": 961536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4249667346496216e-05,
+      "loss": 3.572,
+      "step": 962048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4241281398985696e-05,
+      "loss": 3.584,
+      "step": 962560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4232911830278906e-05,
+      "loss": 3.5733,
+      "step": 963072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4224525882768385e-05,
+      "loss": 3.5849,
+      "step": 963584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4216139935257865e-05,
+      "loss": 3.586,
+      "step": 964096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4207753987747345e-05,
+      "loss": 3.5904,
+      "step": 964608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.419938441904056e-05,
+      "loss": 3.5731,
+      "step": 965120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.419099847153004e-05,
+      "loss": 3.5809,
+      "step": 965632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.418261252401952e-05,
+      "loss": 3.5812,
+      "step": 966144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4174226576509e-05,
+      "loss": 3.5797,
+      "step": 966656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.416585700780221e-05,
+      "loss": 3.5871,
+      "step": 967168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.415747106029169e-05,
+      "loss": 3.5769,
+      "step": 967680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.414908511278117e-05,
+      "loss": 3.5761,
+      "step": 968192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.414069916527065e-05,
+      "loss": 3.5734,
+      "step": 968704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.413232959656386e-05,
+      "loss": 3.5789,
+      "step": 969216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.412394364905334e-05,
+      "loss": 3.5733,
+      "step": 969728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.411555770154282e-05,
+      "loss": 3.5605,
+      "step": 970240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.41071717540323e-05,
+      "loss": 3.58,
+      "step": 970752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4098802185325515e-05,
+      "loss": 3.584,
+      "step": 971264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4090416237814995e-05,
+      "loss": 3.5703,
+      "step": 971776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4082030290304475e-05,
+      "loss": 3.5765,
+      "step": 972288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.407364434279395e-05,
+      "loss": 3.5741,
+      "step": 972800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4065274774087164e-05,
+      "loss": 3.5839,
+      "step": 973312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4056888826576644e-05,
+      "loss": 3.5699,
+      "step": 973824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4048502879066124e-05,
+      "loss": 3.5843,
+      "step": 974336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.40401169315556e-05,
+      "loss": 3.5801,
+      "step": 974848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.403174736284881e-05,
+      "loss": 3.5832,
+      "step": 975360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.402336141533829e-05,
+      "loss": 3.5815,
+      "step": 975872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.401497546782777e-05,
+      "loss": 3.5735,
+      "step": 976384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.400658952031725e-05,
+      "loss": 3.5784,
+      "step": 976896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.399821995161047e-05,
+      "loss": 3.578,
+      "step": 977408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.398983400409995e-05,
+      "loss": 3.5778,
+      "step": 977920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.398144805658942e-05,
+      "loss": 3.5749,
+      "step": 978432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.39730621090789e-05,
+      "loss": 3.5748,
+      "step": 978944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.396469254037212e-05,
+      "loss": 3.58,
+      "step": 979456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.39563065928616e-05,
+      "loss": 3.5723,
+      "step": 979968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.394792064535107e-05,
+      "loss": 3.5727,
+      "step": 980480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.393953469784055e-05,
+      "loss": 3.5678,
+      "step": 980992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3931165129133767e-05,
+      "loss": 3.5697,
+      "step": 981504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3922779181623246e-05,
+      "loss": 3.5876,
+      "step": 982016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3914393234112726e-05,
+      "loss": 3.5744,
+      "step": 982528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3906007286602206e-05,
+      "loss": 3.59,
+      "step": 983040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.389763771789542e-05,
+      "loss": 3.5604,
+      "step": 983552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3889251770384895e-05,
+      "loss": 3.5836,
+      "step": 984064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3880865822874375e-05,
+      "loss": 3.5806,
+      "step": 984576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3872479875363855e-05,
+      "loss": 3.5811,
+      "step": 985088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.386411030665707e-05,
+      "loss": 3.5766,
+      "step": 985600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3855724359146544e-05,
+      "loss": 3.567,
+      "step": 986112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3847338411636024e-05,
+      "loss": 3.5824,
+      "step": 986624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3838952464125504e-05,
+      "loss": 3.5944,
+      "step": 987136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3830582895418713e-05,
+      "loss": 3.5646,
+      "step": 987648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.38221969479082e-05,
+      "loss": 3.5777,
+      "step": 988160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.381381100039768e-05,
+      "loss": 3.5746,
+      "step": 988672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.380542505288716e-05,
+      "loss": 3.5732,
+      "step": 989184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.379705548418037e-05,
+      "loss": 3.5754,
+      "step": 989696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.378866953666985e-05,
+      "loss": 3.5779,
+      "step": 990208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.378028358915933e-05,
+      "loss": 3.5701,
+      "step": 990720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.377189764164881e-05,
+      "loss": 3.5803,
+      "step": 991232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.376352807294202e-05,
+      "loss": 3.5752,
+      "step": 991744
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8422465324401855,
+      "eval_runtime": 306.4284,
+      "eval_samples_per_second": 1245.286,
+      "eval_steps_per_second": 38.916,
+      "step": 992160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.37551421254315e-05,
+      "loss": 3.5804,
+      "step": 992256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.374675617792098e-05,
+      "loss": 3.5638,
+      "step": 992768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.373837023041046e-05,
+      "loss": 3.5821,
+      "step": 993280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.372998428289994e-05,
+      "loss": 3.5867,
+      "step": 993792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3721614714193154e-05,
+      "loss": 3.5768,
+      "step": 994304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3713228766682634e-05,
+      "loss": 3.5777,
+      "step": 994816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3704842819172114e-05,
+      "loss": 3.5759,
+      "step": 995328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3696456871661594e-05,
+      "loss": 3.5776,
+      "step": 995840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.36880873029548e-05,
+      "loss": 3.5657,
+      "step": 996352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.367970135544428e-05,
+      "loss": 3.5817,
+      "step": 996864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.367131540793376e-05,
+      "loss": 3.5835,
+      "step": 997376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.366292946042324e-05,
+      "loss": 3.5779,
+      "step": 997888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.365457627052019e-05,
+      "loss": 3.5887,
+      "step": 998400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.364619032300966e-05,
+      "loss": 3.5592,
+      "step": 998912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.363780437549914e-05,
+      "loss": 3.5756,
+      "step": 999424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.362941842798862e-05,
+      "loss": 3.5636,
+      "step": 999936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.36210324804781e-05,
+      "loss": 3.5728,
+      "step": 1000448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.361264653296759e-05,
+      "loss": 3.57,
+      "step": 1000960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.360426058545707e-05,
+      "loss": 3.565,
+      "step": 1001472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.359587463794655e-05,
+      "loss": 3.5643,
+      "step": 1001984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3587505069239756e-05,
+      "loss": 3.593,
+      "step": 1002496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3579119121729236e-05,
+      "loss": 3.5763,
+      "step": 1003008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3570733174218716e-05,
+      "loss": 3.5721,
+      "step": 1003520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3562347226708196e-05,
+      "loss": 3.568,
+      "step": 1004032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3553977658001405e-05,
+      "loss": 3.5778,
+      "step": 1004544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3545591710490885e-05,
+      "loss": 3.5632,
+      "step": 1005056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3537205762980365e-05,
+      "loss": 3.5597,
+      "step": 1005568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3528819815469845e-05,
+      "loss": 3.5769,
+      "step": 1006080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3520450246763054e-05,
+      "loss": 3.5674,
+      "step": 1006592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.351206429925254e-05,
+      "loss": 3.564,
+      "step": 1007104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.350367835174202e-05,
+      "loss": 3.5698,
+      "step": 1007616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.34952924042315e-05,
+      "loss": 3.5743,
+      "step": 1008128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.348692283552471e-05,
+      "loss": 3.5722,
+      "step": 1008640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.347853688801419e-05,
+      "loss": 3.5706,
+      "step": 1009152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.347015094050367e-05,
+      "loss": 3.57,
+      "step": 1009664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.346176499299315e-05,
+      "loss": 3.574,
+      "step": 1010176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.345339542428636e-05,
+      "loss": 3.58,
+      "step": 1010688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.344500947677584e-05,
+      "loss": 3.5546,
+      "step": 1011200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.343662352926532e-05,
+      "loss": 3.5582,
+      "step": 1011712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.34282375817548e-05,
+      "loss": 3.562,
+      "step": 1012224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.341986801304801e-05,
+      "loss": 3.5653,
+      "step": 1012736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3411482065537495e-05,
+      "loss": 3.5633,
+      "step": 1013248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3403096118026975e-05,
+      "loss": 3.5726,
+      "step": 1013760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3394710170516455e-05,
+      "loss": 3.5674,
+      "step": 1014272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3386340601809664e-05,
+      "loss": 3.5669,
+      "step": 1014784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3377954654299144e-05,
+      "loss": 3.5686,
+      "step": 1015296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3369568706788624e-05,
+      "loss": 3.5743,
+      "step": 1015808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3361182759278104e-05,
+      "loss": 3.5655,
+      "step": 1016320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.335281319057131e-05,
+      "loss": 3.5477,
+      "step": 1016832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.334442724306079e-05,
+      "loss": 3.5754,
+      "step": 1017344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.333604129555027e-05,
+      "loss": 3.5699,
+      "step": 1017856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.332765534803975e-05,
+      "loss": 3.5672,
+      "step": 1018368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.331928577933296e-05,
+      "loss": 3.5648,
+      "step": 1018880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.331089983182245e-05,
+      "loss": 3.5608,
+      "step": 1019392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.330251388431193e-05,
+      "loss": 3.5558,
+      "step": 1019904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.329412793680141e-05,
+      "loss": 3.5583,
+      "step": 1020416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.328575836809462e-05,
+      "loss": 3.576,
+      "step": 1020928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.32773724205841e-05,
+      "loss": 3.5587,
+      "step": 1021440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.326898647307358e-05,
+      "loss": 3.5821,
+      "step": 1021952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.326060052556306e-05,
+      "loss": 3.5718,
+      "step": 1022464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3252230956856266e-05,
+      "loss": 3.5478,
+      "step": 1022976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3243845009345746e-05,
+      "loss": 3.574,
+      "step": 1023488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3235459061835226e-05,
+      "loss": 3.5565,
+      "step": 1024000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3227073114324706e-05,
+      "loss": 3.55,
+      "step": 1024512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3218703545617915e-05,
+      "loss": 3.5571,
+      "step": 1025024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.32103175981074e-05,
+      "loss": 3.5616,
+      "step": 1025536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.320193165059688e-05,
+      "loss": 3.5582,
+      "step": 1026048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.319354570308636e-05,
+      "loss": 3.5615,
+      "step": 1026560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.318517613437957e-05,
+      "loss": 3.5508,
+      "step": 1027072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.317679018686905e-05,
+      "loss": 3.5667,
+      "step": 1027584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.316840423935853e-05,
+      "loss": 3.5729,
+      "step": 1028096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.316001829184801e-05,
+      "loss": 3.5672,
+      "step": 1028608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.315164872314122e-05,
+      "loss": 3.5688,
+      "step": 1029120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.31432627756307e-05,
+      "loss": 3.5656,
+      "step": 1029632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.313487682812018e-05,
+      "loss": 3.5806,
+      "step": 1030144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.312649088060966e-05,
+      "loss": 3.558,
+      "step": 1030656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.311812131190287e-05,
+      "loss": 3.5722,
+      "step": 1031168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3109735364392356e-05,
+      "loss": 3.5674,
+      "step": 1031680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3101349416881836e-05,
+      "loss": 3.5604,
+      "step": 1032192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3092963469371316e-05,
+      "loss": 3.5699,
+      "step": 1032704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3084593900664525e-05,
+      "loss": 3.5598,
+      "step": 1033216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3076207953154005e-05,
+      "loss": 3.5742,
+      "step": 1033728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3067822005643485e-05,
+      "loss": 3.5795,
+      "step": 1034240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3059436058132965e-05,
+      "loss": 3.5629,
+      "step": 1034752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3051066489426174e-05,
+      "loss": 3.5449,
+      "step": 1035264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3042680541915654e-05,
+      "loss": 3.5571,
+      "step": 1035776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3034294594405134e-05,
+      "loss": 3.5705,
+      "step": 1036288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3025908646894614e-05,
+      "loss": 3.5642,
+      "step": 1036800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.301753907818782e-05,
+      "loss": 3.5622,
+      "step": 1037312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.300915313067731e-05,
+      "loss": 3.562,
+      "step": 1037824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.300076718316679e-05,
+      "loss": 3.5596,
+      "step": 1038336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.299238123565627e-05,
+      "loss": 3.5614,
+      "step": 1038848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.298401166694948e-05,
+      "loss": 3.5591,
+      "step": 1039360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.297562571943896e-05,
+      "loss": 3.5659,
+      "step": 1039872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.296723977192844e-05,
+      "loss": 3.5693,
+      "step": 1040384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.295885382441792e-05,
+      "loss": 3.5757,
+      "step": 1040896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.295048425571113e-05,
+      "loss": 3.5546,
+      "step": 1041408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.294209830820061e-05,
+      "loss": 3.5617,
+      "step": 1041920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.293371236069009e-05,
+      "loss": 3.5648,
+      "step": 1042432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.292532641317956e-05,
+      "loss": 3.5626,
+      "step": 1042944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2916956844472776e-05,
+      "loss": 3.5703,
+      "step": 1043456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2908570896962256e-05,
+      "loss": 3.5613,
+      "step": 1043968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.290018494945174e-05,
+      "loss": 3.5551,
+      "step": 1044480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2891799001941216e-05,
+      "loss": 3.5559,
+      "step": 1044992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.288342943323443e-05,
+      "loss": 3.5653,
+      "step": 1045504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.287504348572391e-05,
+      "loss": 3.5586,
+      "step": 1046016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.286665753821339e-05,
+      "loss": 3.5419,
+      "step": 1046528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2858271590702865e-05,
+      "loss": 3.5639,
+      "step": 1047040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.284990202199608e-05,
+      "loss": 3.5665,
+      "step": 1047552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.284151607448556e-05,
+      "loss": 3.5571,
+      "step": 1048064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2833130126975034e-05,
+      "loss": 3.5566,
+      "step": 1048576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2824744179464514e-05,
+      "loss": 3.5568,
+      "step": 1049088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.281637461075773e-05,
+      "loss": 3.5668,
+      "step": 1049600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.280798866324721e-05,
+      "loss": 3.555,
+      "step": 1050112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.279960271573669e-05,
+      "loss": 3.564,
+      "step": 1050624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.279121676822617e-05,
+      "loss": 3.5693,
+      "step": 1051136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2782847199519386e-05,
+      "loss": 3.566,
+      "step": 1051648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2774461252008866e-05,
+      "loss": 3.5623,
+      "step": 1052160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.276607530449834e-05,
+      "loss": 3.5582,
+      "step": 1052672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.275768935698782e-05,
+      "loss": 3.5617,
+      "step": 1053184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2749319788281035e-05,
+      "loss": 3.5611,
+      "step": 1053696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.274093384077051e-05,
+      "loss": 3.5575,
+      "step": 1054208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.273254789325999e-05,
+      "loss": 3.5637,
+      "step": 1054720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.272416194574947e-05,
+      "loss": 3.5569,
+      "step": 1055232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2715792377042684e-05,
+      "loss": 3.5621,
+      "step": 1055744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2707406429532164e-05,
+      "loss": 3.5551,
+      "step": 1056256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2699020482021644e-05,
+      "loss": 3.5594,
+      "step": 1056768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2690634534511124e-05,
+      "loss": 3.5483,
+      "step": 1057280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.268226496580434e-05,
+      "loss": 3.558,
+      "step": 1057792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.267387901829381e-05,
+      "loss": 3.572,
+      "step": 1058304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.266549307078329e-05,
+      "loss": 3.5576,
+      "step": 1058816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.265710712327277e-05,
+      "loss": 3.5735,
+      "step": 1059328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.264873755456598e-05,
+      "loss": 3.5421,
+      "step": 1059840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.264035160705546e-05,
+      "loss": 3.5651,
+      "step": 1060352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.263196565954494e-05,
+      "loss": 3.5677,
+      "step": 1060864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.262357971203442e-05,
+      "loss": 3.561,
+      "step": 1061376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.261521014332764e-05,
+      "loss": 3.5613,
+      "step": 1061888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.260682419581712e-05,
+      "loss": 3.5507,
+      "step": 1062400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.25984382483066e-05,
+      "loss": 3.5679,
+      "step": 1062912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.259005230079608e-05,
+      "loss": 3.5766,
+      "step": 1063424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2581682732089286e-05,
+      "loss": 3.5515,
+      "step": 1063936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2573296784578766e-05,
+      "loss": 3.5563,
+      "step": 1064448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2564910837068246e-05,
+      "loss": 3.5629,
+      "step": 1064960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2556524889557726e-05,
+      "loss": 3.5557,
+      "step": 1065472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2548155320850935e-05,
+      "loss": 3.5549,
+      "step": 1065984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2539769373340415e-05,
+      "loss": 3.5643,
+      "step": 1066496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2531383425829895e-05,
+      "loss": 3.5511,
+      "step": 1067008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2522997478319375e-05,
+      "loss": 3.5692,
+      "step": 1067520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.251462790961259e-05,
+      "loss": 3.5536,
+      "step": 1068032
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.842634439468384,
+      "eval_runtime": 305.3484,
+      "eval_samples_per_second": 1249.69,
+      "eval_steps_per_second": 39.054,
+      "step": 1068480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.250624196210207e-05,
+      "loss": 3.5578,
+      "step": 1068544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.249785601459155e-05,
+      "loss": 3.5484,
+      "step": 1069056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.248947006708103e-05,
+      "loss": 3.5637,
+      "step": 1069568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.248110049837424e-05,
+      "loss": 3.5712,
+      "step": 1070080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.247271455086372e-05,
+      "loss": 3.5624,
+      "step": 1070592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.24643286033532e-05,
+      "loss": 3.56,
+      "step": 1071104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.245594265584268e-05,
+      "loss": 3.557,
+      "step": 1071616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.244757308713589e-05,
+      "loss": 3.5588,
+      "step": 1072128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.243918713962537e-05,
+      "loss": 3.5482,
+      "step": 1072640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.243080119211485e-05,
+      "loss": 3.5694,
+      "step": 1073152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.242241524460433e-05,
+      "loss": 3.5646,
+      "step": 1073664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2414045675897545e-05,
+      "loss": 3.5643,
+      "step": 1074176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.240567610719076e-05,
+      "loss": 3.5729,
+      "step": 1074688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2397290159680234e-05,
+      "loss": 3.5433,
+      "step": 1075200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2388904212169714e-05,
+      "loss": 3.5614,
+      "step": 1075712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2380518264659194e-05,
+      "loss": 3.547,
+      "step": 1076224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2372132317148674e-05,
+      "loss": 3.5578,
+      "step": 1076736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2363746369638154e-05,
+      "loss": 3.5477,
+      "step": 1077248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2355360422127634e-05,
+      "loss": 3.5528,
+      "step": 1077760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2346974474617113e-05,
+      "loss": 3.5475,
+      "step": 1078272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.233860490591032e-05,
+      "loss": 3.5752,
+      "step": 1078784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.23302189583998e-05,
+      "loss": 3.5613,
+      "step": 1079296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.232183301088928e-05,
+      "loss": 3.5571,
+      "step": 1079808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.231344706337876e-05,
+      "loss": 3.5524,
+      "step": 1080320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.230507749467198e-05,
+      "loss": 3.5639,
+      "step": 1080832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.229669154716146e-05,
+      "loss": 3.5463,
+      "step": 1081344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.228830559965094e-05,
+      "loss": 3.5436,
+      "step": 1081856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.227991965214042e-05,
+      "loss": 3.5612,
+      "step": 1082368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.227155008343363e-05,
+      "loss": 3.5487,
+      "step": 1082880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.226316413592311e-05,
+      "loss": 3.549,
+      "step": 1083392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.225477818841259e-05,
+      "loss": 3.5537,
+      "step": 1083904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.224639224090207e-05,
+      "loss": 3.5573,
+      "step": 1084416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2238022672195276e-05,
+      "loss": 3.5588,
+      "step": 1084928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2229636724684756e-05,
+      "loss": 3.5516,
+      "step": 1085440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2221250777174236e-05,
+      "loss": 3.5478,
+      "step": 1085952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2212864829663716e-05,
+      "loss": 3.5622,
+      "step": 1086464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.220449526095693e-05,
+      "loss": 3.5622,
+      "step": 1086976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.219610931344641e-05,
+      "loss": 3.5378,
+      "step": 1087488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.218772336593589e-05,
+      "loss": 3.5469,
+      "step": 1088000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.217933741842537e-05,
+      "loss": 3.5446,
+      "step": 1088512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.217096784971858e-05,
+      "loss": 3.547,
+      "step": 1089024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.216258190220806e-05,
+      "loss": 3.5483,
+      "step": 1089536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.215419595469754e-05,
+      "loss": 3.5575,
+      "step": 1090048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.214581000718702e-05,
+      "loss": 3.5474,
+      "step": 1090560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.213744043848023e-05,
+      "loss": 3.5499,
+      "step": 1091072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.212905449096971e-05,
+      "loss": 3.5569,
+      "step": 1091584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.212066854345919e-05,
+      "loss": 3.5542,
+      "step": 1092096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.211228259594867e-05,
+      "loss": 3.5547,
+      "step": 1092608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2103913027241886e-05,
+      "loss": 3.5315,
+      "step": 1093120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2095527079731366e-05,
+      "loss": 3.5555,
+      "step": 1093632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2087141132220846e-05,
+      "loss": 3.5566,
+      "step": 1094144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2078755184710326e-05,
+      "loss": 3.5514,
+      "step": 1094656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2070385616003535e-05,
+      "loss": 3.5466,
+      "step": 1095168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2061999668493015e-05,
+      "loss": 3.5498,
+      "step": 1095680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2053613720982495e-05,
+      "loss": 3.5326,
+      "step": 1096192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2045227773471974e-05,
+      "loss": 3.5462,
+      "step": 1096704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2036858204765184e-05,
+      "loss": 3.5609,
+      "step": 1097216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2028472257254664e-05,
+      "loss": 3.5449,
+      "step": 1097728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2020086309744143e-05,
+      "loss": 3.5607,
+      "step": 1098240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2011700362233623e-05,
+      "loss": 3.5565,
+      "step": 1098752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.200333079352684e-05,
+      "loss": 3.5332,
+      "step": 1099264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.199494484601632e-05,
+      "loss": 3.5585,
+      "step": 1099776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.19865588985058e-05,
+      "loss": 3.5399,
+      "step": 1100288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.197817295099528e-05,
+      "loss": 3.5365,
+      "step": 1100800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.196980338228849e-05,
+      "loss": 3.5409,
+      "step": 1101312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.196141743477797e-05,
+      "loss": 3.552,
+      "step": 1101824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.195303148726745e-05,
+      "loss": 3.5418,
+      "step": 1102336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.194464553975693e-05,
+      "loss": 3.5406,
+      "step": 1102848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.193627597105014e-05,
+      "loss": 3.5363,
+      "step": 1103360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.192789002353962e-05,
+      "loss": 3.5519,
+      "step": 1103872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.19195040760291e-05,
+      "loss": 3.5555,
+      "step": 1104384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.191111812851858e-05,
+      "loss": 3.5533,
+      "step": 1104896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.190274855981179e-05,
+      "loss": 3.552,
+      "step": 1105408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.189436261230127e-05,
+      "loss": 3.5541,
+      "step": 1105920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.188597666479075e-05,
+      "loss": 3.5588,
+      "step": 1106432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.187759071728023e-05,
+      "loss": 3.5515,
+      "step": 1106944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.186922114857344e-05,
+      "loss": 3.5535,
+      "step": 1107456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.186083520106292e-05,
+      "loss": 3.5512,
+      "step": 1107968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.18524492535524e-05,
+      "loss": 3.5461,
+      "step": 1108480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.184406330604188e-05,
+      "loss": 3.5542,
+      "step": 1108992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.183569373733509e-05,
+      "loss": 3.5462,
+      "step": 1109504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.182730778982457e-05,
+      "loss": 3.5576,
+      "step": 1110016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.181892184231405e-05,
+      "loss": 3.566,
+      "step": 1110528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.181053589480353e-05,
+      "loss": 3.5472,
+      "step": 1111040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.180216632609675e-05,
+      "loss": 3.5274,
+      "step": 1111552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1793780378586227e-05,
+      "loss": 3.5428,
+      "step": 1112064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1785394431075707e-05,
+      "loss": 3.5527,
+      "step": 1112576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1777008483565187e-05,
+      "loss": 3.5461,
+      "step": 1113088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1768638914858396e-05,
+      "loss": 3.5471,
+      "step": 1113600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1760252967347876e-05,
+      "loss": 3.5508,
+      "step": 1114112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1751867019837356e-05,
+      "loss": 3.5382,
+      "step": 1114624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1743497451130565e-05,
+      "loss": 3.5471,
+      "step": 1115136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1735111503620045e-05,
+      "loss": 3.5445,
+      "step": 1115648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1726725556109525e-05,
+      "loss": 3.552,
+      "step": 1116160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1718339608599004e-05,
+      "loss": 3.5496,
+      "step": 1116672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.170997003989222e-05,
+      "loss": 3.562,
+      "step": 1117184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.17015840923817e-05,
+      "loss": 3.5371,
+      "step": 1117696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.169319814487118e-05,
+      "loss": 3.548,
+      "step": 1118208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.168481219736066e-05,
+      "loss": 3.5497,
+      "step": 1118720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.167644262865387e-05,
+      "loss": 3.5492,
+      "step": 1119232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.166805668114335e-05,
+      "loss": 3.5576,
+      "step": 1119744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.165967073363283e-05,
+      "loss": 3.5389,
+      "step": 1120256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.16512847861223e-05,
+      "loss": 3.54,
+      "step": 1120768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.164291521741552e-05,
+      "loss": 3.5396,
+      "step": 1121280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1634529269905e-05,
+      "loss": 3.5505,
+      "step": 1121792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.162614332239448e-05,
+      "loss": 3.54,
+      "step": 1122304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.161775737488396e-05,
+      "loss": 3.5289,
+      "step": 1122816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1609387806177174e-05,
+      "loss": 3.546,
+      "step": 1123328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1601001858666654e-05,
+      "loss": 3.5473,
+      "step": 1123840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1592615911156134e-05,
+      "loss": 3.5436,
+      "step": 1124352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.158422996364561e-05,
+      "loss": 3.5469,
+      "step": 1124864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.157586039493882e-05,
+      "loss": 3.5366,
+      "step": 1125376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.15674744474283e-05,
+      "loss": 3.5533,
+      "step": 1125888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1559088499917776e-05,
+      "loss": 3.5402,
+      "step": 1126400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1550702552407256e-05,
+      "loss": 3.5462,
+      "step": 1126912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.154233298370047e-05,
+      "loss": 3.5535,
+      "step": 1127424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.153394703618995e-05,
+      "loss": 3.5539,
+      "step": 1127936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1525561088679425e-05,
+      "loss": 3.545,
+      "step": 1128448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.151717514116891e-05,
+      "loss": 3.5404,
+      "step": 1128960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.150880557246213e-05,
+      "loss": 3.5452,
+      "step": 1129472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.150041962495161e-05,
+      "loss": 3.5431,
+      "step": 1129984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.149203367744108e-05,
+      "loss": 3.5407,
+      "step": 1130496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.148364772993056e-05,
+      "loss": 3.5504,
+      "step": 1131008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.147527816122378e-05,
+      "loss": 3.5426,
+      "step": 1131520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.146689221371325e-05,
+      "loss": 3.5436,
+      "step": 1132032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.145850626620273e-05,
+      "loss": 3.5383,
+      "step": 1132544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.145012031869221e-05,
+      "loss": 3.543,
+      "step": 1133056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1441750749985426e-05,
+      "loss": 3.5375,
+      "step": 1133568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.14333648024749e-05,
+      "loss": 3.5399,
+      "step": 1134080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.142497885496438e-05,
+      "loss": 3.5532,
+      "step": 1134592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1416592907453865e-05,
+      "loss": 3.5445,
+      "step": 1135104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.140822333874708e-05,
+      "loss": 3.5557,
+      "step": 1135616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1399837391236555e-05,
+      "loss": 3.5328,
+      "step": 1136128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1391451443726034e-05,
+      "loss": 3.5463,
+      "step": 1136640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1383065496215514e-05,
+      "loss": 3.5516,
+      "step": 1137152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1374695927508724e-05,
+      "loss": 3.5476,
+      "step": 1137664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1366309979998204e-05,
+      "loss": 3.5425,
+      "step": 1138176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1357924032487683e-05,
+      "loss": 3.5376,
+      "step": 1138688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1349538084977163e-05,
+      "loss": 3.5535,
+      "step": 1139200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.134116851627037e-05,
+      "loss": 3.5622,
+      "step": 1139712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.133278256875985e-05,
+      "loss": 3.5404,
+      "step": 1140224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.132439662124933e-05,
+      "loss": 3.5389,
+      "step": 1140736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.131601067373882e-05,
+      "loss": 3.5469,
+      "step": 1141248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.130764110503203e-05,
+      "loss": 3.5391,
+      "step": 1141760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.129925515752151e-05,
+      "loss": 3.54,
+      "step": 1142272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.129086921001099e-05,
+      "loss": 3.5484,
+      "step": 1142784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.128248326250047e-05,
+      "loss": 3.5373,
+      "step": 1143296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.127411369379368e-05,
+      "loss": 3.5547,
+      "step": 1143808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.126572774628316e-05,
+      "loss": 3.5367,
+      "step": 1144320
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8429012298583984,
+      "eval_runtime": 309.9767,
+      "eval_samples_per_second": 1231.031,
+      "eval_steps_per_second": 38.471,
+      "step": 1144800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125735817757637e-05,
+      "loss": 3.4994,
+      "step": 1144832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1248972230065846e-05,
+      "loss": 3.5379,
+      "step": 1145344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1240586282555326e-05,
+      "loss": 3.5464,
+      "step": 1145856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1232200335044806e-05,
+      "loss": 3.5543,
+      "step": 1146368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.122383076633802e-05,
+      "loss": 3.5465,
+      "step": 1146880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.12154448188275e-05,
+      "loss": 3.5459,
+      "step": 1147392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.120705887131698e-05,
+      "loss": 3.5435,
+      "step": 1147904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.119867292380646e-05,
+      "loss": 3.5427,
+      "step": 1148416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.119030335509967e-05,
+      "loss": 3.5363,
+      "step": 1148928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.118191740758915e-05,
+      "loss": 3.5509,
+      "step": 1149440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.117353146007863e-05,
+      "loss": 3.5491,
+      "step": 1149952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.116514551256811e-05,
+      "loss": 3.551,
+      "step": 1150464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1156792322665056e-05,
+      "loss": 3.5573,
+      "step": 1150976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1148406375154536e-05,
+      "loss": 3.5279,
+      "step": 1151488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1140020427644016e-05,
+      "loss": 3.5463,
+      "step": 1152000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1131634480133496e-05,
+      "loss": 3.5325,
+      "step": 1152512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1123248532622976e-05,
+      "loss": 3.5458,
+      "step": 1153024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1114862585112456e-05,
+      "loss": 3.5326,
+      "step": 1153536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1106476637601936e-05,
+      "loss": 3.5403,
+      "step": 1154048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1098090690091416e-05,
+      "loss": 3.5346,
+      "step": 1154560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1089721121384625e-05,
+      "loss": 3.5582,
+      "step": 1155072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1081335173874105e-05,
+      "loss": 3.5484,
+      "step": 1155584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1072949226363585e-05,
+      "loss": 3.5379,
+      "step": 1156096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1064563278853065e-05,
+      "loss": 3.5375,
+      "step": 1156608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1056193710146274e-05,
+      "loss": 3.5489,
+      "step": 1157120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1047807762635754e-05,
+      "loss": 3.5301,
+      "step": 1157632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1039421815125234e-05,
+      "loss": 3.5301,
+      "step": 1158144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1031035867614713e-05,
+      "loss": 3.5446,
+      "step": 1158656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.102266629890793e-05,
+      "loss": 3.5325,
+      "step": 1159168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.101428035139741e-05,
+      "loss": 3.5364,
+      "step": 1159680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.100589440388689e-05,
+      "loss": 3.5385,
+      "step": 1160192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.099750845637637e-05,
+      "loss": 3.5397,
+      "step": 1160704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.098913888766958e-05,
+      "loss": 3.5466,
+      "step": 1161216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.098075294015906e-05,
+      "loss": 3.5323,
+      "step": 1161728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.097236699264854e-05,
+      "loss": 3.5376,
+      "step": 1162240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.096398104513802e-05,
+      "loss": 3.5466,
+      "step": 1162752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.095561147643123e-05,
+      "loss": 3.5468,
+      "step": 1163264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.094722552892071e-05,
+      "loss": 3.5233,
+      "step": 1163776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.093883958141019e-05,
+      "loss": 3.5352,
+      "step": 1164288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.093045363389967e-05,
+      "loss": 3.5256,
+      "step": 1164800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.092208406519288e-05,
+      "loss": 3.533,
+      "step": 1165312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.091369811768236e-05,
+      "loss": 3.5337,
+      "step": 1165824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.090531217017184e-05,
+      "loss": 3.5423,
+      "step": 1166336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.089692622266132e-05,
+      "loss": 3.5351,
+      "step": 1166848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.088855665395453e-05,
+      "loss": 3.535,
+      "step": 1167360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.088017070644401e-05,
+      "loss": 3.5414,
+      "step": 1167872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.087178475893349e-05,
+      "loss": 3.5416,
+      "step": 1168384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.086339881142297e-05,
+      "loss": 3.5351,
+      "step": 1168896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.085502924271618e-05,
+      "loss": 3.5182,
+      "step": 1169408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.084664329520566e-05,
+      "loss": 3.5396,
+      "step": 1169920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.083825734769514e-05,
+      "loss": 3.5441,
+      "step": 1170432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.082987140018462e-05,
+      "loss": 3.5368,
+      "step": 1170944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0821518210281566e-05,
+      "loss": 3.5302,
+      "step": 1171456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0813132262771046e-05,
+      "loss": 3.5329,
+      "step": 1171968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0804746315260526e-05,
+      "loss": 3.528,
+      "step": 1172480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0796360367750006e-05,
+      "loss": 3.5219,
+      "step": 1172992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0787974420239486e-05,
+      "loss": 3.5472,
+      "step": 1173504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0779588472728966e-05,
+      "loss": 3.5311,
+      "step": 1174016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0771202525218446e-05,
+      "loss": 3.5425,
+      "step": 1174528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0762816577707926e-05,
+      "loss": 3.5462,
+      "step": 1175040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0754447009001135e-05,
+      "loss": 3.5177,
+      "step": 1175552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0746061061490615e-05,
+      "loss": 3.5377,
+      "step": 1176064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0737675113980095e-05,
+      "loss": 3.5313,
+      "step": 1176576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0729289166469574e-05,
+      "loss": 3.5259,
+      "step": 1177088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.072091959776279e-05,
+      "loss": 3.5196,
+      "step": 1177600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.071253365025227e-05,
+      "loss": 3.5358,
+      "step": 1178112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.070414770274175e-05,
+      "loss": 3.5295,
+      "step": 1178624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.069576175523123e-05,
+      "loss": 3.5199,
+      "step": 1179136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.068739218652444e-05,
+      "loss": 3.5224,
+      "step": 1179648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.067900623901392e-05,
+      "loss": 3.5383,
+      "step": 1180160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.06706202915034e-05,
+      "loss": 3.5423,
+      "step": 1180672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.066223434399288e-05,
+      "loss": 3.5365,
+      "step": 1181184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.065386477528609e-05,
+      "loss": 3.5373,
+      "step": 1181696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.064547882777557e-05,
+      "loss": 3.5375,
+      "step": 1182208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.063709288026505e-05,
+      "loss": 3.5433,
+      "step": 1182720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.062870693275453e-05,
+      "loss": 3.5397,
+      "step": 1183232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0620337364047744e-05,
+      "loss": 3.5375,
+      "step": 1183744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0611951416537224e-05,
+      "loss": 3.5391,
+      "step": 1184256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0603565469026704e-05,
+      "loss": 3.5316,
+      "step": 1184768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0595179521516184e-05,
+      "loss": 3.5374,
+      "step": 1185280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.058680995280939e-05,
+      "loss": 3.5288,
+      "step": 1185792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.057842400529887e-05,
+      "loss": 3.5428,
+      "step": 1186304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.057003805778835e-05,
+      "loss": 3.545,
+      "step": 1186816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.056165211027783e-05,
+      "loss": 3.5354,
+      "step": 1187328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.055328254157104e-05,
+      "loss": 3.5131,
+      "step": 1187840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.054489659406052e-05,
+      "loss": 3.5278,
+      "step": 1188352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.053651064655e-05,
+      "loss": 3.5369,
+      "step": 1188864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.052812469903948e-05,
+      "loss": 3.5344,
+      "step": 1189376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.05197551303327e-05,
+      "loss": 3.5319,
+      "step": 1189888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0511369182822174e-05,
+      "loss": 3.5379,
+      "step": 1190400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0502983235311654e-05,
+      "loss": 3.5308,
+      "step": 1190912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0494613666604867e-05,
+      "loss": 3.5253,
+      "step": 1191424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0486227719094347e-05,
+      "loss": 3.5303,
+      "step": 1191936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0477841771583827e-05,
+      "loss": 3.5393,
+      "step": 1192448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0469455824073307e-05,
+      "loss": 3.533,
+      "step": 1192960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0461086255366516e-05,
+      "loss": 3.546,
+      "step": 1193472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0452700307855996e-05,
+      "loss": 3.525,
+      "step": 1193984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0444314360345476e-05,
+      "loss": 3.535,
+      "step": 1194496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.043592841283496e-05,
+      "loss": 3.536,
+      "step": 1195008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0427558844128168e-05,
+      "loss": 3.5332,
+      "step": 1195520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0419172896617648e-05,
+      "loss": 3.5407,
+      "step": 1196032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0410786949107128e-05,
+      "loss": 3.5306,
+      "step": 1196544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0402401001596608e-05,
+      "loss": 3.5255,
+      "step": 1197056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.039403143288982e-05,
+      "loss": 3.5262,
+      "step": 1197568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.03856454853793e-05,
+      "loss": 3.54,
+      "step": 1198080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.037725953786878e-05,
+      "loss": 3.5215,
+      "step": 1198592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.036887359035826e-05,
+      "loss": 3.5142,
+      "step": 1199104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.036050402165147e-05,
+      "loss": 3.5363,
+      "step": 1199616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.035211807414095e-05,
+      "loss": 3.531,
+      "step": 1200128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.034373212663043e-05,
+      "loss": 3.5268,
+      "step": 1200640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0335346179119913e-05,
+      "loss": 3.5329,
+      "step": 1201152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.032697661041312e-05,
+      "loss": 3.5225,
+      "step": 1201664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.03185906629026e-05,
+      "loss": 3.5379,
+      "step": 1202176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.031020471539208e-05,
+      "loss": 3.5202,
+      "step": 1202688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.030181876788156e-05,
+      "loss": 3.5326,
+      "step": 1203200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0293449199174774e-05,
+      "loss": 3.5409,
+      "step": 1203712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0285063251664254e-05,
+      "loss": 3.5409,
+      "step": 1204224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0276677304153734e-05,
+      "loss": 3.5274,
+      "step": 1204736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0268291356643214e-05,
+      "loss": 3.5262,
+      "step": 1205248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0259921787936423e-05,
+      "loss": 3.5345,
+      "step": 1205760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0251535840425903e-05,
+      "loss": 3.5285,
+      "step": 1206272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0243149892915383e-05,
+      "loss": 3.5246,
+      "step": 1206784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0234763945404866e-05,
+      "loss": 3.5365,
+      "step": 1207296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0226394376698075e-05,
+      "loss": 3.5287,
+      "step": 1207808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0218008429187555e-05,
+      "loss": 3.5299,
+      "step": 1208320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0209622481677035e-05,
+      "loss": 3.5266,
+      "step": 1208832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0201236534166515e-05,
+      "loss": 3.5279,
+      "step": 1209344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0192866965459728e-05,
+      "loss": 3.5287,
+      "step": 1209856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0184481017949208e-05,
+      "loss": 3.5237,
+      "step": 1210368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0176095070438688e-05,
+      "loss": 3.5312,
+      "step": 1210880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0167709122928168e-05,
+      "loss": 3.5377,
+      "step": 1211392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0159339554221377e-05,
+      "loss": 3.5386,
+      "step": 1211904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0150953606710857e-05,
+      "loss": 3.519,
+      "step": 1212416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0142567659200337e-05,
+      "loss": 3.5338,
+      "step": 1212928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.013418171168982e-05,
+      "loss": 3.5361,
+      "step": 1213440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.012581214298303e-05,
+      "loss": 3.5337,
+      "step": 1213952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.011742619547251e-05,
+      "loss": 3.5248,
+      "step": 1214464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.010904024796199e-05,
+      "loss": 3.5228,
+      "step": 1214976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0100654300451465e-05,
+      "loss": 3.5366,
+      "step": 1215488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.009228473174468e-05,
+      "loss": 3.5464,
+      "step": 1216000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.008389878423416e-05,
+      "loss": 3.5276,
+      "step": 1216512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.007551283672364e-05,
+      "loss": 3.5223,
+      "step": 1217024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0067126889213114e-05,
+      "loss": 3.5327,
+      "step": 1217536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.005875732050633e-05,
+      "loss": 3.5222,
+      "step": 1218048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.005037137299581e-05,
+      "loss": 3.5276,
+      "step": 1218560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.004198542548529e-05,
+      "loss": 3.538,
+      "step": 1219072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0033599477974767e-05,
+      "loss": 3.5201,
+      "step": 1219584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0025229909267983e-05,
+      "loss": 3.5391,
+      "step": 1220096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0016843961757463e-05,
+      "loss": 3.5235,
+      "step": 1220608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0008474393050672e-05,
+      "loss": 3.5295,
+      "step": 1221120
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.844369888305664,
+      "eval_runtime": 348.4714,
+      "eval_samples_per_second": 1095.042,
+      "eval_steps_per_second": 34.221,
+      "step": 1221120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.000008844554015e-05,
+      "loss": 3.522,
+      "step": 1221632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9991702498029635e-05,
+      "loss": 3.5296,
+      "step": 1222144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9983316550519115e-05,
+      "loss": 3.5407,
+      "step": 1222656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9974930603008588e-05,
+      "loss": 3.5343,
+      "step": 1223168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9966544655498068e-05,
+      "loss": 3.5334,
+      "step": 1223680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9958158707987548e-05,
+      "loss": 3.5283,
+      "step": 1224192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9949772760477028e-05,
+      "loss": 3.5282,
+      "step": 1224704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.994140319177024e-05,
+      "loss": 3.5167,
+      "step": 1225216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.993301724425972e-05,
+      "loss": 3.5435,
+      "step": 1225728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.99246312967492e-05,
+      "loss": 3.5292,
+      "step": 1226240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.991624534923868e-05,
+      "loss": 3.5405,
+      "step": 1226752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9907892159335625e-05,
+      "loss": 3.5415,
+      "step": 1227264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9899506211825105e-05,
+      "loss": 3.5178,
+      "step": 1227776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.989112026431459e-05,
+      "loss": 3.5275,
+      "step": 1228288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9882734316804062e-05,
+      "loss": 3.5201,
+      "step": 1228800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9874348369293542e-05,
+      "loss": 3.5251,
+      "step": 1229312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9865962421783022e-05,
+      "loss": 3.5238,
+      "step": 1229824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9857576474272502e-05,
+      "loss": 3.522,
+      "step": 1230336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.984919052676198e-05,
+      "loss": 3.5213,
+      "step": 1230848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9840820958055194e-05,
+      "loss": 3.5402,
+      "step": 1231360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9832435010544674e-05,
+      "loss": 3.5327,
+      "step": 1231872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9824049063034154e-05,
+      "loss": 3.5247,
+      "step": 1232384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9815663115523634e-05,
+      "loss": 3.5238,
+      "step": 1232896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9807293546816843e-05,
+      "loss": 3.5376,
+      "step": 1233408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9798907599306326e-05,
+      "loss": 3.5159,
+      "step": 1233920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9790521651795806e-05,
+      "loss": 3.5132,
+      "step": 1234432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9782135704285286e-05,
+      "loss": 3.5296,
+      "step": 1234944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9773766135578495e-05,
+      "loss": 3.5223,
+      "step": 1235456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9765380188067975e-05,
+      "loss": 3.5204,
+      "step": 1235968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9756994240557455e-05,
+      "loss": 3.5204,
+      "step": 1236480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9748608293046935e-05,
+      "loss": 3.527,
+      "step": 1236992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9740238724340148e-05,
+      "loss": 3.5314,
+      "step": 1237504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9731852776829628e-05,
+      "loss": 3.5172,
+      "step": 1238016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9723466829319108e-05,
+      "loss": 3.5237,
+      "step": 1238528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9715080881808588e-05,
+      "loss": 3.5332,
+      "step": 1239040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9706711313101797e-05,
+      "loss": 3.5326,
+      "step": 1239552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.969832536559128e-05,
+      "loss": 3.509,
+      "step": 1240064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.968993941808076e-05,
+      "loss": 3.5194,
+      "step": 1240576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.968155347057024e-05,
+      "loss": 3.5089,
+      "step": 1241088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.967318390186345e-05,
+      "loss": 3.5248,
+      "step": 1241600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.966479795435293e-05,
+      "loss": 3.5175,
+      "step": 1242112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.965641200684241e-05,
+      "loss": 3.5282,
+      "step": 1242624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.964802605933189e-05,
+      "loss": 3.5196,
+      "step": 1243136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.96396564906251e-05,
+      "loss": 3.524,
+      "step": 1243648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.963127054311458e-05,
+      "loss": 3.5234,
+      "step": 1244160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.962288459560406e-05,
+      "loss": 3.5285,
+      "step": 1244672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.961449864809354e-05,
+      "loss": 3.521,
+      "step": 1245184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.960612907938675e-05,
+      "loss": 3.5062,
+      "step": 1245696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9597743131876234e-05,
+      "loss": 3.5242,
+      "step": 1246208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9589357184365714e-05,
+      "loss": 3.5321,
+      "step": 1246720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9580971236855194e-05,
+      "loss": 3.5215,
+      "step": 1247232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9572618046952132e-05,
+      "loss": 3.5158,
+      "step": 1247744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9564232099441612e-05,
+      "loss": 3.516,
+      "step": 1248256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9555846151931095e-05,
+      "loss": 3.5163,
+      "step": 1248768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9547460204420575e-05,
+      "loss": 3.5072,
+      "step": 1249280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9539074256910055e-05,
+      "loss": 3.5343,
+      "step": 1249792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9530688309399535e-05,
+      "loss": 3.5171,
+      "step": 1250304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9522302361889015e-05,
+      "loss": 3.5305,
+      "step": 1250816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9513916414378495e-05,
+      "loss": 3.5277,
+      "step": 1251328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9505546845671704e-05,
+      "loss": 3.5055,
+      "step": 1251840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9497160898161184e-05,
+      "loss": 3.5238,
+      "step": 1252352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9488774950650667e-05,
+      "loss": 3.5181,
+      "step": 1252864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9480389003140147e-05,
+      "loss": 3.5084,
+      "step": 1253376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9472019434433356e-05,
+      "loss": 3.5044,
+      "step": 1253888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9463633486922836e-05,
+      "loss": 3.524,
+      "step": 1254400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9455247539412316e-05,
+      "loss": 3.5106,
+      "step": 1254912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9446861591901796e-05,
+      "loss": 3.5112,
+      "step": 1255424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.943849202319501e-05,
+      "loss": 3.5044,
+      "step": 1255936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.943010607568449e-05,
+      "loss": 3.5246,
+      "step": 1256448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.942172012817397e-05,
+      "loss": 3.5267,
+      "step": 1256960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.941333418066345e-05,
+      "loss": 3.5236,
+      "step": 1257472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9404964611956658e-05,
+      "loss": 3.525,
+      "step": 1257984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9396578664446138e-05,
+      "loss": 3.5267,
+      "step": 1258496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.938819271693562e-05,
+      "loss": 3.5277,
+      "step": 1259008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.93798067694251e-05,
+      "loss": 3.527,
+      "step": 1259520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.937143720071831e-05,
+      "loss": 3.5169,
+      "step": 1260032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.936305125320779e-05,
+      "loss": 3.5276,
+      "step": 1260544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.935466530569727e-05,
+      "loss": 3.52,
+      "step": 1261056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.934627935818675e-05,
+      "loss": 3.5222,
+      "step": 1261568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9337909789479963e-05,
+      "loss": 3.517,
+      "step": 1262080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9329523841969442e-05,
+      "loss": 3.5281,
+      "step": 1262592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9321137894458922e-05,
+      "loss": 3.5305,
+      "step": 1263104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9312751946948402e-05,
+      "loss": 3.5211,
+      "step": 1263616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.930438237824161e-05,
+      "loss": 3.4993,
+      "step": 1264128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.929599643073109e-05,
+      "loss": 3.5122,
+      "step": 1264640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9287610483220575e-05,
+      "loss": 3.5269,
+      "step": 1265152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9279224535710055e-05,
+      "loss": 3.5224,
+      "step": 1265664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9270854967003264e-05,
+      "loss": 3.5191,
+      "step": 1266176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9262469019492744e-05,
+      "loss": 3.5199,
+      "step": 1266688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9254083071982224e-05,
+      "loss": 3.5207,
+      "step": 1267200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9245697124471704e-05,
+      "loss": 3.5105,
+      "step": 1267712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9237327555764916e-05,
+      "loss": 3.5165,
+      "step": 1268224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9228941608254396e-05,
+      "loss": 3.5238,
+      "step": 1268736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9220555660743876e-05,
+      "loss": 3.519,
+      "step": 1269248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9212169713233356e-05,
+      "loss": 3.5321,
+      "step": 1269760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9203800144526565e-05,
+      "loss": 3.5135,
+      "step": 1270272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9195414197016045e-05,
+      "loss": 3.5187,
+      "step": 1270784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.918702824950553e-05,
+      "loss": 3.5214,
+      "step": 1271296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9178658680798738e-05,
+      "loss": 3.5202,
+      "step": 1271808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9170272733288217e-05,
+      "loss": 3.5237,
+      "step": 1272320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9161886785777697e-05,
+      "loss": 3.5174,
+      "step": 1272832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9153500838267177e-05,
+      "loss": 3.511,
+      "step": 1273344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.914513126956039e-05,
+      "loss": 3.5121,
+      "step": 1273856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.913674532204987e-05,
+      "loss": 3.5254,
+      "step": 1274368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.912835937453935e-05,
+      "loss": 3.5097,
+      "step": 1274880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.911997342702883e-05,
+      "loss": 3.5012,
+      "step": 1275392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.911160385832204e-05,
+      "loss": 3.5177,
+      "step": 1275904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.910321791081152e-05,
+      "loss": 3.5218,
+      "step": 1276416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9094831963301e-05,
+      "loss": 3.5119,
+      "step": 1276928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9086446015790482e-05,
+      "loss": 3.5215,
+      "step": 1277440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.907807644708369e-05,
+      "loss": 3.5095,
+      "step": 1277952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.906969049957317e-05,
+      "loss": 3.5227,
+      "step": 1278464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.906130455206265e-05,
+      "loss": 3.507,
+      "step": 1278976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.905291860455213e-05,
+      "loss": 3.5178,
+      "step": 1279488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9044549035845344e-05,
+      "loss": 3.5249,
+      "step": 1280000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9036163088334824e-05,
+      "loss": 3.5276,
+      "step": 1280512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9027777140824303e-05,
+      "loss": 3.5129,
+      "step": 1281024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9019391193313783e-05,
+      "loss": 3.5154,
+      "step": 1281536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9011021624606993e-05,
+      "loss": 3.5178,
+      "step": 1282048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9002635677096472e-05,
+      "loss": 3.5156,
+      "step": 1282560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8994249729585952e-05,
+      "loss": 3.5105,
+      "step": 1283072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8985863782075436e-05,
+      "loss": 3.5189,
+      "step": 1283584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8977494213368645e-05,
+      "loss": 3.5176,
+      "step": 1284096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8969108265858125e-05,
+      "loss": 3.5222,
+      "step": 1284608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8960722318347605e-05,
+      "loss": 3.5106,
+      "step": 1285120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8952336370837085e-05,
+      "loss": 3.5106,
+      "step": 1285632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8943966802130297e-05,
+      "loss": 3.5121,
+      "step": 1286144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8935580854619777e-05,
+      "loss": 3.5164,
+      "step": 1286656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8927194907109257e-05,
+      "loss": 3.5128,
+      "step": 1287168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.891880895959873e-05,
+      "loss": 3.5272,
+      "step": 1287680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8910439390891946e-05,
+      "loss": 3.5228,
+      "step": 1288192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8902053443381426e-05,
+      "loss": 3.5099,
+      "step": 1288704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8893667495870906e-05,
+      "loss": 3.5145,
+      "step": 1289216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8885281548360383e-05,
+      "loss": 3.5261,
+      "step": 1289728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.88769119796536e-05,
+      "loss": 3.521,
+      "step": 1290240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.886852603214308e-05,
+      "loss": 3.514,
+      "step": 1290752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.886014008463256e-05,
+      "loss": 3.5093,
+      "step": 1291264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8851754137122035e-05,
+      "loss": 3.5269,
+      "step": 1291776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.884338456841525e-05,
+      "loss": 3.5276,
+      "step": 1292288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.883499862090473e-05,
+      "loss": 3.5152,
+      "step": 1292800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8826612673394204e-05,
+      "loss": 3.5137,
+      "step": 1293312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8818226725883684e-05,
+      "loss": 3.5154,
+      "step": 1293824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.88098571571769e-05,
+      "loss": 3.5127,
+      "step": 1294336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.880147120966638e-05,
+      "loss": 3.5094,
+      "step": 1294848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8793085262155856e-05,
+      "loss": 3.5242,
+      "step": 1295360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8784699314645336e-05,
+      "loss": 3.5123,
+      "step": 1295872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8776329745938552e-05,
+      "loss": 3.5203,
+      "step": 1296384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8767943798428032e-05,
+      "loss": 3.513,
+      "step": 1296896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8759557850917505e-05,
+      "loss": 3.5108,
+      "step": 1297408
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.845435857772827,
+      "eval_runtime": 345.6394,
+      "eval_samples_per_second": 1104.015,
+      "eval_steps_per_second": 34.501,
+      "step": 1297440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.875117190340699e-05,
+      "loss": 3.5072,
+      "step": 1297920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.874278595589647e-05,
+      "loss": 3.5192,
+      "step": 1298432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.873440000838595e-05,
+      "loss": 3.5267,
+      "step": 1298944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8726030439679158e-05,
+      "loss": 3.5177,
+      "step": 1299456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8717644492168638e-05,
+      "loss": 3.5207,
+      "step": 1299968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8709258544658118e-05,
+      "loss": 3.5152,
+      "step": 1300480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8700872597147598e-05,
+      "loss": 3.5133,
+      "step": 1300992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.869250302844081e-05,
+      "loss": 3.5037,
+      "step": 1301504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.868411708093029e-05,
+      "loss": 3.5295,
+      "step": 1302016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.867573113341977e-05,
+      "loss": 3.5144,
+      "step": 1302528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.866734518590925e-05,
+      "loss": 3.5278,
+      "step": 1303040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.865897561720246e-05,
+      "loss": 3.5221,
+      "step": 1303552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8650606048495675e-05,
+      "loss": 3.5109,
+      "step": 1304064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.864222010098515e-05,
+      "loss": 3.5083,
+      "step": 1304576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.863383415347463e-05,
+      "loss": 3.5117,
+      "step": 1305088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.862544820596411e-05,
+      "loss": 3.508,
+      "step": 1305600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.861706225845359e-05,
+      "loss": 3.5113,
+      "step": 1306112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.860867631094307e-05,
+      "loss": 3.5074,
+      "step": 1306624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.860029036343255e-05,
+      "loss": 3.5049,
+      "step": 1307136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8591920794725764e-05,
+      "loss": 3.5315,
+      "step": 1307648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8583534847215244e-05,
+      "loss": 3.5188,
+      "step": 1308160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8575148899704724e-05,
+      "loss": 3.5125,
+      "step": 1308672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8566762952194204e-05,
+      "loss": 3.5112,
+      "step": 1309184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8558393383487413e-05,
+      "loss": 3.5256,
+      "step": 1309696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8550007435976893e-05,
+      "loss": 3.4972,
+      "step": 1310208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8541621488466376e-05,
+      "loss": 3.503,
+      "step": 1310720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8533235540955856e-05,
+      "loss": 3.5125,
+      "step": 1311232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8524865972249065e-05,
+      "loss": 3.5118,
+      "step": 1311744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8516480024738545e-05,
+      "loss": 3.5057,
+      "step": 1312256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8508094077228025e-05,
+      "loss": 3.5072,
+      "step": 1312768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8499708129717505e-05,
+      "loss": 3.5145,
+      "step": 1313280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8491338561010717e-05,
+      "loss": 3.5203,
+      "step": 1313792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8482952613500197e-05,
+      "loss": 3.5022,
+      "step": 1314304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8474566665989677e-05,
+      "loss": 3.5133,
+      "step": 1314816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8466180718479157e-05,
+      "loss": 3.5188,
+      "step": 1315328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8457811149772366e-05,
+      "loss": 3.5176,
+      "step": 1315840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8449425202261846e-05,
+      "loss": 3.4987,
+      "step": 1316352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.844103925475133e-05,
+      "loss": 3.5022,
+      "step": 1316864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.843265330724081e-05,
+      "loss": 3.4992,
+      "step": 1317376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.842428373853402e-05,
+      "loss": 3.5091,
+      "step": 1317888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.84158977910235e-05,
+      "loss": 3.5037,
+      "step": 1318400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.840751184351298e-05,
+      "loss": 3.5112,
+      "step": 1318912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.839912589600246e-05,
+      "loss": 3.5045,
+      "step": 1319424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.839075632729567e-05,
+      "loss": 3.5103,
+      "step": 1319936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.838237037978515e-05,
+      "loss": 3.5134,
+      "step": 1320448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.837398443227463e-05,
+      "loss": 3.5125,
+      "step": 1320960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.836559848476411e-05,
+      "loss": 3.5106,
+      "step": 1321472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.835722891605732e-05,
+      "loss": 3.4922,
+      "step": 1321984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.83488429685468e-05,
+      "loss": 3.5086,
+      "step": 1322496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8340457021036283e-05,
+      "loss": 3.5147,
+      "step": 1323008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8332071073525763e-05,
+      "loss": 3.51,
+      "step": 1323520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.83237178836227e-05,
+      "loss": 3.5052,
+      "step": 1324032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.831533193611218e-05,
+      "loss": 3.4981,
+      "step": 1324544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.830694598860166e-05,
+      "loss": 3.5062,
+      "step": 1325056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8298560041091145e-05,
+      "loss": 3.4912,
+      "step": 1325568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8290174093580625e-05,
+      "loss": 3.522,
+      "step": 1326080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8281788146070105e-05,
+      "loss": 3.5037,
+      "step": 1326592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8273402198559585e-05,
+      "loss": 3.5165,
+      "step": 1327104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8265016251049065e-05,
+      "loss": 3.5155,
+      "step": 1327616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8256646682342274e-05,
+      "loss": 3.4909,
+      "step": 1328128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8248260734831754e-05,
+      "loss": 3.5094,
+      "step": 1328640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8239874787321237e-05,
+      "loss": 3.5063,
+      "step": 1329152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8231488839810717e-05,
+      "loss": 3.4922,
+      "step": 1329664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8223119271103926e-05,
+      "loss": 3.4917,
+      "step": 1330176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8214733323593406e-05,
+      "loss": 3.511,
+      "step": 1330688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8206347376082886e-05,
+      "loss": 3.5005,
+      "step": 1331200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8197961428572366e-05,
+      "loss": 3.4971,
+      "step": 1331712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.818959185986558e-05,
+      "loss": 3.4918,
+      "step": 1332224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.818120591235506e-05,
+      "loss": 3.5098,
+      "step": 1332736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8172819964844538e-05,
+      "loss": 3.5113,
+      "step": 1333248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8164434017334018e-05,
+      "loss": 3.5105,
+      "step": 1333760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8156064448627227e-05,
+      "loss": 3.5091,
+      "step": 1334272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8147678501116707e-05,
+      "loss": 3.5189,
+      "step": 1334784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.813929255360619e-05,
+      "loss": 3.5129,
+      "step": 1335296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.813090660609567e-05,
+      "loss": 3.5176,
+      "step": 1335808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.812253703738888e-05,
+      "loss": 3.5013,
+      "step": 1336320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.811415108987836e-05,
+      "loss": 3.5162,
+      "step": 1336832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.810576514236784e-05,
+      "loss": 3.5066,
+      "step": 1337344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.809737919485732e-05,
+      "loss": 3.5071,
+      "step": 1337856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8089009626150532e-05,
+      "loss": 3.5061,
+      "step": 1338368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8080623678640012e-05,
+      "loss": 3.511,
+      "step": 1338880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8072237731129492e-05,
+      "loss": 3.5214,
+      "step": 1339392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8063851783618972e-05,
+      "loss": 3.5101,
+      "step": 1339904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.805548221491218e-05,
+      "loss": 3.4893,
+      "step": 1340416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8047112646205393e-05,
+      "loss": 3.4967,
+      "step": 1340928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8038726698694873e-05,
+      "loss": 3.5092,
+      "step": 1341440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8030340751184353e-05,
+      "loss": 3.5094,
+      "step": 1341952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8021954803673833e-05,
+      "loss": 3.5083,
+      "step": 1342464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8013568856163313e-05,
+      "loss": 3.5077,
+      "step": 1342976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8005182908652793e-05,
+      "loss": 3.5044,
+      "step": 1343488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7996796961142273e-05,
+      "loss": 3.4964,
+      "step": 1344000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7988411013631753e-05,
+      "loss": 3.5056,
+      "step": 1344512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7980041444924966e-05,
+      "loss": 3.5048,
+      "step": 1345024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7971655497414446e-05,
+      "loss": 3.5076,
+      "step": 1345536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7963269549903926e-05,
+      "loss": 3.5159,
+      "step": 1346048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7954883602393406e-05,
+      "loss": 3.5077,
+      "step": 1346560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7946514033686615e-05,
+      "loss": 3.5027,
+      "step": 1347072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7938128086176098e-05,
+      "loss": 3.5088,
+      "step": 1347584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7929742138665578e-05,
+      "loss": 3.5093,
+      "step": 1348096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7921356191155058e-05,
+      "loss": 3.5103,
+      "step": 1348608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7912986622448267e-05,
+      "loss": 3.5043,
+      "step": 1349120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7904600674937747e-05,
+      "loss": 3.4991,
+      "step": 1349632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7896214727427227e-05,
+      "loss": 3.4999,
+      "step": 1350144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7887828779916703e-05,
+      "loss": 3.5116,
+      "step": 1350656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.787945921120992e-05,
+      "loss": 3.4968,
+      "step": 1351168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.78710732636994e-05,
+      "loss": 3.4866,
+      "step": 1351680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.786268731618888e-05,
+      "loss": 3.505,
+      "step": 1352192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7854301368678352e-05,
+      "loss": 3.5071,
+      "step": 1352704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7845931799971568e-05,
+      "loss": 3.4969,
+      "step": 1353216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.783754585246105e-05,
+      "loss": 3.5065,
+      "step": 1353728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.782915990495053e-05,
+      "loss": 3.4986,
+      "step": 1354240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7820773957440005e-05,
+      "loss": 3.5093,
+      "step": 1354752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.781240438873322e-05,
+      "loss": 3.496,
+      "step": 1355264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.78040184412227e-05,
+      "loss": 3.504,
+      "step": 1355776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7795632493712177e-05,
+      "loss": 3.5133,
+      "step": 1356288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7787246546201657e-05,
+      "loss": 3.5161,
+      "step": 1356800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7778876977494873e-05,
+      "loss": 3.499,
+      "step": 1357312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7770491029984353e-05,
+      "loss": 3.5022,
+      "step": 1357824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7762105082473826e-05,
+      "loss": 3.5081,
+      "step": 1358336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7753719134963306e-05,
+      "loss": 3.504,
+      "step": 1358848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7745349566256522e-05,
+      "loss": 3.4935,
+      "step": 1359360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7736963618746002e-05,
+      "loss": 3.5062,
+      "step": 1359872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.772857767123548e-05,
+      "loss": 3.4993,
+      "step": 1360384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.772019172372496e-05,
+      "loss": 3.5099,
+      "step": 1360896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7711822155018174e-05,
+      "loss": 3.4947,
+      "step": 1361408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.770343620750765e-05,
+      "loss": 3.4979,
+      "step": 1361920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.769505025999713e-05,
+      "loss": 3.498,
+      "step": 1362432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.768666431248661e-05,
+      "loss": 3.5027,
+      "step": 1362944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7678294743779827e-05,
+      "loss": 3.4983,
+      "step": 1363456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.76699087962693e-05,
+      "loss": 3.5156,
+      "step": 1363968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.766152284875878e-05,
+      "loss": 3.5093,
+      "step": 1364480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.765313690124826e-05,
+      "loss": 3.4981,
+      "step": 1364992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7644767332541472e-05,
+      "loss": 3.4985,
+      "step": 1365504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7636381385030952e-05,
+      "loss": 3.5135,
+      "step": 1366016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7627995437520432e-05,
+      "loss": 3.5098,
+      "step": 1366528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7619609490009912e-05,
+      "loss": 3.4995,
+      "step": 1367040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.761123992130312e-05,
+      "loss": 3.4974,
+      "step": 1367552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.76028539737926e-05,
+      "loss": 3.5115,
+      "step": 1368064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7594468026282085e-05,
+      "loss": 3.5123,
+      "step": 1368576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7586082078771564e-05,
+      "loss": 3.5041,
+      "step": 1369088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7577712510064774e-05,
+      "loss": 3.5009,
+      "step": 1369600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7569326562554254e-05,
+      "loss": 3.5041,
+      "step": 1370112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7560940615043733e-05,
+      "loss": 3.4967,
+      "step": 1370624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7552554667533213e-05,
+      "loss": 3.4965,
+      "step": 1371136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7544185098826426e-05,
+      "loss": 3.5133,
+      "step": 1371648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7535799151315906e-05,
+      "loss": 3.5012,
+      "step": 1372160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7527413203805386e-05,
+      "loss": 3.5077,
+      "step": 1372672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7519027256294866e-05,
+      "loss": 3.5006,
+      "step": 1373184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.751067406639181e-05,
+      "loss": 3.4985,
+      "step": 1373696
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.846585988998413,
+      "eval_runtime": 304.785,
+      "eval_samples_per_second": 1252.001,
+      "eval_steps_per_second": 39.126,
+      "step": 1373760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.750228811888129e-05,
+      "loss": 3.4952,
+      "step": 1374208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.749390217137077e-05,
+      "loss": 3.5037,
+      "step": 1374720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7485516223860247e-05,
+      "loss": 3.5143,
+      "step": 1375232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7477130276349727e-05,
+      "loss": 3.5044,
+      "step": 1375744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7468744328839207e-05,
+      "loss": 3.5104,
+      "step": 1376256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7460358381328687e-05,
+      "loss": 3.502,
+      "step": 1376768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7451972433818167e-05,
+      "loss": 3.5008,
+      "step": 1377280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.744360286511138e-05,
+      "loss": 3.4906,
+      "step": 1377792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.743521691760086e-05,
+      "loss": 3.5135,
+      "step": 1378304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.742683097009034e-05,
+      "loss": 3.5088,
+      "step": 1378816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.741844502257982e-05,
+      "loss": 3.5054,
+      "step": 1379328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.741007545387303e-05,
+      "loss": 3.5105,
+      "step": 1379840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7401705885166244e-05,
+      "loss": 3.4975,
+      "step": 1380352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.739331993765572e-05,
+      "loss": 3.4954,
+      "step": 1380864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.73849339901452e-05,
+      "loss": 3.5035,
+      "step": 1381376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.737654804263468e-05,
+      "loss": 3.4927,
+      "step": 1381888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.736816209512416e-05,
+      "loss": 3.4967,
+      "step": 1382400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.735977614761364e-05,
+      "loss": 3.4943,
+      "step": 1382912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.735139020010312e-05,
+      "loss": 3.4929,
+      "step": 1383424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7343020631396333e-05,
+      "loss": 3.52,
+      "step": 1383936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7334634683885813e-05,
+      "loss": 3.5026,
+      "step": 1384448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7326248736375293e-05,
+      "loss": 3.5058,
+      "step": 1384960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7317862788864773e-05,
+      "loss": 3.497,
+      "step": 1385472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7309493220157982e-05,
+      "loss": 3.511,
+      "step": 1385984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7301107272647462e-05,
+      "loss": 3.4828,
+      "step": 1386496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7292721325136946e-05,
+      "loss": 3.4927,
+      "step": 1387008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7284335377626425e-05,
+      "loss": 3.4945,
+      "step": 1387520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7275965808919635e-05,
+      "loss": 3.502,
+      "step": 1388032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7267579861409115e-05,
+      "loss": 3.4928,
+      "step": 1388544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7259193913898594e-05,
+      "loss": 3.4927,
+      "step": 1389056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7250807966388074e-05,
+      "loss": 3.5027,
+      "step": 1389568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7242438397681287e-05,
+      "loss": 3.5082,
+      "step": 1390080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7234052450170767e-05,
+      "loss": 3.4877,
+      "step": 1390592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7225666502660247e-05,
+      "loss": 3.5017,
+      "step": 1391104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7217280555149727e-05,
+      "loss": 3.5031,
+      "step": 1391616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7208910986442936e-05,
+      "loss": 3.5006,
+      "step": 1392128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7200525038932416e-05,
+      "loss": 3.4869,
+      "step": 1392640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.71921390914219e-05,
+      "loss": 3.4934,
+      "step": 1393152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.718375314391138e-05,
+      "loss": 3.4839,
+      "step": 1393664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7175383575204588e-05,
+      "loss": 3.4993,
+      "step": 1394176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7166997627694068e-05,
+      "loss": 3.4906,
+      "step": 1394688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7158611680183548e-05,
+      "loss": 3.4991,
+      "step": 1395200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7150225732673028e-05,
+      "loss": 3.492,
+      "step": 1395712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.714185616396624e-05,
+      "loss": 3.4935,
+      "step": 1396224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.713347021645572e-05,
+      "loss": 3.4993,
+      "step": 1396736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.71250842689452e-05,
+      "loss": 3.5027,
+      "step": 1397248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.711669832143468e-05,
+      "loss": 3.4973,
+      "step": 1397760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.710832875272789e-05,
+      "loss": 3.4782,
+      "step": 1398272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.709994280521737e-05,
+      "loss": 3.4933,
+      "step": 1398784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7091556857706853e-05,
+      "loss": 3.5047,
+      "step": 1399296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7083170910196333e-05,
+      "loss": 3.4939,
+      "step": 1399808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.707481772029327e-05,
+      "loss": 3.4919,
+      "step": 1400320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.706643177278275e-05,
+      "loss": 3.4878,
+      "step": 1400832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.705804582527223e-05,
+      "loss": 3.491,
+      "step": 1401344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7049659877761714e-05,
+      "loss": 3.4806,
+      "step": 1401856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7041273930251194e-05,
+      "loss": 3.5031,
+      "step": 1402368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7032887982740674e-05,
+      "loss": 3.4949,
+      "step": 1402880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7024502035230154e-05,
+      "loss": 3.5003,
+      "step": 1403392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7016116087719634e-05,
+      "loss": 3.5058,
+      "step": 1403904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7007746519012843e-05,
+      "loss": 3.4804,
+      "step": 1404416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6999360571502323e-05,
+      "loss": 3.4971,
+      "step": 1404928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6990974623991807e-05,
+      "loss": 3.4904,
+      "step": 1405440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6982588676481286e-05,
+      "loss": 3.4846,
+      "step": 1405952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6974219107774496e-05,
+      "loss": 3.4786,
+      "step": 1406464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6965833160263976e-05,
+      "loss": 3.4944,
+      "step": 1406976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6957447212753455e-05,
+      "loss": 3.4911,
+      "step": 1407488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6949061265242935e-05,
+      "loss": 3.4839,
+      "step": 1408000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6940691696536148e-05,
+      "loss": 3.4772,
+      "step": 1408512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6932305749025628e-05,
+      "loss": 3.4978,
+      "step": 1409024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6923919801515108e-05,
+      "loss": 3.5024,
+      "step": 1409536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6915533854004588e-05,
+      "loss": 3.4956,
+      "step": 1410048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6907164285297797e-05,
+      "loss": 3.4959,
+      "step": 1410560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6898778337787277e-05,
+      "loss": 3.5061,
+      "step": 1411072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.689039239027676e-05,
+      "loss": 3.5008,
+      "step": 1411584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.688200644276624e-05,
+      "loss": 3.5057,
+      "step": 1412096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.687363687405945e-05,
+      "loss": 3.4854,
+      "step": 1412608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.686525092654893e-05,
+      "loss": 3.5031,
+      "step": 1413120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.685686497903841e-05,
+      "loss": 3.4975,
+      "step": 1413632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.684847903152789e-05,
+      "loss": 3.4893,
+      "step": 1414144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.68401094628211e-05,
+      "loss": 3.4981,
+      "step": 1414656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.683172351531058e-05,
+      "loss": 3.4974,
+      "step": 1415168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.682333756780006e-05,
+      "loss": 3.509,
+      "step": 1415680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.681495162028954e-05,
+      "loss": 3.4955,
+      "step": 1416192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.680658205158275e-05,
+      "loss": 3.4771,
+      "step": 1416704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.679819610407223e-05,
+      "loss": 3.4831,
+      "step": 1417216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.678981015656171e-05,
+      "loss": 3.4946,
+      "step": 1417728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6781424209051194e-05,
+      "loss": 3.4985,
+      "step": 1418240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6773054640344403e-05,
+      "loss": 3.4976,
+      "step": 1418752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6764668692833883e-05,
+      "loss": 3.4915,
+      "step": 1419264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6756282745323363e-05,
+      "loss": 3.4926,
+      "step": 1419776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6747896797812843e-05,
+      "loss": 3.4855,
+      "step": 1420288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6739527229106055e-05,
+      "loss": 3.4944,
+      "step": 1420800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6731141281595535e-05,
+      "loss": 3.4882,
+      "step": 1421312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6722755334085015e-05,
+      "loss": 3.4963,
+      "step": 1421824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6714369386574495e-05,
+      "loss": 3.5025,
+      "step": 1422336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6705999817867704e-05,
+      "loss": 3.4984,
+      "step": 1422848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6697613870357184e-05,
+      "loss": 3.4882,
+      "step": 1423360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6689227922846664e-05,
+      "loss": 3.4977,
+      "step": 1423872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6680841975336147e-05,
+      "loss": 3.4911,
+      "step": 1424384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6672472406629357e-05,
+      "loss": 3.4986,
+      "step": 1424896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6664086459118837e-05,
+      "loss": 3.4921,
+      "step": 1425408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6655700511608316e-05,
+      "loss": 3.4883,
+      "step": 1425920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6647314564097796e-05,
+      "loss": 3.4832,
+      "step": 1426432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.663894499539101e-05,
+      "loss": 3.4992,
+      "step": 1426944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.663055904788049e-05,
+      "loss": 3.4873,
+      "step": 1427456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.662217310036997e-05,
+      "loss": 3.4774,
+      "step": 1427968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6613803531663178e-05,
+      "loss": 3.4877,
+      "step": 1428480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6605417584152658e-05,
+      "loss": 3.4941,
+      "step": 1428992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6597031636642138e-05,
+      "loss": 3.4852,
+      "step": 1429504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6588645689131618e-05,
+      "loss": 3.4956,
+      "step": 1430016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.658027612042483e-05,
+      "loss": 3.4845,
+      "step": 1430528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.657189017291431e-05,
+      "loss": 3.4972,
+      "step": 1431040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.656350422540379e-05,
+      "loss": 3.4831,
+      "step": 1431552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.655511827789327e-05,
+      "loss": 3.487,
+      "step": 1432064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.654674870918648e-05,
+      "loss": 3.5043,
+      "step": 1432576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6538362761675963e-05,
+      "loss": 3.5006,
+      "step": 1433088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6529976814165443e-05,
+      "loss": 3.487,
+      "step": 1433600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6521590866654916e-05,
+      "loss": 3.4915,
+      "step": 1434112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.651322129794813e-05,
+      "loss": 3.4908,
+      "step": 1434624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.650483535043761e-05,
+      "loss": 3.4906,
+      "step": 1435136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.649644940292709e-05,
+      "loss": 3.4839,
+      "step": 1435648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6488063455416568e-05,
+      "loss": 3.4966,
+      "step": 1436160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6479693886709784e-05,
+      "loss": 3.4871,
+      "step": 1436672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6471307939199264e-05,
+      "loss": 3.4944,
+      "step": 1437184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6462921991688744e-05,
+      "loss": 3.4813,
+      "step": 1437696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6454536044178217e-05,
+      "loss": 3.4904,
+      "step": 1438208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6446166475471433e-05,
+      "loss": 3.4848,
+      "step": 1438720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6437780527960916e-05,
+      "loss": 3.4878,
+      "step": 1439232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.642939458045039e-05,
+      "loss": 3.4871,
+      "step": 1439744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.642100863293987e-05,
+      "loss": 3.5071,
+      "step": 1440256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6412639064233085e-05,
+      "loss": 3.4942,
+      "step": 1440768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6404253116722565e-05,
+      "loss": 3.4896,
+      "step": 1441280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6395867169212042e-05,
+      "loss": 3.4811,
+      "step": 1441792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6387481221701522e-05,
+      "loss": 3.4978,
+      "step": 1442304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6379111652994738e-05,
+      "loss": 3.4959,
+      "step": 1442816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.637072570548421e-05,
+      "loss": 3.491,
+      "step": 1443328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.636233975797369e-05,
+      "loss": 3.4873,
+      "step": 1443840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.635395381046317e-05,
+      "loss": 3.4958,
+      "step": 1444352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6345584241756387e-05,
+      "loss": 3.4977,
+      "step": 1444864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6337198294245863e-05,
+      "loss": 3.4903,
+      "step": 1445376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6328812346735343e-05,
+      "loss": 3.4906,
+      "step": 1445888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6320426399224823e-05,
+      "loss": 3.4909,
+      "step": 1446400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.631205683051804e-05,
+      "loss": 3.4874,
+      "step": 1446912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6303670883007515e-05,
+      "loss": 3.4812,
+      "step": 1447424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6295284935496995e-05,
+      "loss": 3.4966,
+      "step": 1447936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6286898987986475e-05,
+      "loss": 3.4948,
+      "step": 1448448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6278529419279685e-05,
+      "loss": 3.4918,
+      "step": 1448960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6270143471769164e-05,
+      "loss": 3.4859,
+      "step": 1449472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.626177390306238e-05,
+      "loss": 3.4845,
+      "step": 1449984
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8483104705810547,
+      "eval_runtime": 310.9632,
+      "eval_samples_per_second": 1227.126,
+      "eval_steps_per_second": 38.349,
+      "step": 1450080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.625338795555186e-05,
+      "loss": 3.4826,
+      "step": 1450496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6245002008041337e-05,
+      "loss": 3.4904,
+      "step": 1451008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6236616060530817e-05,
+      "loss": 3.503,
+      "step": 1451520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6228230113020297e-05,
+      "loss": 3.4911,
+      "step": 1452032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6219860544313513e-05,
+      "loss": 3.4993,
+      "step": 1452544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6211474596802986e-05,
+      "loss": 3.4876,
+      "step": 1453056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.620308864929247e-05,
+      "loss": 3.4894,
+      "step": 1453568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.619470270178195e-05,
+      "loss": 3.4787,
+      "step": 1454080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6186333133075158e-05,
+      "loss": 3.4999,
+      "step": 1454592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6177947185564638e-05,
+      "loss": 3.4956,
+      "step": 1455104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6169561238054118e-05,
+      "loss": 3.495,
+      "step": 1455616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6161191669347334e-05,
+      "loss": 3.4984,
+      "step": 1456128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6152822100640546e-05,
+      "loss": 3.4831,
+      "step": 1456640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6144436153130026e-05,
+      "loss": 3.4835,
+      "step": 1457152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6136050205619506e-05,
+      "loss": 3.4926,
+      "step": 1457664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6127664258108986e-05,
+      "loss": 3.4832,
+      "step": 1458176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.611927831059846e-05,
+      "loss": 3.482,
+      "step": 1458688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.611089236308794e-05,
+      "loss": 3.4795,
+      "step": 1459200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6102506415577423e-05,
+      "loss": 3.4818,
+      "step": 1459712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6094120468066903e-05,
+      "loss": 3.5089,
+      "step": 1460224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6085750899360112e-05,
+      "loss": 3.4898,
+      "step": 1460736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6077364951849592e-05,
+      "loss": 3.4912,
+      "step": 1461248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6068979004339072e-05,
+      "loss": 3.487,
+      "step": 1461760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6060593056828552e-05,
+      "loss": 3.4931,
+      "step": 1462272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6052223488121764e-05,
+      "loss": 3.4745,
+      "step": 1462784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6043837540611244e-05,
+      "loss": 3.4795,
+      "step": 1463296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6035451593100724e-05,
+      "loss": 3.482,
+      "step": 1463808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6027082024393933e-05,
+      "loss": 3.4931,
+      "step": 1464320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6018696076883413e-05,
+      "loss": 3.4763,
+      "step": 1464832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6010310129372893e-05,
+      "loss": 3.4832,
+      "step": 1465344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6001924181862376e-05,
+      "loss": 3.4886,
+      "step": 1465856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5993554613155586e-05,
+      "loss": 3.4945,
+      "step": 1466368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5985168665645066e-05,
+      "loss": 3.4806,
+      "step": 1466880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5976782718134546e-05,
+      "loss": 3.4873,
+      "step": 1467392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5968396770624025e-05,
+      "loss": 3.4877,
+      "step": 1467904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5960027201917238e-05,
+      "loss": 3.4895,
+      "step": 1468416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5951641254406718e-05,
+      "loss": 3.4736,
+      "step": 1468928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5943255306896198e-05,
+      "loss": 3.4842,
+      "step": 1469440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5934869359385678e-05,
+      "loss": 3.4702,
+      "step": 1469952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5926499790678887e-05,
+      "loss": 3.4876,
+      "step": 1470464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5918113843168367e-05,
+      "loss": 3.4769,
+      "step": 1470976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5909727895657847e-05,
+      "loss": 3.4885,
+      "step": 1471488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.590134194814733e-05,
+      "loss": 3.4828,
+      "step": 1472000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.589297237944054e-05,
+      "loss": 3.4811,
+      "step": 1472512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.588458643193002e-05,
+      "loss": 3.4896,
+      "step": 1473024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.58762004844195e-05,
+      "loss": 3.4876,
+      "step": 1473536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.586781453690898e-05,
+      "loss": 3.4866,
+      "step": 1474048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.585944496820219e-05,
+      "loss": 3.4708,
+      "step": 1474560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.585105902069167e-05,
+      "loss": 3.4769,
+      "step": 1475072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.584267307318115e-05,
+      "loss": 3.4931,
+      "step": 1475584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.583428712567063e-05,
+      "loss": 3.4841,
+      "step": 1476096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5825933935767576e-05,
+      "loss": 3.478,
+      "step": 1476608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5817547988257053e-05,
+      "loss": 3.4783,
+      "step": 1477120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5809162040746533e-05,
+      "loss": 3.4817,
+      "step": 1477632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5800776093236013e-05,
+      "loss": 3.4681,
+      "step": 1478144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5792390145725493e-05,
+      "loss": 3.4865,
+      "step": 1478656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5784004198214973e-05,
+      "loss": 3.4838,
+      "step": 1479168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5775618250704453e-05,
+      "loss": 3.4882,
+      "step": 1479680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5767232303193933e-05,
+      "loss": 3.4916,
+      "step": 1480192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5758862734487145e-05,
+      "loss": 3.4742,
+      "step": 1480704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5750476786976625e-05,
+      "loss": 3.4781,
+      "step": 1481216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5742090839466105e-05,
+      "loss": 3.4821,
+      "step": 1481728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5733704891955585e-05,
+      "loss": 3.4716,
+      "step": 1482240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5725335323248794e-05,
+      "loss": 3.4688,
+      "step": 1482752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5716949375738274e-05,
+      "loss": 3.48,
+      "step": 1483264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5708563428227754e-05,
+      "loss": 3.4807,
+      "step": 1483776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5700177480717234e-05,
+      "loss": 3.4697,
+      "step": 1484288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5691807912010447e-05,
+      "loss": 3.4664,
+      "step": 1484800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5683421964499927e-05,
+      "loss": 3.482,
+      "step": 1485312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5675036016989407e-05,
+      "loss": 3.4896,
+      "step": 1485824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5666650069478886e-05,
+      "loss": 3.4791,
+      "step": 1486336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5658280500772096e-05,
+      "loss": 3.4863,
+      "step": 1486848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.564989455326158e-05,
+      "loss": 3.4934,
+      "step": 1487360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.564150860575106e-05,
+      "loss": 3.4869,
+      "step": 1487872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.563312265824054e-05,
+      "loss": 3.4952,
+      "step": 1488384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5624753089533748e-05,
+      "loss": 3.4727,
+      "step": 1488896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5616367142023228e-05,
+      "loss": 3.4879,
+      "step": 1489408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5607981194512708e-05,
+      "loss": 3.4877,
+      "step": 1489920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5599595247002188e-05,
+      "loss": 3.475,
+      "step": 1490432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.55912256782954e-05,
+      "loss": 3.491,
+      "step": 1490944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.558283973078488e-05,
+      "loss": 3.4834,
+      "step": 1491456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.557445378327436e-05,
+      "loss": 3.4956,
+      "step": 1491968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.556606783576384e-05,
+      "loss": 3.4841,
+      "step": 1492480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.555769826705705e-05,
+      "loss": 3.4638,
+      "step": 1492992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5549328698350262e-05,
+      "loss": 3.474,
+      "step": 1493504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.554094275083974e-05,
+      "loss": 3.4791,
+      "step": 1494016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.553255680332922e-05,
+      "loss": 3.4846,
+      "step": 1494528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.55241708558187e-05,
+      "loss": 3.4854,
+      "step": 1495040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.551578490830818e-05,
+      "loss": 3.4782,
+      "step": 1495552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.550739896079766e-05,
+      "loss": 3.4785,
+      "step": 1496064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.549901301328714e-05,
+      "loss": 3.4722,
+      "step": 1496576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5490627065776625e-05,
+      "loss": 3.4829,
+      "step": 1497088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5482257497069834e-05,
+      "loss": 3.4734,
+      "step": 1497600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5473871549559314e-05,
+      "loss": 3.4847,
+      "step": 1498112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5465485602048794e-05,
+      "loss": 3.4916,
+      "step": 1498624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5457099654538274e-05,
+      "loss": 3.4862,
+      "step": 1499136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5448730085831486e-05,
+      "loss": 3.4753,
+      "step": 1499648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5440344138320966e-05,
+      "loss": 3.4859,
+      "step": 1500160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5431958190810446e-05,
+      "loss": 3.4779,
+      "step": 1500672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5423572243299926e-05,
+      "loss": 3.4875,
+      "step": 1501184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5415202674593135e-05,
+      "loss": 3.48,
+      "step": 1501696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5406816727082615e-05,
+      "loss": 3.4783,
+      "step": 1502208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5398430779572095e-05,
+      "loss": 3.4708,
+      "step": 1502720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.539004483206158e-05,
+      "loss": 3.4892,
+      "step": 1503232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5381675263354788e-05,
+      "loss": 3.4698,
+      "step": 1503744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5373289315844268e-05,
+      "loss": 3.4676,
+      "step": 1504256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5364903368333747e-05,
+      "loss": 3.4729,
+      "step": 1504768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5356517420823227e-05,
+      "loss": 3.4816,
+      "step": 1505280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.534814785211644e-05,
+      "loss": 3.473,
+      "step": 1505792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.533976190460592e-05,
+      "loss": 3.4886,
+      "step": 1506304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.53313759570954e-05,
+      "loss": 3.4738,
+      "step": 1506816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.532299000958488e-05,
+      "loss": 3.477,
+      "step": 1507328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.531462044087809e-05,
+      "loss": 3.4775,
+      "step": 1507840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.530623449336757e-05,
+      "loss": 3.4699,
+      "step": 1508352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.529784854585705e-05,
+      "loss": 3.4951,
+      "step": 1508864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5289462598346532e-05,
+      "loss": 3.4907,
+      "step": 1509376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.528109302963974e-05,
+      "loss": 3.4781,
+      "step": 1509888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.527270708212922e-05,
+      "loss": 3.4799,
+      "step": 1510400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.52643211346187e-05,
+      "loss": 3.4738,
+      "step": 1510912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.525593518710818e-05,
+      "loss": 3.4795,
+      "step": 1511424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5247565618401394e-05,
+      "loss": 3.4718,
+      "step": 1511936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5239179670890874e-05,
+      "loss": 3.4897,
+      "step": 1512448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5230793723380353e-05,
+      "loss": 3.4731,
+      "step": 1512960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5222407775869833e-05,
+      "loss": 3.4792,
+      "step": 1513472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5214038207163043e-05,
+      "loss": 3.4739,
+      "step": 1513984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5205652259652522e-05,
+      "loss": 3.4787,
+      "step": 1514496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5197266312142002e-05,
+      "loss": 3.4718,
+      "step": 1515008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5188880364631486e-05,
+      "loss": 3.476,
+      "step": 1515520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5180510795924695e-05,
+      "loss": 3.4744,
+      "step": 1516032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5172124848414175e-05,
+      "loss": 3.4926,
+      "step": 1516544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5163738900903655e-05,
+      "loss": 3.4837,
+      "step": 1517056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.515535295339313e-05,
+      "loss": 3.4811,
+      "step": 1517568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5146983384686347e-05,
+      "loss": 3.4646,
+      "step": 1518080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5138597437175827e-05,
+      "loss": 3.4862,
+      "step": 1518592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5130211489665307e-05,
+      "loss": 3.483,
+      "step": 1519104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.512182554215478e-05,
+      "loss": 3.4819,
+      "step": 1519616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5113455973447996e-05,
+      "loss": 3.4745,
+      "step": 1520128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5105070025937476e-05,
+      "loss": 3.482,
+      "step": 1520640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5096684078426956e-05,
+      "loss": 3.4843,
+      "step": 1521152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5088298130916433e-05,
+      "loss": 3.4827,
+      "step": 1521664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.507992856220965e-05,
+      "loss": 3.4758,
+      "step": 1522176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.507154261469913e-05,
+      "loss": 3.4815,
+      "step": 1522688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.50631566671886e-05,
+      "loss": 3.4736,
+      "step": 1523200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5054770719678085e-05,
+      "loss": 3.47,
+      "step": 1523712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.50464011509713e-05,
+      "loss": 3.4815,
+      "step": 1524224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.503801520346078e-05,
+      "loss": 3.4884,
+      "step": 1524736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5029629255950254e-05,
+      "loss": 3.4767,
+      "step": 1525248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5021243308439734e-05,
+      "loss": 3.4725,
+      "step": 1525760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.501289011853668e-05,
+      "loss": 3.4798,
+      "step": 1526272
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8489973545074463,
+      "eval_runtime": 304.6768,
+      "eval_samples_per_second": 1252.445,
+      "eval_steps_per_second": 39.14,
+      "step": 1526400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5004504171026162e-05,
+      "loss": 3.4716,
+      "step": 1526784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.499611822351564e-05,
+      "loss": 3.4802,
+      "step": 1527296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.498773227600512e-05,
+      "loss": 3.487,
+      "step": 1527808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.49793463284946e-05,
+      "loss": 3.4804,
+      "step": 1528320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4970976759787808e-05,
+      "loss": 3.4885,
+      "step": 1528832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4962590812277288e-05,
+      "loss": 3.4767,
+      "step": 1529344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.495420486476677e-05,
+      "loss": 3.4776,
+      "step": 1529856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.494581891725625e-05,
+      "loss": 3.4676,
+      "step": 1530368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.493744934854946e-05,
+      "loss": 3.4891,
+      "step": 1530880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.492906340103894e-05,
+      "loss": 3.482,
+      "step": 1531392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.492067745352842e-05,
+      "loss": 3.479,
+      "step": 1531904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.49122915060179e-05,
+      "loss": 3.4882,
+      "step": 1532416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4903938316114845e-05,
+      "loss": 3.4687,
+      "step": 1532928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4895552368604325e-05,
+      "loss": 3.478,
+      "step": 1533440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4887166421093805e-05,
+      "loss": 3.4791,
+      "step": 1533952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.487878047358328e-05,
+      "loss": 3.4695,
+      "step": 1534464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.487039452607276e-05,
+      "loss": 3.4694,
+      "step": 1534976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.486200857856224e-05,
+      "loss": 3.4686,
+      "step": 1535488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4853622631051725e-05,
+      "loss": 3.4728,
+      "step": 1536000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4845236683541205e-05,
+      "loss": 3.4924,
+      "step": 1536512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4836867114834414e-05,
+      "loss": 3.4776,
+      "step": 1537024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4828481167323894e-05,
+      "loss": 3.4777,
+      "step": 1537536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4820095219813374e-05,
+      "loss": 3.4752,
+      "step": 1538048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4811725651106586e-05,
+      "loss": 3.4804,
+      "step": 1538560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4803339703596066e-05,
+      "loss": 3.4654,
+      "step": 1539072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4794953756085546e-05,
+      "loss": 3.4698,
+      "step": 1539584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4786567808575026e-05,
+      "loss": 3.4692,
+      "step": 1540096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4778198239868235e-05,
+      "loss": 3.4816,
+      "step": 1540608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4769812292357715e-05,
+      "loss": 3.4636,
+      "step": 1541120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4761426344847195e-05,
+      "loss": 3.4737,
+      "step": 1541632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.475304039733668e-05,
+      "loss": 3.4769,
+      "step": 1542144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4744670828629888e-05,
+      "loss": 3.4811,
+      "step": 1542656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4736284881119368e-05,
+      "loss": 3.4696,
+      "step": 1543168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4727898933608848e-05,
+      "loss": 3.4741,
+      "step": 1543680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4719512986098328e-05,
+      "loss": 3.4748,
+      "step": 1544192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.471114341739154e-05,
+      "loss": 3.4784,
+      "step": 1544704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.470275746988102e-05,
+      "loss": 3.4676,
+      "step": 1545216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46943715223705e-05,
+      "loss": 3.4681,
+      "step": 1545728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.468598557485998e-05,
+      "loss": 3.4596,
+      "step": 1546240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.467761600615319e-05,
+      "loss": 3.4749,
+      "step": 1546752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.466923005864267e-05,
+      "loss": 3.4638,
+      "step": 1547264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.466084411113215e-05,
+      "loss": 3.4763,
+      "step": 1547776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4652458163621632e-05,
+      "loss": 3.4708,
+      "step": 1548288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.464408859491484e-05,
+      "loss": 3.4709,
+      "step": 1548800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.463570264740432e-05,
+      "loss": 3.4756,
+      "step": 1549312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46273166998938e-05,
+      "loss": 3.4754,
+      "step": 1549824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.461893075238328e-05,
+      "loss": 3.4755,
+      "step": 1550336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.461054480487276e-05,
+      "loss": 3.4619,
+      "step": 1550848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.460215885736224e-05,
+      "loss": 3.4579,
+      "step": 1551360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.459377290985172e-05,
+      "loss": 3.482,
+      "step": 1551872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.45853869623412e-05,
+      "loss": 3.4717,
+      "step": 1552384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4577033772438143e-05,
+      "loss": 3.4666,
+      "step": 1552896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4568647824927623e-05,
+      "loss": 3.4656,
+      "step": 1553408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4560261877417103e-05,
+      "loss": 3.4696,
+      "step": 1553920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4551875929906586e-05,
+      "loss": 3.4558,
+      "step": 1554432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4543489982396062e-05,
+      "loss": 3.474,
+      "step": 1554944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4535104034885542e-05,
+      "loss": 3.4728,
+      "step": 1555456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4526718087375022e-05,
+      "loss": 3.4733,
+      "step": 1555968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4518332139864502e-05,
+      "loss": 3.4814,
+      "step": 1556480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.450996257115771e-05,
+      "loss": 3.4661,
+      "step": 1556992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4501576623647195e-05,
+      "loss": 3.4637,
+      "step": 1557504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4493190676136675e-05,
+      "loss": 3.4771,
+      "step": 1558016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4484804728626155e-05,
+      "loss": 3.4545,
+      "step": 1558528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4476435159919364e-05,
+      "loss": 3.4582,
+      "step": 1559040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4468049212408844e-05,
+      "loss": 3.4692,
+      "step": 1559552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4459663264898324e-05,
+      "loss": 3.4712,
+      "step": 1560064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4451277317387804e-05,
+      "loss": 3.4581,
+      "step": 1560576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4442907748681016e-05,
+      "loss": 3.4566,
+      "step": 1561088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4434521801170496e-05,
+      "loss": 3.4685,
+      "step": 1561600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4426135853659976e-05,
+      "loss": 3.4739,
+      "step": 1562112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4417749906149456e-05,
+      "loss": 3.4732,
+      "step": 1562624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4409380337442665e-05,
+      "loss": 3.4707,
+      "step": 1563136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.440099438993215e-05,
+      "loss": 3.4837,
+      "step": 1563648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.439260844242163e-05,
+      "loss": 3.4743,
+      "step": 1564160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.438422249491111e-05,
+      "loss": 3.4889,
+      "step": 1564672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4375852926204317e-05,
+      "loss": 3.4601,
+      "step": 1565184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4367466978693797e-05,
+      "loss": 3.477,
+      "step": 1565696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4359081031183277e-05,
+      "loss": 3.4748,
+      "step": 1566208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4350695083672757e-05,
+      "loss": 3.4613,
+      "step": 1566720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.434232551496597e-05,
+      "loss": 3.4836,
+      "step": 1567232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.433393956745545e-05,
+      "loss": 3.4706,
+      "step": 1567744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.432555361994493e-05,
+      "loss": 3.4796,
+      "step": 1568256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.431716767243441e-05,
+      "loss": 3.4758,
+      "step": 1568768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.430879810372762e-05,
+      "loss": 3.4549,
+      "step": 1569280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4300412156217102e-05,
+      "loss": 3.4642,
+      "step": 1569792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4292026208706582e-05,
+      "loss": 3.4645,
+      "step": 1570304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.428365663999979e-05,
+      "loss": 3.4752,
+      "step": 1570816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.427527069248927e-05,
+      "loss": 3.4737,
+      "step": 1571328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.426688474497875e-05,
+      "loss": 3.4653,
+      "step": 1571840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.425849879746823e-05,
+      "loss": 3.4702,
+      "step": 1572352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.425011284995771e-05,
+      "loss": 3.4637,
+      "step": 1572864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.424172690244719e-05,
+      "loss": 3.4677,
+      "step": 1573376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.423334095493667e-05,
+      "loss": 3.4614,
+      "step": 1573888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.422495500742615e-05,
+      "loss": 3.4754,
+      "step": 1574400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.421658543871936e-05,
+      "loss": 3.48,
+      "step": 1574912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.420819949120884e-05,
+      "loss": 3.4734,
+      "step": 1575424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.419981354369832e-05,
+      "loss": 3.4641,
+      "step": 1575936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4191427596187803e-05,
+      "loss": 3.4715,
+      "step": 1576448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4183058027481012e-05,
+      "loss": 3.4698,
+      "step": 1576960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4174672079970492e-05,
+      "loss": 3.4772,
+      "step": 1577472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4166286132459972e-05,
+      "loss": 3.4711,
+      "step": 1577984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4157900184949452e-05,
+      "loss": 3.4649,
+      "step": 1578496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4149530616242665e-05,
+      "loss": 3.4587,
+      "step": 1579008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4141144668732145e-05,
+      "loss": 3.4746,
+      "step": 1579520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4132758721221625e-05,
+      "loss": 3.4612,
+      "step": 1580032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4124389152514834e-05,
+      "loss": 3.46,
+      "step": 1580544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4116003205004314e-05,
+      "loss": 3.4565,
+      "step": 1581056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4107617257493794e-05,
+      "loss": 3.4705,
+      "step": 1581568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4099231309983274e-05,
+      "loss": 3.4559,
+      "step": 1582080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4090861741276486e-05,
+      "loss": 3.4801,
+      "step": 1582592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4082475793765966e-05,
+      "loss": 3.4608,
+      "step": 1583104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4074089846255446e-05,
+      "loss": 3.4705,
+      "step": 1583616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4065703898744926e-05,
+      "loss": 3.4659,
+      "step": 1584128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4057334330038135e-05,
+      "loss": 3.4589,
+      "step": 1584640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.404894838252762e-05,
+      "loss": 3.4836,
+      "step": 1585152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4040562435017098e-05,
+      "loss": 3.479,
+      "step": 1585664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4032176487506578e-05,
+      "loss": 3.4653,
+      "step": 1586176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4023806918799787e-05,
+      "loss": 3.4685,
+      "step": 1586688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4015420971289267e-05,
+      "loss": 3.4635,
+      "step": 1587200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4007035023778747e-05,
+      "loss": 3.4694,
+      "step": 1587712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3998649076268227e-05,
+      "loss": 3.4557,
+      "step": 1588224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.399027950756144e-05,
+      "loss": 3.4775,
+      "step": 1588736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.398189356005092e-05,
+      "loss": 3.4671,
+      "step": 1589248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.39735076125404e-05,
+      "loss": 3.4648,
+      "step": 1589760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.396512166502988e-05,
+      "loss": 3.463,
+      "step": 1590272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.395675209632309e-05,
+      "loss": 3.4667,
+      "step": 1590784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3948366148812572e-05,
+      "loss": 3.459,
+      "step": 1591296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3939980201302052e-05,
+      "loss": 3.4666,
+      "step": 1591808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3931594253791532e-05,
+      "loss": 3.462,
+      "step": 1592320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.392322468508474e-05,
+      "loss": 3.478,
+      "step": 1592832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.391483873757422e-05,
+      "loss": 3.4723,
+      "step": 1593344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.39064527900637e-05,
+      "loss": 3.4718,
+      "step": 1593856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.389806684255318e-05,
+      "loss": 3.4544,
+      "step": 1594368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3889697273846393e-05,
+      "loss": 3.4739,
+      "step": 1594880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3881311326335873e-05,
+      "loss": 3.4725,
+      "step": 1595392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3872925378825353e-05,
+      "loss": 3.4702,
+      "step": 1595904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3864539431314833e-05,
+      "loss": 3.4625,
+      "step": 1596416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3856169862608042e-05,
+      "loss": 3.4681,
+      "step": 1596928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3847783915097526e-05,
+      "loss": 3.4732,
+      "step": 1597440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3839397967587006e-05,
+      "loss": 3.4769,
+      "step": 1597952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3831012020076486e-05,
+      "loss": 3.4614,
+      "step": 1598464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3822642451369695e-05,
+      "loss": 3.4708,
+      "step": 1598976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3814256503859175e-05,
+      "loss": 3.4593,
+      "step": 1599488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3805870556348655e-05,
+      "loss": 3.4608,
+      "step": 1600000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3797484608838135e-05,
+      "loss": 3.4738,
+      "step": 1600512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3789115040131347e-05,
+      "loss": 3.4682,
+      "step": 1601024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3780729092620827e-05,
+      "loss": 3.4675,
+      "step": 1601536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3772343145110307e-05,
+      "loss": 3.4608,
+      "step": 1602048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3763957197599783e-05,
+      "loss": 3.4686,
+      "step": 1602560
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.85150146484375,
+      "eval_runtime": 305.1968,
+      "eval_samples_per_second": 1250.311,
+      "eval_steps_per_second": 39.073,
+      "step": 1602720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3755587628892996e-05,
+      "loss": 3.4651,
+      "step": 1603072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.374720168138248e-05,
+      "loss": 3.4651,
+      "step": 1603584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.373881573387196e-05,
+      "loss": 3.4763,
+      "step": 1604096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3730429786361436e-05,
+      "loss": 3.4715,
+      "step": 1604608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3722043838850916e-05,
+      "loss": 3.473,
+      "step": 1605120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3713657891340396e-05,
+      "loss": 3.4693,
+      "step": 1605632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3705271943829876e-05,
+      "loss": 3.4617,
+      "step": 1606144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3696902375123088e-05,
+      "loss": 3.4573,
+      "step": 1606656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3688516427612568e-05,
+      "loss": 3.4723,
+      "step": 1607168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3680130480102048e-05,
+      "loss": 3.4748,
+      "step": 1607680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3671744532591528e-05,
+      "loss": 3.4674,
+      "step": 1608192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3663374963884737e-05,
+      "loss": 3.4759,
+      "step": 1608704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.365500539517795e-05,
+      "loss": 3.4599,
+      "step": 1609216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3646619447667433e-05,
+      "loss": 3.4641,
+      "step": 1609728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.363823350015691e-05,
+      "loss": 3.4709,
+      "step": 1610240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.362984755264639e-05,
+      "loss": 3.4565,
+      "step": 1610752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.362146160513587e-05,
+      "loss": 3.4596,
+      "step": 1611264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.361307565762535e-05,
+      "loss": 3.459,
+      "step": 1611776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.360468971011483e-05,
+      "loss": 3.4618,
+      "step": 1612288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.359630376260431e-05,
+      "loss": 3.4803,
+      "step": 1612800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3587934193897522e-05,
+      "loss": 3.4667,
+      "step": 1613312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3579548246387002e-05,
+      "loss": 3.4695,
+      "step": 1613824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3571162298876482e-05,
+      "loss": 3.462,
+      "step": 1614336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.356277635136596e-05,
+      "loss": 3.4687,
+      "step": 1614848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.355440678265917e-05,
+      "loss": 3.4541,
+      "step": 1615360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.354602083514865e-05,
+      "loss": 3.4584,
+      "step": 1615872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3537634887638134e-05,
+      "loss": 3.4591,
+      "step": 1616384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3529265318931343e-05,
+      "loss": 3.4669,
+      "step": 1616896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3520879371420823e-05,
+      "loss": 3.4533,
+      "step": 1617408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3512493423910303e-05,
+      "loss": 3.4613,
+      "step": 1617920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3504107476399783e-05,
+      "loss": 3.4674,
+      "step": 1618432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3495737907692996e-05,
+      "loss": 3.4701,
+      "step": 1618944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3487351960182475e-05,
+      "loss": 3.4608,
+      "step": 1619456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3478966012671955e-05,
+      "loss": 3.4608,
+      "step": 1619968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3470580065161435e-05,
+      "loss": 3.4647,
+      "step": 1620480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3462210496454644e-05,
+      "loss": 3.4646,
+      "step": 1620992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3453824548944124e-05,
+      "loss": 3.4575,
+      "step": 1621504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3445438601433604e-05,
+      "loss": 3.4548,
+      "step": 1622016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3437052653923088e-05,
+      "loss": 3.4496,
+      "step": 1622528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3428683085216297e-05,
+      "loss": 3.4611,
+      "step": 1623040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3420297137705777e-05,
+      "loss": 3.4569,
+      "step": 1623552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3411911190195257e-05,
+      "loss": 3.4644,
+      "step": 1624064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3403525242684737e-05,
+      "loss": 3.4578,
+      "step": 1624576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.339515567397795e-05,
+      "loss": 3.4602,
+      "step": 1625088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.338676972646743e-05,
+      "loss": 3.4683,
+      "step": 1625600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.337838377895691e-05,
+      "loss": 3.461,
+      "step": 1626112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3369997831446386e-05,
+      "loss": 3.4644,
+      "step": 1626624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3361611883935866e-05,
+      "loss": 3.4499,
+      "step": 1627136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3353225936425346e-05,
+      "loss": 3.4467,
+      "step": 1627648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3344839988914826e-05,
+      "loss": 3.4722,
+      "step": 1628160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3336454041404305e-05,
+      "loss": 3.4614,
+      "step": 1628672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3328084472697518e-05,
+      "loss": 3.4549,
+      "step": 1629184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.331971490399073e-05,
+      "loss": 3.4545,
+      "step": 1629696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.331132895648021e-05,
+      "loss": 3.4593,
+      "step": 1630208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3302943008969687e-05,
+      "loss": 3.4424,
+      "step": 1630720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3294557061459167e-05,
+      "loss": 3.4616,
+      "step": 1631232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.328617111394865e-05,
+      "loss": 3.4606,
+      "step": 1631744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.327778516643813e-05,
+      "loss": 3.4571,
+      "step": 1632256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.326939921892761e-05,
+      "loss": 3.4744,
+      "step": 1632768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.326102965022082e-05,
+      "loss": 3.4528,
+      "step": 1633280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.32526437027103e-05,
+      "loss": 3.4528,
+      "step": 1633792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.324425775519978e-05,
+      "loss": 3.4654,
+      "step": 1634304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.323587180768926e-05,
+      "loss": 3.4488,
+      "step": 1634816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.322750223898247e-05,
+      "loss": 3.447,
+      "step": 1635328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.321911629147195e-05,
+      "loss": 3.4562,
+      "step": 1635840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.321073034396143e-05,
+      "loss": 3.4578,
+      "step": 1636352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.320234439645091e-05,
+      "loss": 3.4498,
+      "step": 1636864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.319397482774412e-05,
+      "loss": 3.4473,
+      "step": 1637376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3185588880233604e-05,
+      "loss": 3.4529,
+      "step": 1637888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3177202932723084e-05,
+      "loss": 3.46,
+      "step": 1638400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3168816985212564e-05,
+      "loss": 3.4627,
+      "step": 1638912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3160447416505773e-05,
+      "loss": 3.4593,
+      "step": 1639424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3152061468995253e-05,
+      "loss": 3.4693,
+      "step": 1639936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3143675521484733e-05,
+      "loss": 3.4651,
+      "step": 1640448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3135289573974213e-05,
+      "loss": 3.4759,
+      "step": 1640960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3126920005267425e-05,
+      "loss": 3.448,
+      "step": 1641472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3118534057756905e-05,
+      "loss": 3.4698,
+      "step": 1641984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3110148110246385e-05,
+      "loss": 3.4644,
+      "step": 1642496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3101762162735862e-05,
+      "loss": 3.4467,
+      "step": 1643008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3093392594029074e-05,
+      "loss": 3.4713,
+      "step": 1643520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3085006646518558e-05,
+      "loss": 3.4592,
+      "step": 1644032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3076620699008038e-05,
+      "loss": 3.4687,
+      "step": 1644544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3068234751497514e-05,
+      "loss": 3.467,
+      "step": 1645056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3059865182790727e-05,
+      "loss": 3.445,
+      "step": 1645568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3051479235280207e-05,
+      "loss": 3.4516,
+      "step": 1646080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.304310966657342e-05,
+      "loss": 3.4535,
+      "step": 1646592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.30347237190629e-05,
+      "loss": 3.4656,
+      "step": 1647104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.302633777155238e-05,
+      "loss": 3.4599,
+      "step": 1647616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.301795182404186e-05,
+      "loss": 3.4569,
+      "step": 1648128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3009565876531335e-05,
+      "loss": 3.4524,
+      "step": 1648640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3001179929020815e-05,
+      "loss": 3.4562,
+      "step": 1649152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2992793981510295e-05,
+      "loss": 3.458,
+      "step": 1649664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.298442441280351e-05,
+      "loss": 3.4495,
+      "step": 1650176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2976038465292988e-05,
+      "loss": 3.4666,
+      "step": 1650688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2967652517782468e-05,
+      "loss": 3.4674,
+      "step": 1651200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.295928294907568e-05,
+      "loss": 3.4679,
+      "step": 1651712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.295089700156516e-05,
+      "loss": 3.4484,
+      "step": 1652224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2942511054054637e-05,
+      "loss": 3.4619,
+      "step": 1652736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.293412510654412e-05,
+      "loss": 3.4554,
+      "step": 1653248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.29257391590336e-05,
+      "loss": 3.4624,
+      "step": 1653760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.291735321152308e-05,
+      "loss": 3.4613,
+      "step": 1654272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.290896726401256e-05,
+      "loss": 3.4578,
+      "step": 1654784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.290059769530577e-05,
+      "loss": 3.4478,
+      "step": 1655296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.289222812659898e-05,
+      "loss": 3.4623,
+      "step": 1655808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.288384217908846e-05,
+      "loss": 3.4525,
+      "step": 1656320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.287545623157794e-05,
+      "loss": 3.446,
+      "step": 1656832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.286707028406742e-05,
+      "loss": 3.4426,
+      "step": 1657344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.28586843365569e-05,
+      "loss": 3.4572,
+      "step": 1657856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.285029838904638e-05,
+      "loss": 3.4508,
+      "step": 1658368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.284191244153586e-05,
+      "loss": 3.4674,
+      "step": 1658880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.283352649402534e-05,
+      "loss": 3.4485,
+      "step": 1659392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2825156925318554e-05,
+      "loss": 3.4571,
+      "step": 1659904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2816770977808034e-05,
+      "loss": 3.4563,
+      "step": 1660416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2808385030297514e-05,
+      "loss": 3.4453,
+      "step": 1660928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.279999908278699e-05,
+      "loss": 3.4705,
+      "step": 1661440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2791629514080203e-05,
+      "loss": 3.4655,
+      "step": 1661952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2783243566569683e-05,
+      "loss": 3.4529,
+      "step": 1662464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2774857619059166e-05,
+      "loss": 3.4577,
+      "step": 1662976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2766471671548643e-05,
+      "loss": 3.4534,
+      "step": 1663488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2758102102841855e-05,
+      "loss": 3.4581,
+      "step": 1664000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2749716155331335e-05,
+      "loss": 3.4419,
+      "step": 1664512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2741330207820815e-05,
+      "loss": 3.4675,
+      "step": 1665024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.273294426031029e-05,
+      "loss": 3.4517,
+      "step": 1665536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2724574691603507e-05,
+      "loss": 3.4525,
+      "step": 1666048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2716188744092987e-05,
+      "loss": 3.4558,
+      "step": 1666560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2707802796582464e-05,
+      "loss": 3.4535,
+      "step": 1667072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2699416849071944e-05,
+      "loss": 3.4527,
+      "step": 1667584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2691047280365156e-05,
+      "loss": 3.4528,
+      "step": 1668096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2682661332854636e-05,
+      "loss": 3.4534,
+      "step": 1668608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2674275385344116e-05,
+      "loss": 3.4647,
+      "step": 1669120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2665889437833596e-05,
+      "loss": 3.4577,
+      "step": 1669632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.265751986912681e-05,
+      "loss": 3.4634,
+      "step": 1670144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.264913392161629e-05,
+      "loss": 3.4433,
+      "step": 1670656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2640747974105765e-05,
+      "loss": 3.4659,
+      "step": 1671168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2632362026595245e-05,
+      "loss": 3.4572,
+      "step": 1671680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.262399245788846e-05,
+      "loss": 3.4583,
+      "step": 1672192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2615606510377938e-05,
+      "loss": 3.4531,
+      "step": 1672704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2607220562867418e-05,
+      "loss": 3.455,
+      "step": 1673216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2598834615356898e-05,
+      "loss": 3.4637,
+      "step": 1673728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.259046504665011e-05,
+      "loss": 3.4658,
+      "step": 1674240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.258207909913959e-05,
+      "loss": 3.4493,
+      "step": 1674752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.257369315162907e-05,
+      "loss": 3.4574,
+      "step": 1675264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2565323582922282e-05,
+      "loss": 3.4522,
+      "step": 1675776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2556937635411762e-05,
+      "loss": 3.4468,
+      "step": 1676288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.254855168790124e-05,
+      "loss": 3.4636,
+      "step": 1676800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.254016574039072e-05,
+      "loss": 3.4563,
+      "step": 1677312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2531796171683935e-05,
+      "loss": 3.46,
+      "step": 1677824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.252341022417341e-05,
+      "loss": 3.444,
+      "step": 1678336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.251502427666289e-05,
+      "loss": 3.4598,
+      "step": 1678848
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.8525948524475098,
+      "eval_runtime": 305.3117,
+      "eval_samples_per_second": 1249.841,
+      "eval_steps_per_second": 39.058,
+      "step": 1679040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2506654707956104e-05,
+      "loss": 3.4512,
+      "step": 1679360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2498268760445584e-05,
+      "loss": 3.45,
+      "step": 1679872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.248988281293506e-05,
+      "loss": 3.4645,
+      "step": 1680384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2481496865424544e-05,
+      "loss": 3.4614,
+      "step": 1680896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2473127296717756e-05,
+      "loss": 3.4611,
+      "step": 1681408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2464741349207233e-05,
+      "loss": 3.4585,
+      "step": 1681920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2456355401696713e-05,
+      "loss": 3.4454,
+      "step": 1682432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2447985832989925e-05,
+      "loss": 3.4516,
+      "step": 1682944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2439599885479405e-05,
+      "loss": 3.4581,
+      "step": 1683456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2431213937968885e-05,
+      "loss": 3.4657,
+      "step": 1683968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2422827990458365e-05,
+      "loss": 3.4547,
+      "step": 1684480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2414442042947845e-05,
+      "loss": 3.4657,
+      "step": 1684992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2406072474241057e-05,
+      "loss": 3.4513,
+      "step": 1685504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2397686526730534e-05,
+      "loss": 3.4502,
+      "step": 1686016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2389300579220014e-05,
+      "loss": 3.4637,
+      "step": 1686528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2380914631709497e-05,
+      "loss": 3.445,
+      "step": 1687040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2372528684198977e-05,
+      "loss": 3.4479,
+      "step": 1687552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2364142736688457e-05,
+      "loss": 3.445,
+      "step": 1688064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2355756789177937e-05,
+      "loss": 3.4509,
+      "step": 1688576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2347370841667414e-05,
+      "loss": 3.4636,
+      "step": 1689088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2339001272960626e-05,
+      "loss": 3.4584,
+      "step": 1689600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2330615325450106e-05,
+      "loss": 3.4594,
+      "step": 1690112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.232222937793959e-05,
+      "loss": 3.4464,
+      "step": 1690624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2313843430429066e-05,
+      "loss": 3.4591,
+      "step": 1691136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.230547386172228e-05,
+      "loss": 3.4458,
+      "step": 1691648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2297104293015488e-05,
+      "loss": 3.4485,
+      "step": 1692160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2288718345504968e-05,
+      "loss": 3.4456,
+      "step": 1692672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.228033239799445e-05,
+      "loss": 3.4521,
+      "step": 1693184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.227194645048393e-05,
+      "loss": 3.4476,
+      "step": 1693696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.226356050297341e-05,
+      "loss": 3.4466,
+      "step": 1694208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.225519093426662e-05,
+      "loss": 3.4583,
+      "step": 1694720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.22468049867561e-05,
+      "loss": 3.4598,
+      "step": 1695232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.223841903924558e-05,
+      "loss": 3.4509,
+      "step": 1695744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.223003309173506e-05,
+      "loss": 3.453,
+      "step": 1696256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222164714422454e-05,
+      "loss": 3.4514,
+      "step": 1696768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.221326119671402e-05,
+      "loss": 3.4529,
+      "step": 1697280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.22048752492035e-05,
+      "loss": 3.4513,
+      "step": 1697792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.219648930169298e-05,
+      "loss": 3.4374,
+      "step": 1698304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.218811973298619e-05,
+      "loss": 3.4373,
+      "step": 1698816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.217973378547567e-05,
+      "loss": 3.4493,
+      "step": 1699328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2171347837965152e-05,
+      "loss": 3.4472,
+      "step": 1699840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2162961890454632e-05,
+      "loss": 3.4513,
+      "step": 1700352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.215459232174784e-05,
+      "loss": 3.4485,
+      "step": 1700864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2146222753041054e-05,
+      "loss": 3.4486,
+      "step": 1701376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2137836805530534e-05,
+      "loss": 3.4571,
+      "step": 1701888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2129450858020014e-05,
+      "loss": 3.4496,
+      "step": 1702400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2121064910509494e-05,
+      "loss": 3.4584,
+      "step": 1702912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2112678962998973e-05,
+      "loss": 3.4374,
+      "step": 1703424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2104293015488453e-05,
+      "loss": 3.4312,
+      "step": 1703936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2095907067977933e-05,
+      "loss": 3.4629,
+      "step": 1704448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2087521120467413e-05,
+      "loss": 3.4523,
+      "step": 1704960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2079151551760622e-05,
+      "loss": 3.4424,
+      "step": 1705472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2070781983053835e-05,
+      "loss": 3.4459,
+      "step": 1705984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2062396035543315e-05,
+      "loss": 3.4454,
+      "step": 1706496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2054010088032795e-05,
+      "loss": 3.4357,
+      "step": 1707008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2045624140522275e-05,
+      "loss": 3.4473,
+      "step": 1707520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2037238193011755e-05,
+      "loss": 3.4506,
+      "step": 1708032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2028852245501235e-05,
+      "loss": 3.4504,
+      "step": 1708544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2020466297990715e-05,
+      "loss": 3.461,
+      "step": 1709056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2012096729283927e-05,
+      "loss": 3.4409,
+      "step": 1709568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2003710781773407e-05,
+      "loss": 3.4431,
+      "step": 1710080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1995324834262887e-05,
+      "loss": 3.4583,
+      "step": 1710592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1986938886752367e-05,
+      "loss": 3.4365,
+      "step": 1711104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1978569318045576e-05,
+      "loss": 3.4315,
+      "step": 1711616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.197018337053506e-05,
+      "loss": 3.4443,
+      "step": 1712128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.196179742302454e-05,
+      "loss": 3.4451,
+      "step": 1712640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1953411475514016e-05,
+      "loss": 3.4368,
+      "step": 1713152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.194504190680723e-05,
+      "loss": 3.4416,
+      "step": 1713664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.193665595929671e-05,
+      "loss": 3.4385,
+      "step": 1714176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.192827001178619e-05,
+      "loss": 3.4558,
+      "step": 1714688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.191988406427567e-05,
+      "loss": 3.451,
+      "step": 1715200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.191151449556888e-05,
+      "loss": 3.4484,
+      "step": 1715712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.190312854805836e-05,
+      "loss": 3.4563,
+      "step": 1716224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.189474260054784e-05,
+      "loss": 3.4517,
+      "step": 1716736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1886356653037317e-05,
+      "loss": 3.4671,
+      "step": 1717248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.187798708433053e-05,
+      "loss": 3.439,
+      "step": 1717760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1869601136820013e-05,
+      "loss": 3.4574,
+      "step": 1718272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.186121518930949e-05,
+      "loss": 3.4526,
+      "step": 1718784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.185282924179897e-05,
+      "loss": 3.4364,
+      "step": 1719296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1844459673092182e-05,
+      "loss": 3.4564,
+      "step": 1719808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1836073725581662e-05,
+      "loss": 3.4496,
+      "step": 1720320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.182768777807114e-05,
+      "loss": 3.4574,
+      "step": 1720832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1819301830560622e-05,
+      "loss": 3.4568,
+      "step": 1721344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1810932261853834e-05,
+      "loss": 3.4373,
+      "step": 1721856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.180254631434331e-05,
+      "loss": 3.4324,
+      "step": 1722368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1794176745636524e-05,
+      "loss": 3.4423,
+      "step": 1722880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1785790798126003e-05,
+      "loss": 3.4556,
+      "step": 1723392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1777404850615483e-05,
+      "loss": 3.4463,
+      "step": 1723904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1769018903104963e-05,
+      "loss": 3.446,
+      "step": 1724416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1760632955594443e-05,
+      "loss": 3.4443,
+      "step": 1724928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1752247008083923e-05,
+      "loss": 3.4448,
+      "step": 1725440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1743861060573403e-05,
+      "loss": 3.4479,
+      "step": 1725952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1735475113062883e-05,
+      "loss": 3.4398,
+      "step": 1726464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1727105544356092e-05,
+      "loss": 3.4547,
+      "step": 1726976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1718719596845576e-05,
+      "loss": 3.4548,
+      "step": 1727488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1710350028138785e-05,
+      "loss": 3.4555,
+      "step": 1728000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1701964080628265e-05,
+      "loss": 3.4367,
+      "step": 1728512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1693578133117745e-05,
+      "loss": 3.4516,
+      "step": 1729024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1685192185607225e-05,
+      "loss": 3.4478,
+      "step": 1729536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1676806238096705e-05,
+      "loss": 3.4518,
+      "step": 1730048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1668420290586185e-05,
+      "loss": 3.4518,
+      "step": 1730560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1660034343075668e-05,
+      "loss": 3.4419,
+      "step": 1731072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1651664774368877e-05,
+      "loss": 3.442,
+      "step": 1731584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1643295205662086e-05,
+      "loss": 3.4495,
+      "step": 1732096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1634909258151566e-05,
+      "loss": 3.4398,
+      "step": 1732608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1626523310641046e-05,
+      "loss": 3.4445,
+      "step": 1733120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.161813736313053e-05,
+      "loss": 3.4258,
+      "step": 1733632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.160975141562001e-05,
+      "loss": 3.4499,
+      "step": 1734144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.160136546810949e-05,
+      "loss": 3.4412,
+      "step": 1734656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1592979520598966e-05,
+      "loss": 3.4504,
+      "step": 1735168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1584593573088446e-05,
+      "loss": 3.4387,
+      "step": 1735680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1576224004381658e-05,
+      "loss": 3.4502,
+      "step": 1736192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1567838056871138e-05,
+      "loss": 3.4469,
+      "step": 1736704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1559452109360618e-05,
+      "loss": 3.4342,
+      "step": 1737216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1551066161850098e-05,
+      "loss": 3.46,
+      "step": 1737728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.154269659314331e-05,
+      "loss": 3.453,
+      "step": 1738240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.153431064563279e-05,
+      "loss": 3.4459,
+      "step": 1738752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1525924698122267e-05,
+      "loss": 3.449,
+      "step": 1739264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1517538750611747e-05,
+      "loss": 3.4362,
+      "step": 1739776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1509169181904963e-05,
+      "loss": 3.4517,
+      "step": 1740288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.150078323439444e-05,
+      "loss": 3.4335,
+      "step": 1740800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.149239728688392e-05,
+      "loss": 3.4541,
+      "step": 1741312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1484027718177132e-05,
+      "loss": 3.4398,
+      "step": 1741824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1475641770666612e-05,
+      "loss": 3.4441,
+      "step": 1742336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1467255823156092e-05,
+      "loss": 3.441,
+      "step": 1742848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1458869875645572e-05,
+      "loss": 3.4441,
+      "step": 1743360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1450483928135052e-05,
+      "loss": 3.4418,
+      "step": 1743872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1442097980624532e-05,
+      "loss": 3.4389,
+      "step": 1744384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1433712033114012e-05,
+      "loss": 3.4435,
+      "step": 1744896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.142532608560349e-05,
+      "loss": 3.4524,
+      "step": 1745408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.14169565168967e-05,
+      "loss": 3.4444,
+      "step": 1745920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1408570569386184e-05,
+      "loss": 3.4567,
+      "step": 1746432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1400184621875664e-05,
+      "loss": 3.4309,
+      "step": 1746944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1391798674365144e-05,
+      "loss": 3.455,
+      "step": 1747456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1383429105658353e-05,
+      "loss": 3.4501,
+      "step": 1747968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1375043158147833e-05,
+      "loss": 3.4447,
+      "step": 1748480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1366657210637313e-05,
+      "loss": 3.4421,
+      "step": 1748992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1358271263126793e-05,
+      "loss": 3.4462,
+      "step": 1749504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1349901694420005e-05,
+      "loss": 3.4487,
+      "step": 1750016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1341515746909485e-05,
+      "loss": 3.4593,
+      "step": 1750528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1333129799398965e-05,
+      "loss": 3.4366,
+      "step": 1751040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1324743851888445e-05,
+      "loss": 3.4491,
+      "step": 1751552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1316374283181654e-05,
+      "loss": 3.4432,
+      "step": 1752064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1307988335671138e-05,
+      "loss": 3.4377,
+      "step": 1752576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1299602388160618e-05,
+      "loss": 3.447,
+      "step": 1753088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1291232819453827e-05,
+      "loss": 3.4467,
+      "step": 1753600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1282846871943307e-05,
+      "loss": 3.4488,
+      "step": 1754112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1274460924432787e-05,
+      "loss": 3.4329,
+      "step": 1754624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1266074976922267e-05,
+      "loss": 3.4531,
+      "step": 1755136
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.8544583320617676,
+      "eval_runtime": 309.7694,
+      "eval_samples_per_second": 1231.855,
+      "eval_steps_per_second": 38.496,
+      "step": 1755360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.125770540821548e-05,
+      "loss": 3.4457,
+      "step": 1755648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.124931946070496e-05,
+      "loss": 3.4408,
+      "step": 1756160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.124093351319444e-05,
+      "loss": 3.4514,
+      "step": 1756672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.123254756568392e-05,
+      "loss": 3.4514,
+      "step": 1757184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1224177996977128e-05,
+      "loss": 3.4509,
+      "step": 1757696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1215792049466608e-05,
+      "loss": 3.449,
+      "step": 1758208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.120740610195609e-05,
+      "loss": 3.4368,
+      "step": 1758720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.11990365332493e-05,
+      "loss": 3.4401,
+      "step": 1759232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.119065058573878e-05,
+      "loss": 3.4484,
+      "step": 1759744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.118226463822826e-05,
+      "loss": 3.4539,
+      "step": 1760256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.117387869071774e-05,
+      "loss": 3.4461,
+      "step": 1760768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1165492743207217e-05,
+      "loss": 3.4528,
+      "step": 1761280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1157123174500433e-05,
+      "loss": 3.4454,
+      "step": 1761792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1148737226989913e-05,
+      "loss": 3.4391,
+      "step": 1762304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.114035127947939e-05,
+      "loss": 3.4467,
+      "step": 1762816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.113196533196887e-05,
+      "loss": 3.4327,
+      "step": 1763328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.112357938445835e-05,
+      "loss": 3.4393,
+      "step": 1763840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.111519343694783e-05,
+      "loss": 3.4365,
+      "step": 1764352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.110680748943731e-05,
+      "loss": 3.4406,
+      "step": 1764864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1098421541926793e-05,
+      "loss": 3.4486,
+      "step": 1765376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.109005197322e-05,
+      "loss": 3.452,
+      "step": 1765888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.108166602570948e-05,
+      "loss": 3.4457,
+      "step": 1766400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.107328007819896e-05,
+      "loss": 3.4391,
+      "step": 1766912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.106489413068844e-05,
+      "loss": 3.4482,
+      "step": 1767424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1056524561981654e-05,
+      "loss": 3.4322,
+      "step": 1767936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1048138614471134e-05,
+      "loss": 3.442,
+      "step": 1768448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1039752666960614e-05,
+      "loss": 3.4315,
+      "step": 1768960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1031366719450094e-05,
+      "loss": 3.4421,
+      "step": 1769472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1022997150743303e-05,
+      "loss": 3.4342,
+      "step": 1769984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1014611203232783e-05,
+      "loss": 3.4402,
+      "step": 1770496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1006241634525995e-05,
+      "loss": 3.4433,
+      "step": 1771008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0997855687015475e-05,
+      "loss": 3.4529,
+      "step": 1771520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0989469739504955e-05,
+      "loss": 3.4409,
+      "step": 1772032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0981083791994435e-05,
+      "loss": 3.4368,
+      "step": 1772544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0972697844483915e-05,
+      "loss": 3.4421,
+      "step": 1773056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0964311896973395e-05,
+      "loss": 3.4395,
+      "step": 1773568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0955925949462875e-05,
+      "loss": 3.4455,
+      "step": 1774080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0947540001952355e-05,
+      "loss": 3.4253,
+      "step": 1774592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0939170433245568e-05,
+      "loss": 3.4287,
+      "step": 1775104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0930784485735044e-05,
+      "loss": 3.436,
+      "step": 1775616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0922398538224524e-05,
+      "loss": 3.441,
+      "step": 1776128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0914012590714004e-05,
+      "loss": 3.4379,
+      "step": 1776640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0905643022007217e-05,
+      "loss": 3.4365,
+      "step": 1777152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.089727345330043e-05,
+      "loss": 3.4378,
+      "step": 1777664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.088888750578991e-05,
+      "loss": 3.4462,
+      "step": 1778176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.088050155827939e-05,
+      "loss": 3.4364,
+      "step": 1778688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.087211561076887e-05,
+      "loss": 3.4542,
+      "step": 1779200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0863729663258345e-05,
+      "loss": 3.4254,
+      "step": 1779712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.085534371574783e-05,
+      "loss": 3.4168,
+      "step": 1780224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.084695776823731e-05,
+      "loss": 3.4546,
+      "step": 1780736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.083857182072679e-05,
+      "loss": 3.4403,
+      "step": 1781248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0830202252019998e-05,
+      "loss": 3.4359,
+      "step": 1781760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.082183268331321e-05,
+      "loss": 3.4374,
+      "step": 1782272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.081344673580269e-05,
+      "loss": 3.4332,
+      "step": 1782784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.080506078829217e-05,
+      "loss": 3.4248,
+      "step": 1783296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.079667484078165e-05,
+      "loss": 3.4339,
+      "step": 1783808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.078828889327113e-05,
+      "loss": 3.4375,
+      "step": 1784320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.077990294576061e-05,
+      "loss": 3.4388,
+      "step": 1784832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.077151699825009e-05,
+      "loss": 3.4513,
+      "step": 1785344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.076313105073957e-05,
+      "loss": 3.4322,
+      "step": 1785856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.075476148203278e-05,
+      "loss": 3.4307,
+      "step": 1786368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0746375534522262e-05,
+      "loss": 3.45,
+      "step": 1786880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.073800596581547e-05,
+      "loss": 3.4232,
+      "step": 1787392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.072962001830495e-05,
+      "loss": 3.4216,
+      "step": 1787904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.072123407079443e-05,
+      "loss": 3.4333,
+      "step": 1788416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.071284812328391e-05,
+      "loss": 3.4354,
+      "step": 1788928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.070446217577339e-05,
+      "loss": 3.425,
+      "step": 1789440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.069607622826287e-05,
+      "loss": 3.432,
+      "step": 1789952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.068769028075235e-05,
+      "loss": 3.427,
+      "step": 1790464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.067930433324183e-05,
+      "loss": 3.4431,
+      "step": 1790976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0670934764535044e-05,
+      "loss": 3.4428,
+      "step": 1791488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0662548817024524e-05,
+      "loss": 3.4354,
+      "step": 1792000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0654162869514e-05,
+      "loss": 3.4446,
+      "step": 1792512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0645776922003484e-05,
+      "loss": 3.4427,
+      "step": 1793024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0637423732100425e-05,
+      "loss": 3.4558,
+      "step": 1793536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0629037784589905e-05,
+      "loss": 3.4272,
+      "step": 1794048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0620651837079385e-05,
+      "loss": 3.45,
+      "step": 1794560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0612265889568865e-05,
+      "loss": 3.4372,
+      "step": 1795072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0603879942058345e-05,
+      "loss": 3.4285,
+      "step": 1795584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0595493994547825e-05,
+      "loss": 3.4459,
+      "step": 1796096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0587108047037305e-05,
+      "loss": 3.4409,
+      "step": 1796608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0578722099526785e-05,
+      "loss": 3.4456,
+      "step": 1797120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0570352530819994e-05,
+      "loss": 3.4498,
+      "step": 1797632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0561966583309474e-05,
+      "loss": 3.4248,
+      "step": 1798144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0553580635798954e-05,
+      "loss": 3.4207,
+      "step": 1798656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.054521106709217e-05,
+      "loss": 3.4343,
+      "step": 1799168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0536825119581646e-05,
+      "loss": 3.4459,
+      "step": 1799680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0528439172071126e-05,
+      "loss": 3.4382,
+      "step": 1800192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0520053224560606e-05,
+      "loss": 3.4344,
+      "step": 1800704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0511667277050086e-05,
+      "loss": 3.4286,
+      "step": 1801216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0503281329539566e-05,
+      "loss": 3.4367,
+      "step": 1801728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0494895382029046e-05,
+      "loss": 3.4357,
+      "step": 1802240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0486509434518526e-05,
+      "loss": 3.4312,
+      "step": 1802752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.047813986581174e-05,
+      "loss": 3.4434,
+      "step": 1803264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.046975391830122e-05,
+      "loss": 3.4457,
+      "step": 1803776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0461384349594428e-05,
+      "loss": 3.4438,
+      "step": 1804288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0452998402083908e-05,
+      "loss": 3.4297,
+      "step": 1804800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0444612454573387e-05,
+      "loss": 3.4432,
+      "step": 1805312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.043622650706287e-05,
+      "loss": 3.4309,
+      "step": 1805824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.042784055955235e-05,
+      "loss": 3.4413,
+      "step": 1806336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0419454612041827e-05,
+      "loss": 3.4408,
+      "step": 1806848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0411068664531307e-05,
+      "loss": 3.4354,
+      "step": 1807360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.040269909582452e-05,
+      "loss": 3.4268,
+      "step": 1807872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0394313148314e-05,
+      "loss": 3.433,
+      "step": 1808384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.038592720080348e-05,
+      "loss": 3.4367,
+      "step": 1808896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.037754125329296e-05,
+      "loss": 3.4301,
+      "step": 1809408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0369171684586172e-05,
+      "loss": 3.4167,
+      "step": 1809920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.036078573707565e-05,
+      "loss": 3.4385,
+      "step": 1810432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.035239978956513e-05,
+      "loss": 3.4329,
+      "step": 1810944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.034401384205461e-05,
+      "loss": 3.4362,
+      "step": 1811456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0335644273347824e-05,
+      "loss": 3.4272,
+      "step": 1811968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.03272583258373e-05,
+      "loss": 3.4443,
+      "step": 1812480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.031887237832678e-05,
+      "loss": 3.4339,
+      "step": 1812992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.031048643081626e-05,
+      "loss": 3.4251,
+      "step": 1813504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0302116862109473e-05,
+      "loss": 3.4473,
+      "step": 1814016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0293730914598953e-05,
+      "loss": 3.4426,
+      "step": 1814528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0285344967088433e-05,
+      "loss": 3.4325,
+      "step": 1815040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0276959019577913e-05,
+      "loss": 3.4399,
+      "step": 1815552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0268589450871122e-05,
+      "loss": 3.4258,
+      "step": 1816064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0260203503360602e-05,
+      "loss": 3.4412,
+      "step": 1816576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0251817555850082e-05,
+      "loss": 3.4274,
+      "step": 1817088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0243431608339562e-05,
+      "loss": 3.4412,
+      "step": 1817600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0235062039632775e-05,
+      "loss": 3.4299,
+      "step": 1818112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0226676092122255e-05,
+      "loss": 3.433,
+      "step": 1818624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0218290144611735e-05,
+      "loss": 3.4312,
+      "step": 1819136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0209904197101215e-05,
+      "loss": 3.4333,
+      "step": 1819648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0201518249590695e-05,
+      "loss": 3.428,
+      "step": 1820160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0193132302080175e-05,
+      "loss": 3.4301,
+      "step": 1820672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0184746354569654e-05,
+      "loss": 3.435,
+      "step": 1821184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0176360407059134e-05,
+      "loss": 3.4443,
+      "step": 1821696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0167990838352347e-05,
+      "loss": 3.429,
+      "step": 1822208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0159604890841827e-05,
+      "loss": 3.4499,
+      "step": 1822720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0151218943331303e-05,
+      "loss": 3.4193,
+      "step": 1823232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0142832995820783e-05,
+      "loss": 3.4446,
+      "step": 1823744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0134463427113996e-05,
+      "loss": 3.4373,
+      "step": 1824256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.012607747960348e-05,
+      "loss": 3.4363,
+      "step": 1824768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0117691532092956e-05,
+      "loss": 3.43,
+      "step": 1825280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0109305584582436e-05,
+      "loss": 3.4373,
+      "step": 1825792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0100936015875648e-05,
+      "loss": 3.4381,
+      "step": 1826304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0092550068365128e-05,
+      "loss": 3.4485,
+      "step": 1826816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0084164120854608e-05,
+      "loss": 3.4259,
+      "step": 1827328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0075778173344088e-05,
+      "loss": 3.4363,
+      "step": 1827840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.00674086046373e-05,
+      "loss": 3.4318,
+      "step": 1828352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0059022657126777e-05,
+      "loss": 3.4331,
+      "step": 1828864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0050636709616257e-05,
+      "loss": 3.4317,
+      "step": 1829376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0042250762105737e-05,
+      "loss": 3.4355,
+      "step": 1829888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.003388119339895e-05,
+      "loss": 3.4414,
+      "step": 1830400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.002549524588843e-05,
+      "loss": 3.4204,
+      "step": 1830912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.001710929837791e-05,
+      "loss": 3.4432,
+      "step": 1831424
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.855501413345337,
+      "eval_runtime": 304.4846,
+      "eval_samples_per_second": 1253.236,
+      "eval_steps_per_second": 39.165,
+      "step": 1831680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0008739729671122e-05,
+      "loss": 3.4342,
+      "step": 1831936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000353782160602e-05,
+      "loss": 3.4318,
+      "step": 1832448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999196783465008e-05,
+      "loss": 3.4377,
+      "step": 1832960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9983581887139562e-05,
+      "loss": 3.4382,
+      "step": 1833472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9975212318432774e-05,
+      "loss": 3.4435,
+      "step": 1833984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.996682637092225e-05,
+      "loss": 3.4362,
+      "step": 1834496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.995844042341173e-05,
+      "loss": 3.4282,
+      "step": 1835008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.995005447590121e-05,
+      "loss": 3.4331,
+      "step": 1835520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9941684907194423e-05,
+      "loss": 3.4331,
+      "step": 1836032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9933298959683903e-05,
+      "loss": 3.444,
+      "step": 1836544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9924913012173383e-05,
+      "loss": 3.4372,
+      "step": 1837056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9916527064662863e-05,
+      "loss": 3.44,
+      "step": 1837568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9908173874759805e-05,
+      "loss": 3.4284,
+      "step": 1838080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9899787927249285e-05,
+      "loss": 3.4275,
+      "step": 1838592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9891401979738765e-05,
+      "loss": 3.4377,
+      "step": 1839104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9883016032228248e-05,
+      "loss": 3.4227,
+      "step": 1839616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9874630084717725e-05,
+      "loss": 3.4294,
+      "step": 1840128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9866244137207205e-05,
+      "loss": 3.4251,
+      "step": 1840640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9857858189696685e-05,
+      "loss": 3.4304,
+      "step": 1841152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9849472242186164e-05,
+      "loss": 3.4354,
+      "step": 1841664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9841102673479377e-05,
+      "loss": 3.4436,
+      "step": 1842176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9832716725968857e-05,
+      "loss": 3.4345,
+      "step": 1842688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9824330778458337e-05,
+      "loss": 3.4271,
+      "step": 1843200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9815944830947817e-05,
+      "loss": 3.4331,
+      "step": 1843712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9807575262241026e-05,
+      "loss": 3.4285,
+      "step": 1844224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9799189314730506e-05,
+      "loss": 3.4303,
+      "step": 1844736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9790803367219986e-05,
+      "loss": 3.42,
+      "step": 1845248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.978241741970947e-05,
+      "loss": 3.4328,
+      "step": 1845760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9774047851002678e-05,
+      "loss": 3.431,
+      "step": 1846272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9765661903492158e-05,
+      "loss": 3.4248,
+      "step": 1846784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.975729233478537e-05,
+      "loss": 3.4307,
+      "step": 1847296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9748906387274847e-05,
+      "loss": 3.4443,
+      "step": 1847808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.974052043976433e-05,
+      "loss": 3.4281,
+      "step": 1848320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.973213449225381e-05,
+      "loss": 3.4274,
+      "step": 1848832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.972374854474329e-05,
+      "loss": 3.431,
+      "step": 1849344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.971536259723277e-05,
+      "loss": 3.4351,
+      "step": 1849856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.970697664972225e-05,
+      "loss": 3.4304,
+      "step": 1850368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9698590702211727e-05,
+      "loss": 3.4156,
+      "step": 1850880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.969022113350494e-05,
+      "loss": 3.4187,
+      "step": 1851392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9681851564798152e-05,
+      "loss": 3.4255,
+      "step": 1851904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9673465617287632e-05,
+      "loss": 3.4272,
+      "step": 1852416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9665079669777112e-05,
+      "loss": 3.4267,
+      "step": 1852928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9656693722266592e-05,
+      "loss": 3.4286,
+      "step": 1853440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.96483241535598e-05,
+      "loss": 3.4288,
+      "step": 1853952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.963993820604928e-05,
+      "loss": 3.4332,
+      "step": 1854464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9631552258538764e-05,
+      "loss": 3.4243,
+      "step": 1854976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9623166311028244e-05,
+      "loss": 3.4388,
+      "step": 1855488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9614780363517724e-05,
+      "loss": 3.4184,
+      "step": 1856000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.96063944160072e-05,
+      "loss": 3.4061,
+      "step": 1856512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.959800846849668e-05,
+      "loss": 3.4438,
+      "step": 1857024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.958962252098616e-05,
+      "loss": 3.428,
+      "step": 1857536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9581252952279373e-05,
+      "loss": 3.43,
+      "step": 1858048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9572883383572586e-05,
+      "loss": 3.4243,
+      "step": 1858560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9564497436062066e-05,
+      "loss": 3.4217,
+      "step": 1859072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9556111488551546e-05,
+      "loss": 3.4155,
+      "step": 1859584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9547725541041025e-05,
+      "loss": 3.4211,
+      "step": 1860096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9539339593530502e-05,
+      "loss": 3.4316,
+      "step": 1860608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9530953646019985e-05,
+      "loss": 3.4292,
+      "step": 1861120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9522567698509465e-05,
+      "loss": 3.441,
+      "step": 1861632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9514181750998945e-05,
+      "loss": 3.4197,
+      "step": 1862144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9505812182292154e-05,
+      "loss": 3.4176,
+      "step": 1862656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9497426234781634e-05,
+      "loss": 3.4422,
+      "step": 1863168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9489056666074847e-05,
+      "loss": 3.4116,
+      "step": 1863680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9480670718564327e-05,
+      "loss": 3.4095,
+      "step": 1864192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9472284771053807e-05,
+      "loss": 3.4264,
+      "step": 1864704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9463898823543287e-05,
+      "loss": 3.4237,
+      "step": 1865216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9455512876032767e-05,
+      "loss": 3.4136,
+      "step": 1865728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9447126928522247e-05,
+      "loss": 3.4213,
+      "step": 1866240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9438740981011727e-05,
+      "loss": 3.4164,
+      "step": 1866752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9430355033501207e-05,
+      "loss": 3.4327,
+      "step": 1867264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.942198546479442e-05,
+      "loss": 3.4311,
+      "step": 1867776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94135995172839e-05,
+      "loss": 3.4283,
+      "step": 1868288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.940521356977338e-05,
+      "loss": 3.4319,
+      "step": 1868800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9396827622262855e-05,
+      "loss": 3.4326,
+      "step": 1869312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9388458053556068e-05,
+      "loss": 3.4428,
+      "step": 1869824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9380072106045548e-05,
+      "loss": 3.4182,
+      "step": 1870336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9371686158535028e-05,
+      "loss": 3.4394,
+      "step": 1870848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9363300211024508e-05,
+      "loss": 3.4222,
+      "step": 1871360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.935493064231772e-05,
+      "loss": 3.4178,
+      "step": 1871872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.93465446948072e-05,
+      "loss": 3.4358,
+      "step": 1872384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.933815874729668e-05,
+      "loss": 3.4321,
+      "step": 1872896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9329772799786157e-05,
+      "loss": 3.4341,
+      "step": 1873408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9321403231079373e-05,
+      "loss": 3.439,
+      "step": 1873920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9313017283568853e-05,
+      "loss": 3.4142,
+      "step": 1874432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.930463133605833e-05,
+      "loss": 3.4073,
+      "step": 1874944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.929626176735154e-05,
+      "loss": 3.4248,
+      "step": 1875456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.928787581984102e-05,
+      "loss": 3.4351,
+      "step": 1875968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.92794898723305e-05,
+      "loss": 3.4286,
+      "step": 1876480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.927110392481998e-05,
+      "loss": 3.4262,
+      "step": 1876992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.926271797730946e-05,
+      "loss": 3.4189,
+      "step": 1877504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.925433202979894e-05,
+      "loss": 3.4238,
+      "step": 1878016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.924594608228842e-05,
+      "loss": 3.4278,
+      "step": 1878528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.92375601347779e-05,
+      "loss": 3.4204,
+      "step": 1879040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.922919056607111e-05,
+      "loss": 3.4313,
+      "step": 1879552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9220804618560594e-05,
+      "loss": 3.4341,
+      "step": 1880064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9212435049853803e-05,
+      "loss": 3.4349,
+      "step": 1880576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9204049102343283e-05,
+      "loss": 3.4213,
+      "step": 1881088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9195663154832763e-05,
+      "loss": 3.4281,
+      "step": 1881600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9187277207322243e-05,
+      "loss": 3.4211,
+      "step": 1882112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9178891259811723e-05,
+      "loss": 3.4298,
+      "step": 1882624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9170505312301203e-05,
+      "loss": 3.4322,
+      "step": 1883136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9162119364790686e-05,
+      "loss": 3.4277,
+      "step": 1883648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9153733417280163e-05,
+      "loss": 3.4194,
+      "step": 1884160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9145363848573375e-05,
+      "loss": 3.4173,
+      "step": 1884672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9136977901062855e-05,
+      "loss": 3.4278,
+      "step": 1885184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.912859195355233e-05,
+      "loss": 3.4177,
+      "step": 1885696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9120222384845547e-05,
+      "loss": 3.4051,
+      "step": 1886208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9111836437335027e-05,
+      "loss": 3.4312,
+      "step": 1886720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9103450489824507e-05,
+      "loss": 3.4224,
+      "step": 1887232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9095080921117716e-05,
+      "loss": 3.4262,
+      "step": 1887744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9086694973607196e-05,
+      "loss": 3.4135,
+      "step": 1888256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9078309026096676e-05,
+      "loss": 3.4301,
+      "step": 1888768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9069923078586156e-05,
+      "loss": 3.4261,
+      "step": 1889280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9061537131075636e-05,
+      "loss": 3.4135,
+      "step": 1889792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9053151183565116e-05,
+      "loss": 3.4375,
+      "step": 1890304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9044765236054596e-05,
+      "loss": 3.4334,
+      "step": 1890816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9036379288544076e-05,
+      "loss": 3.4226,
+      "step": 1891328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9028009719837285e-05,
+      "loss": 3.4284,
+      "step": 1891840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.90196401511305e-05,
+      "loss": 3.4135,
+      "step": 1892352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.901125420361998e-05,
+      "loss": 3.4299,
+      "step": 1892864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9002868256109458e-05,
+      "loss": 3.4194,
+      "step": 1893376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8994482308598938e-05,
+      "loss": 3.4275,
+      "step": 1893888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.898611273989215e-05,
+      "loss": 3.4186,
+      "step": 1894400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.897772679238163e-05,
+      "loss": 3.426,
+      "step": 1894912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.896934084487111e-05,
+      "loss": 3.42,
+      "step": 1895424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.896095489736059e-05,
+      "loss": 3.4206,
+      "step": 1895936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.895256894985007e-05,
+      "loss": 3.4159,
+      "step": 1896448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.894418300233955e-05,
+      "loss": 3.4223,
+      "step": 1896960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.893579705482903e-05,
+      "loss": 3.4201,
+      "step": 1897472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.892741110731851e-05,
+      "loss": 3.4299,
+      "step": 1897984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.891904153861172e-05,
+      "loss": 3.4212,
+      "step": 1898496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8910655591101202e-05,
+      "loss": 3.4413,
+      "step": 1899008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.890228602239441e-05,
+      "loss": 3.409,
+      "step": 1899520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.889390007488389e-05,
+      "loss": 3.4305,
+      "step": 1900032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.888551412737337e-05,
+      "loss": 3.4294,
+      "step": 1900544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.887712817986285e-05,
+      "loss": 3.427,
+      "step": 1901056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.886874223235233e-05,
+      "loss": 3.4191,
+      "step": 1901568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.886035628484181e-05,
+      "loss": 3.4247,
+      "step": 1902080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.885197033733129e-05,
+      "loss": 3.4246,
+      "step": 1902592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.884358438982077e-05,
+      "loss": 3.4423,
+      "step": 1903104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8835214821113983e-05,
+      "loss": 3.4156,
+      "step": 1903616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.882682887360346e-05,
+      "loss": 3.4234,
+      "step": 1904128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.881844292609294e-05,
+      "loss": 3.4194,
+      "step": 1904640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.881005697858242e-05,
+      "loss": 3.4244,
+      "step": 1905152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8801687409875636e-05,
+      "loss": 3.4176,
+      "step": 1905664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8793301462365112e-05,
+      "loss": 3.428,
+      "step": 1906176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8784931893658325e-05,
+      "loss": 3.4266,
+      "step": 1906688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8776545946147805e-05,
+      "loss": 3.419,
+      "step": 1907200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8768159998637285e-05,
+      "loss": 3.4275,
+      "step": 1907712
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.857713460922241,
+      "eval_runtime": 631.031,
+      "eval_samples_per_second": 604.71,
+      "eval_steps_per_second": 18.898,
+      "step": 1908000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8759774051126765e-05,
+      "loss": 3.4171,
+      "step": 1908224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8751388103616245e-05,
+      "loss": 3.4187,
+      "step": 1908736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8743002156105725e-05,
+      "loss": 3.4305,
+      "step": 1909248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8734616208595205e-05,
+      "loss": 3.4307,
+      "step": 1909760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8726230261084685e-05,
+      "loss": 3.4335,
+      "step": 1910272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8717860692377894e-05,
+      "loss": 3.4244,
+      "step": 1910784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8709474744867374e-05,
+      "loss": 3.4196,
+      "step": 1911296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8701088797356857e-05,
+      "loss": 3.4242,
+      "step": 1911808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8692702849846337e-05,
+      "loss": 3.42,
+      "step": 1912320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8684333281139546e-05,
+      "loss": 3.4361,
+      "step": 1912832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8675947333629026e-05,
+      "loss": 3.4282,
+      "step": 1913344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8667561386118506e-05,
+      "loss": 3.4267,
+      "step": 1913856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.865919181741172e-05,
+      "loss": 3.4298,
+      "step": 1914368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.86508058699012e-05,
+      "loss": 3.4133,
+      "step": 1914880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.864241992239068e-05,
+      "loss": 3.4275,
+      "step": 1915392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8634033974880158e-05,
+      "loss": 3.4132,
+      "step": 1915904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8625648027369638e-05,
+      "loss": 3.4157,
+      "step": 1916416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8617262079859115e-05,
+      "loss": 3.4179,
+      "step": 1916928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8608876132348595e-05,
+      "loss": 3.4156,
+      "step": 1917440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8600490184838075e-05,
+      "loss": 3.4262,
+      "step": 1917952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.859212061613129e-05,
+      "loss": 3.4365,
+      "step": 1918464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8583734668620767e-05,
+      "loss": 3.4294,
+      "step": 1918976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8575348721110247e-05,
+      "loss": 3.418,
+      "step": 1919488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8566962773599727e-05,
+      "loss": 3.4191,
+      "step": 1920000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8558593204892936e-05,
+      "loss": 3.4157,
+      "step": 1920512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8550223636186152e-05,
+      "loss": 3.4212,
+      "step": 1921024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8541837688675632e-05,
+      "loss": 3.4109,
+      "step": 1921536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8533451741165112e-05,
+      "loss": 3.4271,
+      "step": 1922048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852506579365459e-05,
+      "loss": 3.4152,
+      "step": 1922560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851667984614407e-05,
+      "loss": 3.4169,
+      "step": 1923072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850831027743728e-05,
+      "loss": 3.417,
+      "step": 1923584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8499924329926764e-05,
+      "loss": 3.4328,
+      "step": 1924096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849153838241624e-05,
+      "loss": 3.4202,
+      "step": 1924608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848315243490572e-05,
+      "loss": 3.4168,
+      "step": 1925120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.84747664873952e-05,
+      "loss": 3.4191,
+      "step": 1925632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.846638053988468e-05,
+      "loss": 3.4206,
+      "step": 1926144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.845799459237416e-05,
+      "loss": 3.4242,
+      "step": 1926656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.844960864486364e-05,
+      "loss": 3.4071,
+      "step": 1927168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8441239076156853e-05,
+      "loss": 3.4074,
+      "step": 1927680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8432853128646333e-05,
+      "loss": 3.4133,
+      "step": 1928192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8424467181135813e-05,
+      "loss": 3.4139,
+      "step": 1928704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8416081233625293e-05,
+      "loss": 3.4152,
+      "step": 1929216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.840769528611477e-05,
+      "loss": 3.4189,
+      "step": 1929728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8399325717407982e-05,
+      "loss": 3.4226,
+      "step": 1930240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8390939769897465e-05,
+      "loss": 3.4174,
+      "step": 1930752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8382553822386945e-05,
+      "loss": 3.4203,
+      "step": 1931264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8374167874876422e-05,
+      "loss": 3.4276,
+      "step": 1931776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8365781927365902e-05,
+      "loss": 3.4123,
+      "step": 1932288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8357395979855382e-05,
+      "loss": 3.3941,
+      "step": 1932800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8349010032344862e-05,
+      "loss": 3.4293,
+      "step": 1933312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8340640463638074e-05,
+      "loss": 3.4166,
+      "step": 1933824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8332254516127554e-05,
+      "loss": 3.4227,
+      "step": 1934336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8323868568617034e-05,
+      "loss": 3.4172,
+      "step": 1934848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8315482621106514e-05,
+      "loss": 3.4095,
+      "step": 1935360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8307096673595994e-05,
+      "loss": 3.4038,
+      "step": 1935872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8298710726085474e-05,
+      "loss": 3.4112,
+      "step": 1936384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.829032477857495e-05,
+      "loss": 3.4232,
+      "step": 1936896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8281938831064434e-05,
+      "loss": 3.414,
+      "step": 1937408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8273569262357646e-05,
+      "loss": 3.4291,
+      "step": 1937920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8265199693650856e-05,
+      "loss": 3.4147,
+      "step": 1938432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8256813746140336e-05,
+      "loss": 3.405,
+      "step": 1938944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8248427798629815e-05,
+      "loss": 3.4316,
+      "step": 1939456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8240041851119295e-05,
+      "loss": 3.4039,
+      "step": 1939968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8231655903608775e-05,
+      "loss": 3.3971,
+      "step": 1940480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8223269956098255e-05,
+      "loss": 3.4143,
+      "step": 1940992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8214884008587735e-05,
+      "loss": 3.4129,
+      "step": 1941504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8206498061077215e-05,
+      "loss": 3.4062,
+      "step": 1942016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8198128492370424e-05,
+      "loss": 3.4094,
+      "step": 1942528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8189742544859904e-05,
+      "loss": 3.4083,
+      "step": 1943040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8181356597349388e-05,
+      "loss": 3.421,
+      "step": 1943552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8172970649838868e-05,
+      "loss": 3.4235,
+      "step": 1944064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8164601081132077e-05,
+      "loss": 3.4163,
+      "step": 1944576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8156215133621557e-05,
+      "loss": 3.4241,
+      "step": 1945088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8147829186111037e-05,
+      "loss": 3.4204,
+      "step": 1945600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8139443238600517e-05,
+      "loss": 3.4311,
+      "step": 1946112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.813107366989373e-05,
+      "loss": 3.4124,
+      "step": 1946624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.812268772238321e-05,
+      "loss": 3.431,
+      "step": 1947136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.811430177487269e-05,
+      "loss": 3.4115,
+      "step": 1947648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.810591582736217e-05,
+      "loss": 3.4104,
+      "step": 1948160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8097546258655378e-05,
+      "loss": 3.42,
+      "step": 1948672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8089160311144858e-05,
+      "loss": 3.4217,
+      "step": 1949184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.808077436363434e-05,
+      "loss": 3.4251,
+      "step": 1949696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.807238841612382e-05,
+      "loss": 3.427,
+      "step": 1950208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.806401884741703e-05,
+      "loss": 3.4005,
+      "step": 1950720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.805563289990651e-05,
+      "loss": 3.4019,
+      "step": 1951232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.804726333119972e-05,
+      "loss": 3.4154,
+      "step": 1951744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8038877383689203e-05,
+      "loss": 3.4224,
+      "step": 1952256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8030491436178683e-05,
+      "loss": 3.416,
+      "step": 1952768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8022105488668163e-05,
+      "loss": 3.415,
+      "step": 1953280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8013719541157643e-05,
+      "loss": 3.4118,
+      "step": 1953792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8005333593647123e-05,
+      "loss": 3.4166,
+      "step": 1954304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7996947646136603e-05,
+      "loss": 3.414,
+      "step": 1954816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.798856169862608e-05,
+      "loss": 3.4115,
+      "step": 1955328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.798019212991929e-05,
+      "loss": 3.421,
+      "step": 1955840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7971806182408775e-05,
+      "loss": 3.4214,
+      "step": 1956352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7963436613701984e-05,
+      "loss": 3.426,
+      "step": 1956864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7955050666191464e-05,
+      "loss": 3.4113,
+      "step": 1957376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7946664718680944e-05,
+      "loss": 3.4152,
+      "step": 1957888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7938278771170424e-05,
+      "loss": 3.4135,
+      "step": 1958400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7929892823659904e-05,
+      "loss": 3.4174,
+      "step": 1958912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7921506876149384e-05,
+      "loss": 3.4204,
+      "step": 1959424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7913120928638864e-05,
+      "loss": 3.4188,
+      "step": 1959936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7904734981128344e-05,
+      "loss": 3.4114,
+      "step": 1960448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7896365412421553e-05,
+      "loss": 3.4045,
+      "step": 1960960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7887979464911033e-05,
+      "loss": 3.4204,
+      "step": 1961472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7879593517400513e-05,
+      "loss": 3.4086,
+      "step": 1961984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7871207569889996e-05,
+      "loss": 3.3938,
+      "step": 1962496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7862838001183205e-05,
+      "loss": 3.4165,
+      "step": 1963008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7854452053672685e-05,
+      "loss": 3.421,
+      "step": 1963520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7846082484965898e-05,
+      "loss": 3.4121,
+      "step": 1964032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7837696537455374e-05,
+      "loss": 3.4076,
+      "step": 1964544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7829310589944858e-05,
+      "loss": 3.415,
+      "step": 1965056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7820924642434337e-05,
+      "loss": 3.4156,
+      "step": 1965568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7812538694923817e-05,
+      "loss": 3.404,
+      "step": 1966080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7804152747413297e-05,
+      "loss": 3.4286,
+      "step": 1966592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7795766799902777e-05,
+      "loss": 3.4258,
+      "step": 1967104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7787380852392257e-05,
+      "loss": 3.4106,
+      "step": 1967616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7779011283685466e-05,
+      "loss": 3.4178,
+      "step": 1968128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.777064171497868e-05,
+      "loss": 3.4065,
+      "step": 1968640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.776225576746816e-05,
+      "loss": 3.418,
+      "step": 1969152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.775386981995764e-05,
+      "loss": 3.4069,
+      "step": 1969664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.774548387244712e-05,
+      "loss": 3.4157,
+      "step": 1970176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7737114303740328e-05,
+      "loss": 3.4096,
+      "step": 1970688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.772872835622981e-05,
+      "loss": 3.4182,
+      "step": 1971200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.772034240871929e-05,
+      "loss": 3.4122,
+      "step": 1971712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.771195646120877e-05,
+      "loss": 3.4118,
+      "step": 1972224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.770357051369825e-05,
+      "loss": 3.4047,
+      "step": 1972736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.769518456618773e-05,
+      "loss": 3.4114,
+      "step": 1973248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7686798618677208e-05,
+      "loss": 3.4074,
+      "step": 1973760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7678412671166688e-05,
+      "loss": 3.4205,
+      "step": 1974272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.76700431024599e-05,
+      "loss": 3.4135,
+      "step": 1974784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7661673533753113e-05,
+      "loss": 3.4276,
+      "step": 1975296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7653287586242592e-05,
+      "loss": 3.397,
+      "step": 1975808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7644901638732072e-05,
+      "loss": 3.4251,
+      "step": 1976320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7636515691221552e-05,
+      "loss": 3.4151,
+      "step": 1976832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.762812974371103e-05,
+      "loss": 3.4148,
+      "step": 1977344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7619743796200512e-05,
+      "loss": 3.4104,
+      "step": 1977856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7611357848689992e-05,
+      "loss": 3.4111,
+      "step": 1978368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7602971901179472e-05,
+      "loss": 3.4176,
+      "step": 1978880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.759460233247268e-05,
+      "loss": 3.4344,
+      "step": 1979392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.758621638496216e-05,
+      "loss": 3.4019,
+      "step": 1979904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.757783043745164e-05,
+      "loss": 3.416,
+      "step": 1980416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.756944448994112e-05,
+      "loss": 3.4101,
+      "step": 1980928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7561074921234334e-05,
+      "loss": 3.413,
+      "step": 1981440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7552688973723814e-05,
+      "loss": 3.4066,
+      "step": 1981952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7544303026213294e-05,
+      "loss": 3.42,
+      "step": 1982464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7535933457506503e-05,
+      "loss": 3.4144,
+      "step": 1982976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7527547509995983e-05,
+      "loss": 3.4145,
+      "step": 1983488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7519161562485466e-05,
+      "loss": 3.4146,
+      "step": 1984000
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.859578847885132,
+      "eval_runtime": 637.0796,
+      "eval_samples_per_second": 598.969,
+      "eval_steps_per_second": 18.718,
+      "step": 1984320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.751079199377868e-05,
+      "loss": 3.4086,
+      "step": 1984512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7502406046268155e-05,
+      "loss": 3.4092,
+      "step": 1985024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7494020098757635e-05,
+      "loss": 3.418,
+      "step": 1985536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7485634151247115e-05,
+      "loss": 3.4221,
+      "step": 1986048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7477264582540327e-05,
+      "loss": 3.4233,
+      "step": 1986560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7468878635029807e-05,
+      "loss": 3.4143,
+      "step": 1987072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.746050906632302e-05,
+      "loss": 3.4096,
+      "step": 1987584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.74521231188125e-05,
+      "loss": 3.4136,
+      "step": 1988096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7443737171301976e-05,
+      "loss": 3.4102,
+      "step": 1988608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7435351223791456e-05,
+      "loss": 3.4247,
+      "step": 1989120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7426965276280936e-05,
+      "loss": 3.4173,
+      "step": 1989632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.741857932877042e-05,
+      "loss": 3.4186,
+      "step": 1990144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.741020976006363e-05,
+      "loss": 3.4216,
+      "step": 1990656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.740182381255311e-05,
+      "loss": 3.398,
+      "step": 1991168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.739343786504259e-05,
+      "loss": 3.4209,
+      "step": 1991680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.738505191753207e-05,
+      "loss": 3.4004,
+      "step": 1992192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.737666597002155e-05,
+      "loss": 3.4065,
+      "step": 1992704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.736828002251103e-05,
+      "loss": 3.4053,
+      "step": 1993216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.735989407500051e-05,
+      "loss": 3.4059,
+      "step": 1993728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.735150812748999e-05,
+      "loss": 3.4146,
+      "step": 1994240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.73431385587832e-05,
+      "loss": 3.422,
+      "step": 1994752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.733475261127268e-05,
+      "loss": 3.4204,
+      "step": 1995264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7326366663762157e-05,
+      "loss": 3.4084,
+      "step": 1995776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7317980716251637e-05,
+      "loss": 3.4059,
+      "step": 1996288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7309611147544853e-05,
+      "loss": 3.4106,
+      "step": 1996800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.730122520003433e-05,
+      "loss": 3.4095,
+      "step": 1997312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.729283925252381e-05,
+      "loss": 3.4009,
+      "step": 1997824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.728445330501329e-05,
+      "loss": 3.4186,
+      "step": 1998336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7276083736306502e-05,
+      "loss": 3.4026,
+      "step": 1998848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7267697788795982e-05,
+      "loss": 3.4109,
+      "step": 1999360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7259328220089195e-05,
+      "loss": 3.4041,
+      "step": 1999872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7250942272578675e-05,
+      "loss": 3.4198,
+      "step": 2000384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7242556325068155e-05,
+      "loss": 3.415,
+      "step": 2000896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.723417037755763e-05,
+      "loss": 3.4033,
+      "step": 2001408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.722578443004711e-05,
+      "loss": 3.4128,
+      "step": 2001920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.721739848253659e-05,
+      "loss": 3.4038,
+      "step": 2002432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7209012535026074e-05,
+      "loss": 3.4159,
+      "step": 2002944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7200626587515554e-05,
+      "loss": 3.4032,
+      "step": 2003456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7192257018808763e-05,
+      "loss": 3.3932,
+      "step": 2003968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7183871071298243e-05,
+      "loss": 3.4017,
+      "step": 2004480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7175485123787723e-05,
+      "loss": 3.4103,
+      "step": 2004992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7167099176277203e-05,
+      "loss": 3.402,
+      "step": 2005504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7158729607570416e-05,
+      "loss": 3.4071,
+      "step": 2006016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7150343660059896e-05,
+      "loss": 3.4122,
+      "step": 2006528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7141957712549376e-05,
+      "loss": 3.4067,
+      "step": 2007040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7133571765038856e-05,
+      "loss": 3.411,
+      "step": 2007552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7125185817528336e-05,
+      "loss": 3.421,
+      "step": 2008064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7116799870017812e-05,
+      "loss": 3.4004,
+      "step": 2008576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7108413922507292e-05,
+      "loss": 3.3837,
+      "step": 2009088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7100027974996775e-05,
+      "loss": 3.4177,
+      "step": 2009600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7091658406289985e-05,
+      "loss": 3.4044,
+      "step": 2010112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7083272458779465e-05,
+      "loss": 3.4112,
+      "step": 2010624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7074902890072677e-05,
+      "loss": 3.4053,
+      "step": 2011136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7066516942562157e-05,
+      "loss": 3.4013,
+      "step": 2011648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7058130995051637e-05,
+      "loss": 3.395,
+      "step": 2012160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7049745047541117e-05,
+      "loss": 3.3973,
+      "step": 2012672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7041359100030597e-05,
+      "loss": 3.4173,
+      "step": 2013184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.703298953132381e-05,
+      "loss": 3.402,
+      "step": 2013696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7024603583813286e-05,
+      "loss": 3.4207,
+      "step": 2014208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7016217636302766e-05,
+      "loss": 3.4064,
+      "step": 2014720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7007831688792246e-05,
+      "loss": 3.3928,
+      "step": 2015232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.699944574128173e-05,
+      "loss": 3.417,
+      "step": 2015744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.699105979377121e-05,
+      "loss": 3.4016,
+      "step": 2016256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6982690225064418e-05,
+      "loss": 3.385,
+      "step": 2016768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6974304277553898e-05,
+      "loss": 3.4051,
+      "step": 2017280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6965918330043378e-05,
+      "loss": 3.4031,
+      "step": 2017792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6957532382532858e-05,
+      "loss": 3.3942,
+      "step": 2018304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6949146435022338e-05,
+      "loss": 3.3986,
+      "step": 2018816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6940760487511818e-05,
+      "loss": 3.3996,
+      "step": 2019328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6932374540001298e-05,
+      "loss": 3.4095,
+      "step": 2019840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6923988592490778e-05,
+      "loss": 3.4135,
+      "step": 2020352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.691561902378399e-05,
+      "loss": 3.4027,
+      "step": 2020864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6907233076273467e-05,
+      "loss": 3.416,
+      "step": 2021376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6898847128762947e-05,
+      "loss": 3.4105,
+      "step": 2021888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6890477560056163e-05,
+      "loss": 3.4216,
+      "step": 2022400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6882107991349372e-05,
+      "loss": 3.401,
+      "step": 2022912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6873722043838852e-05,
+      "loss": 3.422,
+      "step": 2023424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6865336096328332e-05,
+      "loss": 3.4005,
+      "step": 2023936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6856950148817812e-05,
+      "loss": 3.3991,
+      "step": 2024448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.684856420130729e-05,
+      "loss": 3.4134,
+      "step": 2024960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.684017825379677e-05,
+      "loss": 3.4066,
+      "step": 2025472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.683179230628625e-05,
+      "loss": 3.4173,
+      "step": 2025984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.682342273757946e-05,
+      "loss": 3.4165,
+      "step": 2026496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.681503679006894e-05,
+      "loss": 3.3941,
+      "step": 2027008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.680665084255842e-05,
+      "loss": 3.3908,
+      "step": 2027520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6798281273851636e-05,
+      "loss": 3.402,
+      "step": 2028032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6789895326341113e-05,
+      "loss": 3.4168,
+      "step": 2028544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6781509378830593e-05,
+      "loss": 3.4041,
+      "step": 2029056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6773123431320073e-05,
+      "loss": 3.4059,
+      "step": 2029568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6764737483809553e-05,
+      "loss": 3.4057,
+      "step": 2030080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6756351536299033e-05,
+      "loss": 3.399,
+      "step": 2030592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6747965588788513e-05,
+      "loss": 3.4094,
+      "step": 2031104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6739579641277993e-05,
+      "loss": 3.3998,
+      "step": 2031616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6731210072571205e-05,
+      "loss": 3.4103,
+      "step": 2032128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6722824125060685e-05,
+      "loss": 3.4097,
+      "step": 2032640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6714454556353894e-05,
+      "loss": 3.418,
+      "step": 2033152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6706068608843374e-05,
+      "loss": 3.3988,
+      "step": 2033664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6697682661332854e-05,
+      "loss": 3.409,
+      "step": 2034176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6689296713822338e-05,
+      "loss": 3.3989,
+      "step": 2034688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6680910766311818e-05,
+      "loss": 3.408,
+      "step": 2035200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6672524818801294e-05,
+      "loss": 3.4113,
+      "step": 2035712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6664138871290774e-05,
+      "loss": 3.4053,
+      "step": 2036224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6655752923780254e-05,
+      "loss": 3.4032,
+      "step": 2036736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6647383355073466e-05,
+      "loss": 3.3963,
+      "step": 2037248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.663901378636668e-05,
+      "loss": 3.4057,
+      "step": 2037760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.663062783885616e-05,
+      "loss": 3.4004,
+      "step": 2038272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.662224189134564e-05,
+      "loss": 3.384,
+      "step": 2038784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6613855943835115e-05,
+      "loss": 3.4083,
+      "step": 2039296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6605469996324595e-05,
+      "loss": 3.4071,
+      "step": 2039808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6597100427617808e-05,
+      "loss": 3.4012,
+      "step": 2040320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.658871448010729e-05,
+      "loss": 3.3975,
+      "step": 2040832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6580328532596768e-05,
+      "loss": 3.4029,
+      "step": 2041344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6571942585086248e-05,
+      "loss": 3.4067,
+      "step": 2041856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6563556637575728e-05,
+      "loss": 3.3913,
+      "step": 2042368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6555170690065208e-05,
+      "loss": 3.42,
+      "step": 2042880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6546784742554688e-05,
+      "loss": 3.4134,
+      "step": 2043392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6538398795044168e-05,
+      "loss": 3.4024,
+      "step": 2043904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.653002922633738e-05,
+      "loss": 3.4091,
+      "step": 2044416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.652165965763059e-05,
+      "loss": 3.3968,
+      "step": 2044928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.651327371012007e-05,
+      "loss": 3.4086,
+      "step": 2045440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.650488776260955e-05,
+      "loss": 3.4013,
+      "step": 2045952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.649650181509903e-05,
+      "loss": 3.4015,
+      "step": 2046464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.648813224639224e-05,
+      "loss": 3.4034,
+      "step": 2046976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.647974629888172e-05,
+      "loss": 3.4034,
+      "step": 2047488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.64713603513712e-05,
+      "loss": 3.4054,
+      "step": 2048000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.646297440386068e-05,
+      "loss": 3.3983,
+      "step": 2048512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.645458845635016e-05,
+      "loss": 3.4001,
+      "step": 2049024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.644620250883964e-05,
+      "loss": 3.3981,
+      "step": 2049536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.643781656132912e-05,
+      "loss": 3.3992,
+      "step": 2050048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.64294306138186e-05,
+      "loss": 3.4115,
+      "step": 2050560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6421061045111814e-05,
+      "loss": 3.4047,
+      "step": 2051072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6412675097601294e-05,
+      "loss": 3.4168,
+      "step": 2051584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.640428915009077e-05,
+      "loss": 3.3853,
+      "step": 2052096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.639590320258025e-05,
+      "loss": 3.4164,
+      "step": 2052608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6387533633873463e-05,
+      "loss": 3.4053,
+      "step": 2053120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6379147686362946e-05,
+      "loss": 3.4063,
+      "step": 2053632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6370761738852423e-05,
+      "loss": 3.4033,
+      "step": 2054144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6362392170145635e-05,
+      "loss": 3.3979,
+      "step": 2054656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6354006222635115e-05,
+      "loss": 3.4068,
+      "step": 2055168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6345620275124595e-05,
+      "loss": 3.4223,
+      "step": 2055680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6337234327614075e-05,
+      "loss": 3.3953,
+      "step": 2056192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6328848380103555e-05,
+      "loss": 3.4069,
+      "step": 2056704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6320462432593035e-05,
+      "loss": 3.3982,
+      "step": 2057216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6312076485082515e-05,
+      "loss": 3.4011,
+      "step": 2057728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6303690537571995e-05,
+      "loss": 3.3993,
+      "step": 2058240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6295320968865204e-05,
+      "loss": 3.4073,
+      "step": 2058752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6286935021354684e-05,
+      "loss": 3.4053,
+      "step": 2059264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6278549073844164e-05,
+      "loss": 3.4069,
+      "step": 2059776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6270163126333647e-05,
+      "loss": 3.4027,
+      "step": 2060288
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.85927677154541,
+      "eval_runtime": 302.9986,
+      "eval_samples_per_second": 1259.382,
+      "eval_steps_per_second": 39.357,
+      "step": 2060640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.626180993643059e-05,
+      "loss": 3.3969,
+      "step": 2060800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.625342398892007e-05,
+      "loss": 3.3991,
+      "step": 2061312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6245038041409545e-05,
+      "loss": 3.4096,
+      "step": 2061824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6236652093899025e-05,
+      "loss": 3.412,
+      "step": 2062336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.622826614638851e-05,
+      "loss": 3.4114,
+      "step": 2062848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.621988019887799e-05,
+      "loss": 3.4034,
+      "step": 2063360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.621149425136747e-05,
+      "loss": 3.4028,
+      "step": 2063872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.620310830385695e-05,
+      "loss": 3.4055,
+      "step": 2064384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6194738735150157e-05,
+      "loss": 3.3982,
+      "step": 2064896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6186352787639637e-05,
+      "loss": 3.4101,
+      "step": 2065408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6177966840129117e-05,
+      "loss": 3.408,
+      "step": 2065920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.61695808926186e-05,
+      "loss": 3.4118,
+      "step": 2066432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.616122770271554e-05,
+      "loss": 3.4069,
+      "step": 2066944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.615284175520502e-05,
+      "loss": 3.3915,
+      "step": 2067456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.61444558076945e-05,
+      "loss": 3.4101,
+      "step": 2067968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.613606986018398e-05,
+      "loss": 3.389,
+      "step": 2068480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6127683912673462e-05,
+      "loss": 3.4,
+      "step": 2068992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6119297965162942e-05,
+      "loss": 3.3982,
+      "step": 2069504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6110912017652422e-05,
+      "loss": 3.3929,
+      "step": 2070016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.61025260701419e-05,
+      "loss": 3.4007,
+      "step": 2070528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.609415650143511e-05,
+      "loss": 3.4165,
+      "step": 2071040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.608577055392459e-05,
+      "loss": 3.4062,
+      "step": 2071552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.607738460641407e-05,
+      "loss": 3.4047,
+      "step": 2072064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.606899865890355e-05,
+      "loss": 3.398,
+      "step": 2072576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6060629090196764e-05,
+      "loss": 3.3984,
+      "step": 2073088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6052243142686243e-05,
+      "loss": 3.3973,
+      "step": 2073600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.604385719517572e-05,
+      "loss": 3.3895,
+      "step": 2074112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.60354712476652e-05,
+      "loss": 3.4066,
+      "step": 2074624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6027101678958416e-05,
+      "loss": 3.3933,
+      "step": 2075136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6018715731447896e-05,
+      "loss": 3.4009,
+      "step": 2075648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6010346162741105e-05,
+      "loss": 3.3948,
+      "step": 2076160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6001960215230585e-05,
+      "loss": 3.41,
+      "step": 2076672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5993574267720065e-05,
+      "loss": 3.3991,
+      "step": 2077184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5985188320209545e-05,
+      "loss": 3.3995,
+      "step": 2077696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5976802372699025e-05,
+      "loss": 3.4016,
+      "step": 2078208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5968416425188505e-05,
+      "loss": 3.3967,
+      "step": 2078720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5960030477677985e-05,
+      "loss": 3.409,
+      "step": 2079232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5951644530167465e-05,
+      "loss": 3.3874,
+      "step": 2079744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5943274961460674e-05,
+      "loss": 3.3866,
+      "step": 2080256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5934889013950154e-05,
+      "loss": 3.3929,
+      "step": 2080768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5926503066439634e-05,
+      "loss": 3.4004,
+      "step": 2081280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5918117118929117e-05,
+      "loss": 3.3918,
+      "step": 2081792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5909747550222326e-05,
+      "loss": 3.3998,
+      "step": 2082304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5901361602711806e-05,
+      "loss": 3.4007,
+      "step": 2082816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5892975655201286e-05,
+      "loss": 3.3966,
+      "step": 2083328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5884589707690766e-05,
+      "loss": 3.3999,
+      "step": 2083840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5876203760180246e-05,
+      "loss": 3.406,
+      "step": 2084352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.586783419147346e-05,
+      "loss": 3.3947,
+      "step": 2084864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.585944824396294e-05,
+      "loss": 3.3718,
+      "step": 2085376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5851062296452418e-05,
+      "loss": 3.4073,
+      "step": 2085888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5842676348941898e-05,
+      "loss": 3.3971,
+      "step": 2086400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5834290401431375e-05,
+      "loss": 3.3975,
+      "step": 2086912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5825920832724587e-05,
+      "loss": 3.395,
+      "step": 2087424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.581753488521407e-05,
+      "loss": 3.3917,
+      "step": 2087936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.580914893770355e-05,
+      "loss": 3.3858,
+      "step": 2088448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5800762990193027e-05,
+      "loss": 3.3883,
+      "step": 2088960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5792377042682507e-05,
+      "loss": 3.4089,
+      "step": 2089472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5783991095171987e-05,
+      "loss": 3.3916,
+      "step": 2089984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5775605147661467e-05,
+      "loss": 3.4121,
+      "step": 2090496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.576723557895468e-05,
+      "loss": 3.3971,
+      "step": 2091008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.575884963144416e-05,
+      "loss": 3.3806,
+      "step": 2091520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.575046368393364e-05,
+      "loss": 3.4101,
+      "step": 2092032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.574207773642312e-05,
+      "loss": 3.3898,
+      "step": 2092544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.57336917889126e-05,
+      "loss": 3.381,
+      "step": 2093056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.572530584140208e-05,
+      "loss": 3.3931,
+      "step": 2093568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5716919893891556e-05,
+      "loss": 3.3912,
+      "step": 2094080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.570853394638104e-05,
+      "loss": 3.3841,
+      "step": 2094592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.570016437767425e-05,
+      "loss": 3.3931,
+      "step": 2095104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.569177843016373e-05,
+      "loss": 3.3904,
+      "step": 2095616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5683392482653208e-05,
+      "loss": 3.3965,
+      "step": 2096128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5675006535142688e-05,
+      "loss": 3.403,
+      "step": 2096640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.56666369664359e-05,
+      "loss": 3.3962,
+      "step": 2097152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.565825101892538e-05,
+      "loss": 3.4031,
+      "step": 2097664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.564986507141486e-05,
+      "loss": 3.3985,
+      "step": 2098176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.564147912390434e-05,
+      "loss": 3.4125,
+      "step": 2098688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5633109555197553e-05,
+      "loss": 3.3927,
+      "step": 2099200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.562472360768703e-05,
+      "loss": 3.412,
+      "step": 2099712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.561633766017651e-05,
+      "loss": 3.391,
+      "step": 2100224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5607951712665993e-05,
+      "loss": 3.3891,
+      "step": 2100736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5599582143959205e-05,
+      "loss": 3.4009,
+      "step": 2101248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5591212575252414e-05,
+      "loss": 3.3992,
+      "step": 2101760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5582826627741894e-05,
+      "loss": 3.4053,
+      "step": 2102272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5574440680231374e-05,
+      "loss": 3.4107,
+      "step": 2102784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5566054732720854e-05,
+      "loss": 3.3887,
+      "step": 2103296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5557668785210334e-05,
+      "loss": 3.3803,
+      "step": 2103808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5549299216503547e-05,
+      "loss": 3.3898,
+      "step": 2104320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5540913268993027e-05,
+      "loss": 3.408,
+      "step": 2104832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5532527321482503e-05,
+      "loss": 3.3913,
+      "step": 2105344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5524141373971983e-05,
+      "loss": 3.3977,
+      "step": 2105856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5515755426461463e-05,
+      "loss": 3.3975,
+      "step": 2106368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5507369478950947e-05,
+      "loss": 3.3879,
+      "step": 2106880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5498983531440426e-05,
+      "loss": 3.3986,
+      "step": 2107392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5490597583929906e-05,
+      "loss": 3.3923,
+      "step": 2107904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5482228015223116e-05,
+      "loss": 3.3954,
+      "step": 2108416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5473842067712595e-05,
+      "loss": 3.4052,
+      "step": 2108928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5465472499005808e-05,
+      "loss": 3.4054,
+      "step": 2109440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5457086551495288e-05,
+      "loss": 3.3913,
+      "step": 2109952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5448700603984768e-05,
+      "loss": 3.3987,
+      "step": 2110464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5440314656474248e-05,
+      "loss": 3.3903,
+      "step": 2110976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5431928708963728e-05,
+      "loss": 3.3981,
+      "step": 2111488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5423542761453208e-05,
+      "loss": 3.4001,
+      "step": 2112000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5415156813942684e-05,
+      "loss": 3.396,
+      "step": 2112512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5406770866432164e-05,
+      "loss": 3.3931,
+      "step": 2113024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539840129772538e-05,
+      "loss": 3.386,
+      "step": 2113536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539003172901859e-05,
+      "loss": 3.3984,
+      "step": 2114048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538164578150807e-05,
+      "loss": 3.3894,
+      "step": 2114560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537325983399755e-05,
+      "loss": 3.3759,
+      "step": 2115072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536487388648703e-05,
+      "loss": 3.3977,
+      "step": 2115584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535648793897651e-05,
+      "loss": 3.3972,
+      "step": 2116096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534811837026972e-05,
+      "loss": 3.3911,
+      "step": 2116608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53397324227592e-05,
+      "loss": 3.3887,
+      "step": 2117120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533134647524868e-05,
+      "loss": 3.3961,
+      "step": 2117632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5322960527738158e-05,
+      "loss": 3.3962,
+      "step": 2118144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5314574580227638e-05,
+      "loss": 3.3814,
+      "step": 2118656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5306188632717118e-05,
+      "loss": 3.4063,
+      "step": 2119168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.52978026852066e-05,
+      "loss": 3.4007,
+      "step": 2119680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.528941673769608e-05,
+      "loss": 3.3959,
+      "step": 2120192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.528104716898929e-05,
+      "loss": 3.3994,
+      "step": 2120704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5272677600282503e-05,
+      "loss": 3.3897,
+      "step": 2121216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.526429165277198e-05,
+      "loss": 3.3962,
+      "step": 2121728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5255905705261461e-05,
+      "loss": 3.3941,
+      "step": 2122240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5247519757750941e-05,
+      "loss": 3.3916,
+      "step": 2122752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5239150189044155e-05,
+      "loss": 3.3936,
+      "step": 2123264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5230764241533632e-05,
+      "loss": 3.3924,
+      "step": 2123776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5222378294023112e-05,
+      "loss": 3.3943,
+      "step": 2124288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5213992346512593e-05,
+      "loss": 3.3897,
+      "step": 2124800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5205606399002073e-05,
+      "loss": 3.3882,
+      "step": 2125312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5197220451491553e-05,
+      "loss": 3.3897,
+      "step": 2125824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5188834503981033e-05,
+      "loss": 3.3865,
+      "step": 2126336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5180448556470513e-05,
+      "loss": 3.404,
+      "step": 2126848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5172078987763724e-05,
+      "loss": 3.3913,
+      "step": 2127360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5163693040253204e-05,
+      "loss": 3.4085,
+      "step": 2127872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5155323471546415e-05,
+      "loss": 3.3767,
+      "step": 2128384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5146937524035895e-05,
+      "loss": 3.4077,
+      "step": 2128896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5138551576525376e-05,
+      "loss": 3.3914,
+      "step": 2129408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5130165629014856e-05,
+      "loss": 3.4017,
+      "step": 2129920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5121779681504336e-05,
+      "loss": 3.3922,
+      "step": 2130432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5113393733993814e-05,
+      "loss": 3.3839,
+      "step": 2130944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5105007786483294e-05,
+      "loss": 3.3998,
+      "step": 2131456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5096621838972774e-05,
+      "loss": 3.4175,
+      "step": 2131968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5088252270265987e-05,
+      "loss": 3.3796,
+      "step": 2132480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5079866322755465e-05,
+      "loss": 3.3963,
+      "step": 2132992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5071480375244945e-05,
+      "loss": 3.392,
+      "step": 2133504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5063094427734425e-05,
+      "loss": 3.3881,
+      "step": 2134016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5054724859027636e-05,
+      "loss": 3.3909,
+      "step": 2134528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5046338911517116e-05,
+      "loss": 3.3984,
+      "step": 2135040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5037969342810328e-05,
+      "loss": 3.39,
+      "step": 2135552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.502958339529981e-05,
+      "loss": 3.3966,
+      "step": 2136064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5021197447789286e-05,
+      "loss": 3.3953,
+      "step": 2136576
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.861250638961792,
+      "eval_runtime": 302.6574,
+      "eval_samples_per_second": 1260.802,
+      "eval_steps_per_second": 39.401,
+      "step": 2136960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5012811500278766e-05,
+      "loss": 3.396,
+      "step": 2137088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5004425552768248e-05,
+      "loss": 3.3853,
+      "step": 2137600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4996039605257728e-05,
+      "loss": 3.3991,
+      "step": 2138112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4987653657747208e-05,
+      "loss": 3.4037,
+      "step": 2138624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4979284089040419e-05,
+      "loss": 3.3988,
+      "step": 2139136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4970898141529899e-05,
+      "loss": 3.3927,
+      "step": 2139648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4962512194019379e-05,
+      "loss": 3.3964,
+      "step": 2140160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4954126246508859e-05,
+      "loss": 3.3913,
+      "step": 2140672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.494575667780207e-05,
+      "loss": 3.3891,
+      "step": 2141184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4937387109095282e-05,
+      "loss": 3.3961,
+      "step": 2141696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.492900116158476e-05,
+      "loss": 3.402,
+      "step": 2142208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.492061521407424e-05,
+      "loss": 3.4015,
+      "step": 2142720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.491222926656372e-05,
+      "loss": 3.4003,
+      "step": 2143232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4903843319053202e-05,
+      "loss": 3.3804,
+      "step": 2143744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4895457371542682e-05,
+      "loss": 3.398,
+      "step": 2144256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4887071424032162e-05,
+      "loss": 3.3816,
+      "step": 2144768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.487868547652164e-05,
+      "loss": 3.3898,
+      "step": 2145280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4870315907814852e-05,
+      "loss": 3.3869,
+      "step": 2145792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4861929960304332e-05,
+      "loss": 3.3843,
+      "step": 2146304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4853544012793812e-05,
+      "loss": 3.3919,
+      "step": 2146816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4845174444087023e-05,
+      "loss": 3.4067,
+      "step": 2147328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4836788496576503e-05,
+      "loss": 3.3963,
+      "step": 2147840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4828402549065985e-05,
+      "loss": 3.3953,
+      "step": 2148352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4820016601555465e-05,
+      "loss": 3.3869,
+      "step": 2148864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4811647032848674e-05,
+      "loss": 3.3937,
+      "step": 2149376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4803261085338155e-05,
+      "loss": 3.3861,
+      "step": 2149888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4794891516631365e-05,
+      "loss": 3.3758,
+      "step": 2150400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4786505569120846e-05,
+      "loss": 3.3983,
+      "step": 2150912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4778119621610326e-05,
+      "loss": 3.3839,
+      "step": 2151424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4769733674099806e-05,
+      "loss": 3.3896,
+      "step": 2151936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4761364105393017e-05,
+      "loss": 3.3907,
+      "step": 2152448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4752978157882497e-05,
+      "loss": 3.3965,
+      "step": 2152960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4744592210371977e-05,
+      "loss": 3.3869,
+      "step": 2153472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4736206262861457e-05,
+      "loss": 3.3908,
+      "step": 2153984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4727820315350937e-05,
+      "loss": 3.3911,
+      "step": 2154496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4719434367840415e-05,
+      "loss": 3.3866,
+      "step": 2155008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4711048420329895e-05,
+      "loss": 3.4001,
+      "step": 2155520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4702662472819375e-05,
+      "loss": 3.3749,
+      "step": 2156032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4694292904112586e-05,
+      "loss": 3.3823,
+      "step": 2156544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4685906956602066e-05,
+      "loss": 3.3829,
+      "step": 2157056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4677521009091547e-05,
+      "loss": 3.3866,
+      "step": 2157568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4669135061581027e-05,
+      "loss": 3.3812,
+      "step": 2158080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4660749114070507e-05,
+      "loss": 3.3898,
+      "step": 2158592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4652379545363718e-05,
+      "loss": 3.3931,
+      "step": 2159104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4643993597853198e-05,
+      "loss": 3.3874,
+      "step": 2159616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4635607650342678e-05,
+      "loss": 3.3876,
+      "step": 2160128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4627221702832158e-05,
+      "loss": 3.3969,
+      "step": 2160640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.461883575532164e-05,
+      "loss": 3.3837,
+      "step": 2161152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.461044980781112e-05,
+      "loss": 3.366,
+      "step": 2161664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4602063860300596e-05,
+      "loss": 3.3986,
+      "step": 2162176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4593677912790078e-05,
+      "loss": 3.3893,
+      "step": 2162688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.458530834408329e-05,
+      "loss": 3.3872,
+      "step": 2163200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4576938775376501e-05,
+      "loss": 3.3872,
+      "step": 2163712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4568552827865981e-05,
+      "loss": 3.3827,
+      "step": 2164224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4560166880355461e-05,
+      "loss": 3.3747,
+      "step": 2164736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.455178093284494e-05,
+      "loss": 3.3782,
+      "step": 2165248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4543394985334419e-05,
+      "loss": 3.3989,
+      "step": 2165760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4535009037823899e-05,
+      "loss": 3.3817,
+      "step": 2166272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4526623090313379e-05,
+      "loss": 3.4037,
+      "step": 2166784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4518237142802859e-05,
+      "loss": 3.3889,
+      "step": 2167296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.450986757409607e-05,
+      "loss": 3.3686,
+      "step": 2167808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.450148162658555e-05,
+      "loss": 3.3967,
+      "step": 2168320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4493095679075031e-05,
+      "loss": 3.3808,
+      "step": 2168832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4484709731564511e-05,
+      "loss": 3.3739,
+      "step": 2169344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4476323784053991e-05,
+      "loss": 3.3804,
+      "step": 2169856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4467937836543471e-05,
+      "loss": 3.3839,
+      "step": 2170368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.445955188903295e-05,
+      "loss": 3.3749,
+      "step": 2170880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4451182320326162e-05,
+      "loss": 3.3842,
+      "step": 2171392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4442796372815642e-05,
+      "loss": 3.3763,
+      "step": 2171904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4434410425305122e-05,
+      "loss": 3.3889,
+      "step": 2172416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44260244777946e-05,
+      "loss": 3.3923,
+      "step": 2172928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4417654909087813e-05,
+      "loss": 3.3848,
+      "step": 2173440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4409268961577294e-05,
+      "loss": 3.3943,
+      "step": 2173952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4400883014066774e-05,
+      "loss": 3.3886,
+      "step": 2174464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.439249706655625e-05,
+      "loss": 3.4023,
+      "step": 2174976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4384127497849465e-05,
+      "loss": 3.3843,
+      "step": 2175488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4375741550338945e-05,
+      "loss": 3.3981,
+      "step": 2176000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4367355602828421e-05,
+      "loss": 3.3856,
+      "step": 2176512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4358969655317903e-05,
+      "loss": 3.3793,
+      "step": 2177024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4350600086611116e-05,
+      "loss": 3.3948,
+      "step": 2177536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4342214139100596e-05,
+      "loss": 3.3857,
+      "step": 2178048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4333828191590074e-05,
+      "loss": 3.3957,
+      "step": 2178560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4325442244079554e-05,
+      "loss": 3.3995,
+      "step": 2179072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4317072675372766e-05,
+      "loss": 3.3821,
+      "step": 2179584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4308686727862245e-05,
+      "loss": 3.3653,
+      "step": 2180096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4300317159155457e-05,
+      "loss": 3.3794,
+      "step": 2180608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4291931211644937e-05,
+      "loss": 3.3942,
+      "step": 2181120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4283545264134419e-05,
+      "loss": 3.3834,
+      "step": 2181632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4275159316623895e-05,
+      "loss": 3.3851,
+      "step": 2182144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4266773369113375e-05,
+      "loss": 3.3882,
+      "step": 2182656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4258387421602857e-05,
+      "loss": 3.377,
+      "step": 2183168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4250001474092337e-05,
+      "loss": 3.392,
+      "step": 2183680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4241615526581817e-05,
+      "loss": 3.3814,
+      "step": 2184192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.423326233667876e-05,
+      "loss": 3.3912,
+      "step": 2184704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.422487638916824e-05,
+      "loss": 3.3939,
+      "step": 2185216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4216490441657718e-05,
+      "loss": 3.3944,
+      "step": 2185728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4208104494147198e-05,
+      "loss": 3.3795,
+      "step": 2186240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4199718546636678e-05,
+      "loss": 3.3872,
+      "step": 2186752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4191332599126158e-05,
+      "loss": 3.3858,
+      "step": 2187264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.418294665161564e-05,
+      "loss": 3.3867,
+      "step": 2187776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.417456070410512e-05,
+      "loss": 3.3915,
+      "step": 2188288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4166191135398329e-05,
+      "loss": 3.3852,
+      "step": 2188800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.415780518788781e-05,
+      "loss": 3.3826,
+      "step": 2189312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.414941924037729e-05,
+      "loss": 3.3795,
+      "step": 2189824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4141049671670501e-05,
+      "loss": 3.3859,
+      "step": 2190336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4132663724159981e-05,
+      "loss": 3.3803,
+      "step": 2190848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4124277776649461e-05,
+      "loss": 3.3682,
+      "step": 2191360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4115891829138941e-05,
+      "loss": 3.3864,
+      "step": 2191872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4107505881628421e-05,
+      "loss": 3.3916,
+      "step": 2192384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4099136312921632e-05,
+      "loss": 3.3776,
+      "step": 2192896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4090750365411112e-05,
+      "loss": 3.3816,
+      "step": 2193408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4082364417900593e-05,
+      "loss": 3.3805,
+      "step": 2193920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4073978470390073e-05,
+      "loss": 3.3902,
+      "step": 2194432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.406559252287955e-05,
+      "loss": 3.3749,
+      "step": 2194944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.405720657536903e-05,
+      "loss": 3.3905,
+      "step": 2195456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4048820627858512e-05,
+      "loss": 3.3883,
+      "step": 2195968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4040434680347992e-05,
+      "loss": 3.3928,
+      "step": 2196480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4032065111641202e-05,
+      "loss": 3.3893,
+      "step": 2196992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4023679164130682e-05,
+      "loss": 3.3793,
+      "step": 2197504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4015309595423895e-05,
+      "loss": 3.3853,
+      "step": 2198016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4006923647913373e-05,
+      "loss": 3.3837,
+      "step": 2198528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3998537700402853e-05,
+      "loss": 3.3838,
+      "step": 2199040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3990151752892333e-05,
+      "loss": 3.3844,
+      "step": 2199552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3981765805381813e-05,
+      "loss": 3.3814,
+      "step": 2200064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3973379857871295e-05,
+      "loss": 3.3862,
+      "step": 2200576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3964993910360774e-05,
+      "loss": 3.3794,
+      "step": 2201088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3956624341653984e-05,
+      "loss": 3.3801,
+      "step": 2201600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3948254772947198e-05,
+      "loss": 3.3749,
+      "step": 2202112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3939868825436674e-05,
+      "loss": 3.3768,
+      "step": 2202624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3931482877926156e-05,
+      "loss": 3.3951,
+      "step": 2203136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3923096930415636e-05,
+      "loss": 3.3837,
+      "step": 2203648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3914710982905116e-05,
+      "loss": 3.4005,
+      "step": 2204160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3906325035394596e-05,
+      "loss": 3.3697,
+      "step": 2204672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3897939087884076e-05,
+      "loss": 3.3909,
+      "step": 2205184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3889569519177287e-05,
+      "loss": 3.3882,
+      "step": 2205696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3881199950470497e-05,
+      "loss": 3.3905,
+      "step": 2206208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3872814002959977e-05,
+      "loss": 3.3842,
+      "step": 2206720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3864428055449457e-05,
+      "loss": 3.3757,
+      "step": 2207232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3856042107938937e-05,
+      "loss": 3.3878,
+      "step": 2207744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3847656160428419e-05,
+      "loss": 3.4025,
+      "step": 2208256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3839270212917899e-05,
+      "loss": 3.3751,
+      "step": 2208768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3830884265407379e-05,
+      "loss": 3.3842,
+      "step": 2209280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3822498317896857e-05,
+      "loss": 3.3831,
+      "step": 2209792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.381412874919007e-05,
+      "loss": 3.38,
+      "step": 2210304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.380574280167955e-05,
+      "loss": 3.384,
+      "step": 2210816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3797356854169028e-05,
+      "loss": 3.3862,
+      "step": 2211328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3788970906658508e-05,
+      "loss": 3.3795,
+      "step": 2211840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.378060133795172e-05,
+      "loss": 3.3877,
+      "step": 2212352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3772215390441202e-05,
+      "loss": 3.3835,
+      "step": 2212864
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.861837863922119,
+      "eval_runtime": 304.9052,
+      "eval_samples_per_second": 1251.507,
+      "eval_steps_per_second": 39.111,
+      "step": 2213280
+    }
+  ],
+  "logging_steps": 512,
+  "max_steps": 3052726,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 10,
+  "total_flos": 1.546857469920043e+18,
+  "trial_name": null,
+  "trial_params": null
+}