diff --git "a/checkpoint-2442240/trainer_state.json" "b/checkpoint-2442240/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2442240/trainer_state.json"
@@ -0,0 +1,28901 @@
+{
+  "best_metric": 3.9805712699890137,
+  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/passive/lstm/3/checkpoints/checkpoint-2442240",
+  "epoch": 0.025000606015738065,
+  "eval_steps": 10,
+  "global_step": 2442240,
+  "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": 10.8209,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999161405248948e-05,
+      "loss": 7.5568,
+      "step": 512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998322810497896e-05,
+      "loss": 7.0563,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997484215746844e-05,
+      "loss": 6.9949,
+      "step": 1536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996645620995792e-05,
+      "loss": 6.9457,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99580702624474e-05,
+      "loss": 6.8667,
+      "step": 2560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994968431493688e-05,
+      "loss": 6.7087,
+      "step": 3072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994129836742636e-05,
+      "loss": 6.6099,
+      "step": 3584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993291241991584e-05,
+      "loss": 6.5194,
+      "step": 4096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992452647240532e-05,
+      "loss": 6.4426,
+      "step": 4608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99161405248948e-05,
+      "loss": 6.3859,
+      "step": 5120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.990775457738428e-05,
+      "loss": 6.3178,
+      "step": 5632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989936862987376e-05,
+      "loss": 6.2549,
+      "step": 6144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989098268236324e-05,
+      "loss": 6.1917,
+      "step": 6656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.988259673485272e-05,
+      "loss": 6.14,
+      "step": 7168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.98742107873422e-05,
+      "loss": 6.0751,
+      "step": 7680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986582483983168e-05,
+      "loss": 6.0418,
+      "step": 8192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.985743889232116e-05,
+      "loss": 5.9947,
+      "step": 8704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984905294481064e-05,
+      "loss": 5.9555,
+      "step": 9216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984066699730012e-05,
+      "loss": 5.917,
+      "step": 9728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.983229742859333e-05,
+      "loss": 5.8862,
+      "step": 10240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.982391148108281e-05,
+      "loss": 5.8472,
+      "step": 10752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.981552553357229e-05,
+      "loss": 5.8198,
+      "step": 11264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9807155964865506e-05,
+      "loss": 5.778,
+      "step": 11776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9798770017354986e-05,
+      "loss": 5.7659,
+      "step": 12288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9790384069844466e-05,
+      "loss": 5.7291,
+      "step": 12800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9781998122333946e-05,
+      "loss": 5.7097,
+      "step": 13312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9773612174823426e-05,
+      "loss": 5.6721,
+      "step": 13824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.97652262273129e-05,
+      "loss": 5.6552,
+      "step": 14336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.975684027980238e-05,
+      "loss": 5.6355,
+      "step": 14848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.974845433229186e-05,
+      "loss": 5.6104,
+      "step": 15360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.974008476358507e-05,
+      "loss": 5.6008,
+      "step": 15872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973169881607455e-05,
+      "loss": 5.5742,
+      "step": 16384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972331286856403e-05,
+      "loss": 5.5579,
+      "step": 16896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9714926921053515e-05,
+      "loss": 5.5499,
+      "step": 17408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9706540973542995e-05,
+      "loss": 5.5264,
+      "step": 17920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9698171404836204e-05,
+      "loss": 5.5173,
+      "step": 18432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9689785457325684e-05,
+      "loss": 5.471,
+      "step": 18944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9681399509815164e-05,
+      "loss": 5.477,
+      "step": 19456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9673013562304644e-05,
+      "loss": 5.4379,
+      "step": 19968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9664627614794124e-05,
+      "loss": 5.4437,
+      "step": 20480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965625804608733e-05,
+      "loss": 5.4203,
+      "step": 20992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964787209857681e-05,
+      "loss": 5.4179,
+      "step": 21504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963948615106629e-05,
+      "loss": 5.3885,
+      "step": 22016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963110020355577e-05,
+      "loss": 5.3944,
+      "step": 22528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962271425604525e-05,
+      "loss": 5.3831,
+      "step": 23040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.961432830853473e-05,
+      "loss": 5.3749,
+      "step": 23552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960595873982795e-05,
+      "loss": 5.3678,
+      "step": 24064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.959757279231743e-05,
+      "loss": 5.3383,
+      "step": 24576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958918684480691e-05,
+      "loss": 5.3293,
+      "step": 25088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958080089729639e-05,
+      "loss": 5.3294,
+      "step": 25600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.957241494978587e-05,
+      "loss": 5.3153,
+      "step": 26112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.956402900227535e-05,
+      "loss": 5.3102,
+      "step": 26624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.955564305476483e-05,
+      "loss": 5.2874,
+      "step": 27136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.954727348605804e-05,
+      "loss": 5.2838,
+      "step": 27648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953888753854752e-05,
+      "loss": 5.2638,
+      "step": 28160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9530501591037e-05,
+      "loss": 5.2849,
+      "step": 28672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.952211564352648e-05,
+      "loss": 5.2387,
+      "step": 29184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951372969601596e-05,
+      "loss": 5.2546,
+      "step": 29696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.950534374850543e-05,
+      "loss": 5.2441,
+      "step": 30208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.949695780099492e-05,
+      "loss": 5.2235,
+      "step": 30720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948858823228813e-05,
+      "loss": 5.2221,
+      "step": 31232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948020228477761e-05,
+      "loss": 5.2082,
+      "step": 31744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9471816337267086e-05,
+      "loss": 5.1982,
+      "step": 32256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9463430389756566e-05,
+      "loss": 5.191,
+      "step": 32768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.945507719985351e-05,
+      "loss": 5.2049,
+      "step": 33280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.944669125234299e-05,
+      "loss": 5.1792,
+      "step": 33792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.943830530483247e-05,
+      "loss": 5.1701,
+      "step": 34304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.942991935732195e-05,
+      "loss": 5.1601,
+      "step": 34816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.942153340981143e-05,
+      "loss": 5.157,
+      "step": 35328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9413147462300904e-05,
+      "loss": 5.1632,
+      "step": 35840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9404761514790384e-05,
+      "loss": 5.1577,
+      "step": 36352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.939637556727987e-05,
+      "loss": 5.1337,
+      "step": 36864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.938798961976935e-05,
+      "loss": 5.1489,
+      "step": 37376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9379636429866296e-05,
+      "loss": 5.1414,
+      "step": 37888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9371250482355776e-05,
+      "loss": 5.1144,
+      "step": 38400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9362864534845256e-05,
+      "loss": 5.1127,
+      "step": 38912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.935447858733473e-05,
+      "loss": 5.103,
+      "step": 39424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.934609263982421e-05,
+      "loss": 5.0947,
+      "step": 39936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.933770669231369e-05,
+      "loss": 5.0891,
+      "step": 40448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.932932074480317e-05,
+      "loss": 5.0944,
+      "step": 40960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.932093479729265e-05,
+      "loss": 5.0902,
+      "step": 41472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931254884978213e-05,
+      "loss": 5.0915,
+      "step": 41984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.930417928107534e-05,
+      "loss": 5.0665,
+      "step": 42496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9295793333564825e-05,
+      "loss": 5.0488,
+      "step": 43008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9287423764858034e-05,
+      "loss": 5.0552,
+      "step": 43520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9279037817347514e-05,
+      "loss": 5.0559,
+      "step": 44032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9270651869836994e-05,
+      "loss": 5.0544,
+      "step": 44544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9262265922326474e-05,
+      "loss": 5.0371,
+      "step": 45056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9253879974815954e-05,
+      "loss": 5.0355,
+      "step": 45568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9245494027305433e-05,
+      "loss": 5.0355,
+      "step": 46080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9237108079794913e-05,
+      "loss": 5.0281,
+      "step": 46592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922872213228439e-05,
+      "loss": 5.0165,
+      "step": 47104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922033618477387e-05,
+      "loss": 5.0128,
+      "step": 47616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921195023726335e-05,
+      "loss": 5.0001,
+      "step": 48128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.920356428975283e-05,
+      "loss": 5.0129,
+      "step": 48640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919519472104604e-05,
+      "loss": 5.0006,
+      "step": 49152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918680877353552e-05,
+      "loss": 4.979,
+      "step": 49664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917842282602501e-05,
+      "loss": 4.9847,
+      "step": 50176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917003687851449e-05,
+      "loss": 4.982,
+      "step": 50688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.916165093100397e-05,
+      "loss": 4.9889,
+      "step": 51200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915328136229718e-05,
+      "loss": 4.9709,
+      "step": 51712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914489541478666e-05,
+      "loss": 4.9677,
+      "step": 52224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.913650946727614e-05,
+      "loss": 4.9518,
+      "step": 52736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.912812351976562e-05,
+      "loss": 4.9576,
+      "step": 53248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911973757225509e-05,
+      "loss": 4.9371,
+      "step": 53760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911136800354831e-05,
+      "loss": 4.9483,
+      "step": 54272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.910298205603779e-05,
+      "loss": 4.9349,
+      "step": 54784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.909459610852727e-05,
+      "loss": 4.9384,
+      "step": 55296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.908621016101675e-05,
+      "loss": 4.9274,
+      "step": 55808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.907782421350623e-05,
+      "loss": 4.9304,
+      "step": 56320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906943826599571e-05,
+      "loss": 4.9166,
+      "step": 56832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906105231848519e-05,
+      "loss": 4.9207,
+      "step": 57344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.905266637097467e-05,
+      "loss": 4.9128,
+      "step": 57856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9044296802267876e-05,
+      "loss": 4.9061,
+      "step": 58368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9035910854757356e-05,
+      "loss": 4.9177,
+      "step": 58880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9027524907246836e-05,
+      "loss": 4.9003,
+      "step": 59392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9019155338540045e-05,
+      "loss": 4.9002,
+      "step": 59904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9010769391029525e-05,
+      "loss": 4.8912,
+      "step": 60416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9002383443519005e-05,
+      "loss": 4.8944,
+      "step": 60928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8994013874812214e-05,
+      "loss": 4.8892,
+      "step": 61440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89856279273017e-05,
+      "loss": 4.8825,
+      "step": 61952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897724197979118e-05,
+      "loss": 4.8876,
+      "step": 62464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.896885603228066e-05,
+      "loss": 4.874,
+      "step": 62976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.896047008477014e-05,
+      "loss": 4.8753,
+      "step": 63488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.895208413725962e-05,
+      "loss": 4.8537,
+      "step": 64000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89436981897491e-05,
+      "loss": 4.8603,
+      "step": 64512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.893531224223858e-05,
+      "loss": 4.8618,
+      "step": 65024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.892694267353179e-05,
+      "loss": 4.8521,
+      "step": 65536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891855672602127e-05,
+      "loss": 4.8429,
+      "step": 66048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891017077851075e-05,
+      "loss": 4.8658,
+      "step": 66560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.890178483100023e-05,
+      "loss": 4.8481,
+      "step": 67072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.889339888348971e-05,
+      "loss": 4.8407,
+      "step": 67584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.888501293597919e-05,
+      "loss": 4.8387,
+      "step": 68096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88766433672724e-05,
+      "loss": 4.8334,
+      "step": 68608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.886825741976188e-05,
+      "loss": 4.8353,
+      "step": 69120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8859871472251365e-05,
+      "loss": 4.8324,
+      "step": 69632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8851485524740845e-05,
+      "loss": 4.8275,
+      "step": 70144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8843099577230325e-05,
+      "loss": 4.829,
+      "step": 70656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8834713629719805e-05,
+      "loss": 4.828,
+      "step": 71168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.882632768220928e-05,
+      "loss": 4.8157,
+      "step": 71680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.881794173469876e-05,
+      "loss": 4.8062,
+      "step": 72192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8809572165991974e-05,
+      "loss": 4.8178,
+      "step": 72704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.880120259728518e-05,
+      "loss": 4.8037,
+      "step": 73216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879281664977466e-05,
+      "loss": 4.8074,
+      "step": 73728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.878443070226414e-05,
+      "loss": 4.7987,
+      "step": 74240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.877604475475362e-05,
+      "loss": 4.7962,
+      "step": 74752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.87676588072431e-05,
+      "loss": 4.8014,
+      "step": 75264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875927285973258e-05,
+      "loss": 4.7887,
+      "step": 75776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.87509032910258e-05,
+      "loss": 4.7862,
+      "step": 76288
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.75404691696167,
+      "eval_runtime": 294.4935,
+      "eval_samples_per_second": 1295.754,
+      "eval_steps_per_second": 40.493,
+      "step": 76320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.874251734351528e-05,
+      "loss": 4.7728,
+      "step": 76800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.873414777480849e-05,
+      "loss": 4.7719,
+      "step": 77312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.872576182729797e-05,
+      "loss": 4.7959,
+      "step": 77824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.871737587978745e-05,
+      "loss": 4.7803,
+      "step": 78336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.870898993227693e-05,
+      "loss": 4.7712,
+      "step": 78848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.87006039847664e-05,
+      "loss": 4.7681,
+      "step": 79360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.869221803725588e-05,
+      "loss": 4.7691,
+      "step": 79872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.868383208974536e-05,
+      "loss": 4.7519,
+      "step": 80384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8675462521038576e-05,
+      "loss": 4.7617,
+      "step": 80896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8667076573528056e-05,
+      "loss": 4.7605,
+      "step": 81408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8658690626017536e-05,
+      "loss": 4.7619,
+      "step": 81920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8650304678507016e-05,
+      "loss": 4.7615,
+      "step": 82432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8641935109800225e-05,
+      "loss": 4.7428,
+      "step": 82944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8633549162289705e-05,
+      "loss": 4.7437,
+      "step": 83456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8625163214779185e-05,
+      "loss": 4.7324,
+      "step": 83968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8616777267268665e-05,
+      "loss": 4.7343,
+      "step": 84480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8608391319758145e-05,
+      "loss": 4.7403,
+      "step": 84992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8600005372247625e-05,
+      "loss": 4.7298,
+      "step": 85504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8591619424737105e-05,
+      "loss": 4.7359,
+      "step": 86016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8583233477226585e-05,
+      "loss": 4.7484,
+      "step": 86528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8574847529716065e-05,
+      "loss": 4.7308,
+      "step": 87040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8566477961009274e-05,
+      "loss": 4.7333,
+      "step": 87552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8558092013498754e-05,
+      "loss": 4.7154,
+      "step": 88064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.854970606598824e-05,
+      "loss": 4.7366,
+      "step": 88576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.854132011847772e-05,
+      "loss": 4.7145,
+      "step": 89088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.85329341709672e-05,
+      "loss": 4.7163,
+      "step": 89600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.852454822345668e-05,
+      "loss": 4.7035,
+      "step": 90112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.851617865474989e-05,
+      "loss": 4.7129,
+      "step": 90624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.850779270723937e-05,
+      "loss": 4.7032,
+      "step": 91136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.849940675972885e-05,
+      "loss": 4.6993,
+      "step": 91648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.849102081221833e-05,
+      "loss": 4.7045,
+      "step": 92160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.848263486470781e-05,
+      "loss": 4.7088,
+      "step": 92672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.847426529600102e-05,
+      "loss": 4.702,
+      "step": 93184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.84658793484905e-05,
+      "loss": 4.7027,
+      "step": 93696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.845750977978371e-05,
+      "loss": 4.6948,
+      "step": 94208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8449123832273194e-05,
+      "loss": 4.702,
+      "step": 94720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8440737884762674e-05,
+      "loss": 4.6742,
+      "step": 95232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8432351937252154e-05,
+      "loss": 4.6881,
+      "step": 95744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8423965989741634e-05,
+      "loss": 4.6658,
+      "step": 96256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8415580042231114e-05,
+      "loss": 4.6852,
+      "step": 96768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.840719409472059e-05,
+      "loss": 4.669,
+      "step": 97280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.839880814721007e-05,
+      "loss": 4.6818,
+      "step": 97792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.839043857850328e-05,
+      "loss": 4.6603,
+      "step": 98304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.838205263099276e-05,
+      "loss": 4.6788,
+      "step": 98816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8373666683482236e-05,
+      "loss": 4.6764,
+      "step": 99328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8365280735971716e-05,
+      "loss": 4.6732,
+      "step": 99840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8356894788461196e-05,
+      "loss": 4.6797,
+      "step": 100352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834852521975441e-05,
+      "loss": 4.6501,
+      "step": 100864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834013927224389e-05,
+      "loss": 4.6538,
+      "step": 101376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.833175332473337e-05,
+      "loss": 4.6663,
+      "step": 101888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832336737722285e-05,
+      "loss": 4.6582,
+      "step": 102400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831498142971233e-05,
+      "loss": 4.6559,
+      "step": 102912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.830659548220181e-05,
+      "loss": 4.6488,
+      "step": 103424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.829820953469129e-05,
+      "loss": 4.6436,
+      "step": 103936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.828982358718077e-05,
+      "loss": 4.6389,
+      "step": 104448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.828145401847398e-05,
+      "loss": 4.6534,
+      "step": 104960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.827306807096346e-05,
+      "loss": 4.6291,
+      "step": 105472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.826468212345294e-05,
+      "loss": 4.641,
+      "step": 105984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.825629617594242e-05,
+      "loss": 4.6469,
+      "step": 106496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.82479102284319e-05,
+      "loss": 4.6295,
+      "step": 107008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.823952428092138e-05,
+      "loss": 4.6298,
+      "step": 107520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.823113833341086e-05,
+      "loss": 4.6277,
+      "step": 108032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.822275238590034e-05,
+      "loss": 4.6148,
+      "step": 108544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8214382817193557e-05,
+      "loss": 4.6224,
+      "step": 109056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8205996869683037e-05,
+      "loss": 4.6378,
+      "step": 109568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8197610922172516e-05,
+      "loss": 4.6195,
+      "step": 110080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8189224974661996e-05,
+      "loss": 4.6139,
+      "step": 110592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8180855405955206e-05,
+      "loss": 4.6128,
+      "step": 111104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8172469458444685e-05,
+      "loss": 4.6137,
+      "step": 111616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8164083510934165e-05,
+      "loss": 4.6279,
+      "step": 112128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8155697563423645e-05,
+      "loss": 4.6195,
+      "step": 112640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8147327994716854e-05,
+      "loss": 4.6104,
+      "step": 113152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8138942047206334e-05,
+      "loss": 4.624,
+      "step": 113664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8130556099695814e-05,
+      "loss": 4.6227,
+      "step": 114176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.812218653098903e-05,
+      "loss": 4.6056,
+      "step": 114688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.811380058347851e-05,
+      "loss": 4.6045,
+      "step": 115200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.810541463596799e-05,
+      "loss": 4.6032,
+      "step": 115712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.809702868845747e-05,
+      "loss": 4.6,
+      "step": 116224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.808864274094695e-05,
+      "loss": 4.5967,
+      "step": 116736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.808027317224016e-05,
+      "loss": 4.6001,
+      "step": 117248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.807188722472964e-05,
+      "loss": 4.6069,
+      "step": 117760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.806350127721912e-05,
+      "loss": 4.6072,
+      "step": 118272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.80551153297086e-05,
+      "loss": 4.595,
+      "step": 118784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.804672938219807e-05,
+      "loss": 4.5728,
+      "step": 119296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.803834343468755e-05,
+      "loss": 4.5889,
+      "step": 119808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.802995748717703e-05,
+      "loss": 4.5938,
+      "step": 120320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.802157153966652e-05,
+      "loss": 4.595,
+      "step": 120832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.801320197095973e-05,
+      "loss": 4.587,
+      "step": 121344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.800481602344921e-05,
+      "loss": 4.5837,
+      "step": 121856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7996446454742424e-05,
+      "loss": 4.5903,
+      "step": 122368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.79880605072319e-05,
+      "loss": 4.5806,
+      "step": 122880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.797967455972138e-05,
+      "loss": 4.5765,
+      "step": 123392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.797128861221086e-05,
+      "loss": 4.5836,
+      "step": 123904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.796290266470034e-05,
+      "loss": 4.5677,
+      "step": 124416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7954533095993546e-05,
+      "loss": 4.5869,
+      "step": 124928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7946147148483026e-05,
+      "loss": 4.5784,
+      "step": 125440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7937761200972506e-05,
+      "loss": 4.5583,
+      "step": 125952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7929375253461986e-05,
+      "loss": 4.5695,
+      "step": 126464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.79210056847552e-05,
+      "loss": 4.5689,
+      "step": 126976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.791261973724468e-05,
+      "loss": 4.5807,
+      "step": 127488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.790423378973416e-05,
+      "loss": 4.5667,
+      "step": 128000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.789584784222364e-05,
+      "loss": 4.562,
+      "step": 128512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.788746189471312e-05,
+      "loss": 4.5555,
+      "step": 129024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.787909232600633e-05,
+      "loss": 4.5643,
+      "step": 129536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.787070637849581e-05,
+      "loss": 4.5466,
+      "step": 130048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.786232043098529e-05,
+      "loss": 4.555,
+      "step": 130560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.785393448347477e-05,
+      "loss": 4.5523,
+      "step": 131072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.784554853596425e-05,
+      "loss": 4.5537,
+      "step": 131584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.783716258845373e-05,
+      "loss": 4.5444,
+      "step": 132096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.782877664094321e-05,
+      "loss": 4.5542,
+      "step": 132608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.782039069343269e-05,
+      "loss": 4.5456,
+      "step": 133120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.781200474592217e-05,
+      "loss": 4.5466,
+      "step": 133632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7803635177215386e-05,
+      "loss": 4.5481,
+      "step": 134144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7795249229704866e-05,
+      "loss": 4.5391,
+      "step": 134656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7786863282194346e-05,
+      "loss": 4.5621,
+      "step": 135168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7778493713487555e-05,
+      "loss": 4.5475,
+      "step": 135680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7770107765977035e-05,
+      "loss": 4.5449,
+      "step": 136192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7761721818466515e-05,
+      "loss": 4.5359,
+      "step": 136704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7753335870955995e-05,
+      "loss": 4.5425,
+      "step": 137216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7744966302249204e-05,
+      "loss": 4.5402,
+      "step": 137728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7736580354738684e-05,
+      "loss": 4.532,
+      "step": 138240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7728194407228164e-05,
+      "loss": 4.5474,
+      "step": 138752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7719808459717644e-05,
+      "loss": 4.537,
+      "step": 139264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.771143889101086e-05,
+      "loss": 4.5293,
+      "step": 139776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.770305294350034e-05,
+      "loss": 4.5245,
+      "step": 140288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.769466699598982e-05,
+      "loss": 4.5262,
+      "step": 140800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.76862810484793e-05,
+      "loss": 4.5262,
+      "step": 141312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.767789510096878e-05,
+      "loss": 4.5286,
+      "step": 141824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.766950915345826e-05,
+      "loss": 4.5142,
+      "step": 142336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.766112320594773e-05,
+      "loss": 4.5427,
+      "step": 142848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.765273725843721e-05,
+      "loss": 4.5329,
+      "step": 143360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.764435131092669e-05,
+      "loss": 4.5214,
+      "step": 143872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.763599812102364e-05,
+      "loss": 4.5208,
+      "step": 144384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.762761217351312e-05,
+      "loss": 4.5145,
+      "step": 144896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.76192262260026e-05,
+      "loss": 4.5281,
+      "step": 145408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.761084027849208e-05,
+      "loss": 4.5241,
+      "step": 145920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.760245433098156e-05,
+      "loss": 4.5192,
+      "step": 146432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.759406838347104e-05,
+      "loss": 4.5236,
+      "step": 146944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.758568243596052e-05,
+      "loss": 4.5215,
+      "step": 147456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.757731286725373e-05,
+      "loss": 4.5189,
+      "step": 147968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7568926919743206e-05,
+      "loss": 4.5086,
+      "step": 148480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7560540972232686e-05,
+      "loss": 4.5205,
+      "step": 148992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7552155024722166e-05,
+      "loss": 4.5103,
+      "step": 149504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7543769077211646e-05,
+      "loss": 4.5139,
+      "step": 150016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7535383129701126e-05,
+      "loss": 4.5072,
+      "step": 150528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7526997182190606e-05,
+      "loss": 4.5097,
+      "step": 151040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7518611234680086e-05,
+      "loss": 4.5106,
+      "step": 151552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7510241665973295e-05,
+      "loss": 4.4977,
+      "step": 152064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.750185571846278e-05,
+      "loss": 4.5018,
+      "step": 152576
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.476864814758301,
+      "eval_runtime": 295.321,
+      "eval_samples_per_second": 1292.123,
+      "eval_steps_per_second": 40.38,
+      "step": 152640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.749346977095226e-05,
+      "loss": 4.4904,
+      "step": 153088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.748508382344174e-05,
+      "loss": 4.4898,
+      "step": 153600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.747669787593122e-05,
+      "loss": 4.5159,
+      "step": 154112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.74683119284207e-05,
+      "loss": 4.5074,
+      "step": 154624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.745992598091018e-05,
+      "loss": 4.4951,
+      "step": 155136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.745154003339966e-05,
+      "loss": 4.4946,
+      "step": 155648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.744315408588914e-05,
+      "loss": 4.4962,
+      "step": 156160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.743476813837862e-05,
+      "loss": 4.4806,
+      "step": 156672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7426382190868095e-05,
+      "loss": 4.4997,
+      "step": 157184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7417996243357575e-05,
+      "loss": 4.4921,
+      "step": 157696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7409610295847055e-05,
+      "loss": 4.4957,
+      "step": 158208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7401224348336535e-05,
+      "loss": 4.4977,
+      "step": 158720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.739285477962975e-05,
+      "loss": 4.4807,
+      "step": 159232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.738446883211923e-05,
+      "loss": 4.4857,
+      "step": 159744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.737608288460871e-05,
+      "loss": 4.479,
+      "step": 160256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.736769693709819e-05,
+      "loss": 4.4784,
+      "step": 160768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.735931098958767e-05,
+      "loss": 4.4827,
+      "step": 161280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.735092504207715e-05,
+      "loss": 4.4749,
+      "step": 161792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.734253909456663e-05,
+      "loss": 4.481,
+      "step": 162304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.733415314705611e-05,
+      "loss": 4.5013,
+      "step": 162816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.732576719954559e-05,
+      "loss": 4.4794,
+      "step": 163328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.731738125203507e-05,
+      "loss": 4.4827,
+      "step": 163840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.730899530452455e-05,
+      "loss": 4.4755,
+      "step": 164352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.730062573581776e-05,
+      "loss": 4.4884,
+      "step": 164864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.729223978830724e-05,
+      "loss": 4.4684,
+      "step": 165376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.728385384079672e-05,
+      "loss": 4.4794,
+      "step": 165888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.72754678932862e-05,
+      "loss": 4.4631,
+      "step": 166400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.726708194577568e-05,
+      "loss": 4.4686,
+      "step": 166912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.725869599826516e-05,
+      "loss": 4.465,
+      "step": 167424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.725031005075464e-05,
+      "loss": 4.4623,
+      "step": 167936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.724192410324412e-05,
+      "loss": 4.4693,
+      "step": 168448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7233570913341064e-05,
+      "loss": 4.4752,
+      "step": 168960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7225184965830544e-05,
+      "loss": 4.4674,
+      "step": 169472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7216799018320024e-05,
+      "loss": 4.4732,
+      "step": 169984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.720842944961323e-05,
+      "loss": 4.4586,
+      "step": 170496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.720004350210271e-05,
+      "loss": 4.4756,
+      "step": 171008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.719165755459219e-05,
+      "loss": 4.443,
+      "step": 171520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.718327160708167e-05,
+      "loss": 4.4657,
+      "step": 172032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.717488565957115e-05,
+      "loss": 4.4399,
+      "step": 172544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.716649971206063e-05,
+      "loss": 4.4612,
+      "step": 173056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.715811376455011e-05,
+      "loss": 4.4485,
+      "step": 173568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.714972781703959e-05,
+      "loss": 4.4624,
+      "step": 174080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.714134186952907e-05,
+      "loss": 4.4441,
+      "step": 174592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.713295592201855e-05,
+      "loss": 4.456,
+      "step": 175104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.712456997450803e-05,
+      "loss": 4.4587,
+      "step": 175616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.711618402699751e-05,
+      "loss": 4.4576,
+      "step": 176128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.710781445829072e-05,
+      "loss": 4.4598,
+      "step": 176640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.70994285107802e-05,
+      "loss": 4.4386,
+      "step": 177152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.709104256326968e-05,
+      "loss": 4.4354,
+      "step": 177664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.708265661575916e-05,
+      "loss": 4.4563,
+      "step": 178176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.707428704705237e-05,
+      "loss": 4.4488,
+      "step": 178688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.706590109954185e-05,
+      "loss": 4.4428,
+      "step": 179200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705751515203134e-05,
+      "loss": 4.441,
+      "step": 179712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7049145583324546e-05,
+      "loss": 4.4372,
+      "step": 180224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7040759635814026e-05,
+      "loss": 4.4278,
+      "step": 180736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7032373688303506e-05,
+      "loss": 4.443,
+      "step": 181248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7023987740792986e-05,
+      "loss": 4.4233,
+      "step": 181760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7015601793282466e-05,
+      "loss": 4.4356,
+      "step": 182272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7007215845771946e-05,
+      "loss": 4.4463,
+      "step": 182784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6998829898261426e-05,
+      "loss": 4.4244,
+      "step": 183296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6990443950750906e-05,
+      "loss": 4.4282,
+      "step": 183808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6982074382044115e-05,
+      "loss": 4.4301,
+      "step": 184320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6973688434533595e-05,
+      "loss": 4.4157,
+      "step": 184832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6965302487023075e-05,
+      "loss": 4.4185,
+      "step": 185344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6956916539512555e-05,
+      "loss": 4.4405,
+      "step": 185856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6948530592002035e-05,
+      "loss": 4.4234,
+      "step": 186368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.694016102329525e-05,
+      "loss": 4.417,
+      "step": 186880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.693179145458846e-05,
+      "loss": 4.4173,
+      "step": 187392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.692340550707794e-05,
+      "loss": 4.4221,
+      "step": 187904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.691501955956742e-05,
+      "loss": 4.4354,
+      "step": 188416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.69066336120569e-05,
+      "loss": 4.4297,
+      "step": 188928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.689824766454638e-05,
+      "loss": 4.4174,
+      "step": 189440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.688986171703586e-05,
+      "loss": 4.4331,
+      "step": 189952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.688147576952534e-05,
+      "loss": 4.4278,
+      "step": 190464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.687308982201481e-05,
+      "loss": 4.4187,
+      "step": 190976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.686472025330803e-05,
+      "loss": 4.415,
+      "step": 191488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.685633430579751e-05,
+      "loss": 4.4204,
+      "step": 192000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6847964737090725e-05,
+      "loss": 4.4136,
+      "step": 192512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6839578789580204e-05,
+      "loss": 4.4126,
+      "step": 193024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6831192842069684e-05,
+      "loss": 4.4129,
+      "step": 193536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6822806894559164e-05,
+      "loss": 4.4213,
+      "step": 194048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6814420947048644e-05,
+      "loss": 4.4263,
+      "step": 194560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.680603499953812e-05,
+      "loss": 4.4119,
+      "step": 195072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6797665430831333e-05,
+      "loss": 4.3973,
+      "step": 195584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.678929586212454e-05,
+      "loss": 4.4047,
+      "step": 196096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.678090991461402e-05,
+      "loss": 4.4111,
+      "step": 196608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.67725239671035e-05,
+      "loss": 4.4173,
+      "step": 197120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.676413801959298e-05,
+      "loss": 4.4071,
+      "step": 197632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.675575207208246e-05,
+      "loss": 4.4053,
+      "step": 198144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.674736612457194e-05,
+      "loss": 4.4154,
+      "step": 198656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.673898017706142e-05,
+      "loss": 4.4014,
+      "step": 199168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.673061060835464e-05,
+      "loss": 4.4028,
+      "step": 199680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.672222466084412e-05,
+      "loss": 4.4053,
+      "step": 200192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.671383871333359e-05,
+      "loss": 4.3974,
+      "step": 200704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.670545276582307e-05,
+      "loss": 4.4155,
+      "step": 201216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.669706681831255e-05,
+      "loss": 4.4004,
+      "step": 201728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.668868087080203e-05,
+      "loss": 4.3895,
+      "step": 202240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.668029492329151e-05,
+      "loss": 4.3991,
+      "step": 202752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.667190897578099e-05,
+      "loss": 4.3966,
+      "step": 203264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.666352302827047e-05,
+      "loss": 4.412,
+      "step": 203776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.665513708075995e-05,
+      "loss": 4.4012,
+      "step": 204288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.664675113324943e-05,
+      "loss": 4.3922,
+      "step": 204800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.663838156454265e-05,
+      "loss": 4.3852,
+      "step": 205312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.662999561703213e-05,
+      "loss": 4.3967,
+      "step": 205824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.662160966952161e-05,
+      "loss": 4.3808,
+      "step": 206336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.661322372201109e-05,
+      "loss": 4.3932,
+      "step": 206848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6604837774500567e-05,
+      "loss": 4.3884,
+      "step": 207360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6596468205793776e-05,
+      "loss": 4.3886,
+      "step": 207872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6588082258283256e-05,
+      "loss": 4.3789,
+      "step": 208384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6579696310772736e-05,
+      "loss": 4.398,
+      "step": 208896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6571326742065945e-05,
+      "loss": 4.3804,
+      "step": 209408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6562940794555425e-05,
+      "loss": 4.3823,
+      "step": 209920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6554554847044905e-05,
+      "loss": 4.3901,
+      "step": 210432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6546168899534385e-05,
+      "loss": 4.3801,
+      "step": 210944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6537782952023865e-05,
+      "loss": 4.3996,
+      "step": 211456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6529397004513344e-05,
+      "loss": 4.3879,
+      "step": 211968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.652101105700283e-05,
+      "loss": 4.3885,
+      "step": 212480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6512625109492304e-05,
+      "loss": 4.3786,
+      "step": 212992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6504239161981784e-05,
+      "loss": 4.3841,
+      "step": 213504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6495869593275e-05,
+      "loss": 4.3851,
+      "step": 214016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.648748364576447e-05,
+      "loss": 4.3739,
+      "step": 214528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.647909769825395e-05,
+      "loss": 4.3891,
+      "step": 215040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.647071175074343e-05,
+      "loss": 4.3822,
+      "step": 215552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.646232580323291e-05,
+      "loss": 4.3759,
+      "step": 216064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.645393985572239e-05,
+      "loss": 4.378,
+      "step": 216576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.644555390821187e-05,
+      "loss": 4.3759,
+      "step": 217088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.643716796070135e-05,
+      "loss": 4.3674,
+      "step": 217600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642879839199457e-05,
+      "loss": 4.3816,
+      "step": 218112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642041244448405e-05,
+      "loss": 4.3651,
+      "step": 218624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.641202649697353e-05,
+      "loss": 4.3877,
+      "step": 219136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.640364054946301e-05,
+      "loss": 4.3838,
+      "step": 219648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.639527098075622e-05,
+      "loss": 4.3703,
+      "step": 220160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.63868850332457e-05,
+      "loss": 4.3692,
+      "step": 220672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.637851546453891e-05,
+      "loss": 4.3682,
+      "step": 221184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.637012951702839e-05,
+      "loss": 4.3788,
+      "step": 221696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.636174356951787e-05,
+      "loss": 4.3756,
+      "step": 222208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6353374000811076e-05,
+      "loss": 4.3718,
+      "step": 222720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6344988053300556e-05,
+      "loss": 4.3748,
+      "step": 223232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6336602105790036e-05,
+      "loss": 4.3754,
+      "step": 223744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.632823253708325e-05,
+      "loss": 4.3754,
+      "step": 224256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.631984658957273e-05,
+      "loss": 4.3619,
+      "step": 224768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.631146064206221e-05,
+      "loss": 4.3801,
+      "step": 225280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.630307469455169e-05,
+      "loss": 4.3694,
+      "step": 225792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.629468874704117e-05,
+      "loss": 4.3655,
+      "step": 226304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.628630279953065e-05,
+      "loss": 4.367,
+      "step": 226816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.627791685202013e-05,
+      "loss": 4.3685,
+      "step": 227328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626953090450961e-05,
+      "loss": 4.3661,
+      "step": 227840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626114495699909e-05,
+      "loss": 4.3543,
+      "step": 228352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.625275900948857e-05,
+      "loss": 4.3581,
+      "step": 228864
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.344208240509033,
+      "eval_runtime": 292.4539,
+      "eval_samples_per_second": 1304.79,
+      "eval_steps_per_second": 40.776,
+      "step": 228960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.624437306197805e-05,
+      "loss": 4.36,
+      "step": 229376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.623598711446753e-05,
+      "loss": 4.3485,
+      "step": 229888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.622760116695701e-05,
+      "loss": 4.3753,
+      "step": 230400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.621923159825022e-05,
+      "loss": 4.3674,
+      "step": 230912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.62108456507397e-05,
+      "loss": 4.3625,
+      "step": 231424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.620245970322919e-05,
+      "loss": 4.35,
+      "step": 231936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.619407375571867e-05,
+      "loss": 4.3623,
+      "step": 232448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.618568780820814e-05,
+      "loss": 4.3426,
+      "step": 232960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.617730186069762e-05,
+      "loss": 4.3645,
+      "step": 233472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.61689159131871e-05,
+      "loss": 4.3567,
+      "step": 233984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.616052996567658e-05,
+      "loss": 4.3561,
+      "step": 234496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.615216039696979e-05,
+      "loss": 4.3661,
+      "step": 235008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6143790828263005e-05,
+      "loss": 4.3443,
+      "step": 235520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6135404880752485e-05,
+      "loss": 4.3529,
+      "step": 236032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.612701893324196e-05,
+      "loss": 4.3412,
+      "step": 236544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.611863298573144e-05,
+      "loss": 4.3482,
+      "step": 237056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6110247038220925e-05,
+      "loss": 4.3444,
+      "step": 237568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6101861090710405e-05,
+      "loss": 4.3446,
+      "step": 238080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6093491522003614e-05,
+      "loss": 4.3489,
+      "step": 238592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6085105574493094e-05,
+      "loss": 4.3674,
+      "step": 239104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6076719626982574e-05,
+      "loss": 4.3482,
+      "step": 239616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6068333679472054e-05,
+      "loss": 4.3517,
+      "step": 240128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6059947731961534e-05,
+      "loss": 4.3491,
+      "step": 240640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.605157816325474e-05,
+      "loss": 4.3591,
+      "step": 241152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.604319221574422e-05,
+      "loss": 4.3414,
+      "step": 241664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.60348062682337e-05,
+      "loss": 4.3471,
+      "step": 242176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.602642032072318e-05,
+      "loss": 4.3367,
+      "step": 242688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.601803437321266e-05,
+      "loss": 4.3369,
+      "step": 243200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.600964842570214e-05,
+      "loss": 4.3359,
+      "step": 243712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.600127885699536e-05,
+      "loss": 4.3405,
+      "step": 244224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.599289290948484e-05,
+      "loss": 4.3385,
+      "step": 244736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.598450696197432e-05,
+      "loss": 4.3492,
+      "step": 245248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.59761210144638e-05,
+      "loss": 4.342,
+      "step": 245760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.596773506695328e-05,
+      "loss": 4.344,
+      "step": 246272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.595936549824649e-05,
+      "loss": 4.3367,
+      "step": 246784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.595097955073597e-05,
+      "loss": 4.3469,
+      "step": 247296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.594259360322545e-05,
+      "loss": 4.3278,
+      "step": 247808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.593420765571493e-05,
+      "loss": 4.3391,
+      "step": 248320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.592582170820441e-05,
+      "loss": 4.3163,
+      "step": 248832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5917452139497616e-05,
+      "loss": 4.3366,
+      "step": 249344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5909066191987096e-05,
+      "loss": 4.3284,
+      "step": 249856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5900680244476576e-05,
+      "loss": 4.3418,
+      "step": 250368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.589229429696606e-05,
+      "loss": 4.3235,
+      "step": 250880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.588392472825927e-05,
+      "loss": 4.3332,
+      "step": 251392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.587553878074875e-05,
+      "loss": 4.3368,
+      "step": 251904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.586715283323823e-05,
+      "loss": 4.3404,
+      "step": 252416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.585876688572771e-05,
+      "loss": 4.3333,
+      "step": 252928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.585038093821719e-05,
+      "loss": 4.3218,
+      "step": 253440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.584199499070667e-05,
+      "loss": 4.3127,
+      "step": 253952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.583362542199988e-05,
+      "loss": 4.3365,
+      "step": 254464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.582523947448936e-05,
+      "loss": 4.3311,
+      "step": 254976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.581685352697884e-05,
+      "loss": 4.3247,
+      "step": 255488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.580846757946832e-05,
+      "loss": 4.3203,
+      "step": 256000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.580009801076153e-05,
+      "loss": 4.3239,
+      "step": 256512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5791712063251017e-05,
+      "loss": 4.3065,
+      "step": 257024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5783326115740497e-05,
+      "loss": 4.3296,
+      "step": 257536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.577494016822997e-05,
+      "loss": 4.3049,
+      "step": 258048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.576655422071945e-05,
+      "loss": 4.3198,
+      "step": 258560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.575816827320893e-05,
+      "loss": 4.3288,
+      "step": 259072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.574978232569841e-05,
+      "loss": 4.3148,
+      "step": 259584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.574139637818789e-05,
+      "loss": 4.3094,
+      "step": 260096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.57330268094811e-05,
+      "loss": 4.3166,
+      "step": 260608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.572464086197058e-05,
+      "loss": 4.3013,
+      "step": 261120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.571625491446006e-05,
+      "loss": 4.3098,
+      "step": 261632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.570786896694954e-05,
+      "loss": 4.3226,
+      "step": 262144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.569948301943902e-05,
+      "loss": 4.3112,
+      "step": 262656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5691113450732234e-05,
+      "loss": 4.3049,
+      "step": 263168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5682743882025443e-05,
+      "loss": 4.302,
+      "step": 263680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.567437431331866e-05,
+      "loss": 4.3105,
+      "step": 264192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.566600474461187e-05,
+      "loss": 4.317,
+      "step": 264704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.565761879710135e-05,
+      "loss": 4.3175,
+      "step": 265216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.564923284959083e-05,
+      "loss": 4.3103,
+      "step": 265728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.564084690208031e-05,
+      "loss": 4.3191,
+      "step": 266240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.563246095456979e-05,
+      "loss": 4.3205,
+      "step": 266752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.562407500705927e-05,
+      "loss": 4.3064,
+      "step": 267264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.561568905954874e-05,
+      "loss": 4.3048,
+      "step": 267776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.560730311203822e-05,
+      "loss": 4.3122,
+      "step": 268288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.559891716452771e-05,
+      "loss": 4.3077,
+      "step": 268800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.559053121701719e-05,
+      "loss": 4.3058,
+      "step": 269312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.558214526950667e-05,
+      "loss": 4.3046,
+      "step": 269824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.557375932199615e-05,
+      "loss": 4.311,
+      "step": 270336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.556538975328936e-05,
+      "loss": 4.3171,
+      "step": 270848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.555700380577884e-05,
+      "loss": 4.3059,
+      "step": 271360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.554861785826832e-05,
+      "loss": 4.2915,
+      "step": 271872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.55402319107578e-05,
+      "loss": 4.2928,
+      "step": 272384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5531862342051006e-05,
+      "loss": 4.3019,
+      "step": 272896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5523476394540486e-05,
+      "loss": 4.3105,
+      "step": 273408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5515090447029966e-05,
+      "loss": 4.3034,
+      "step": 273920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5506704499519446e-05,
+      "loss": 4.2949,
+      "step": 274432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5498318552008926e-05,
+      "loss": 4.3086,
+      "step": 274944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5489932604498406e-05,
+      "loss": 4.3,
+      "step": 275456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5481546656987886e-05,
+      "loss": 4.2974,
+      "step": 275968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.547319346708483e-05,
+      "loss": 4.2956,
+      "step": 276480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.546480751957431e-05,
+      "loss": 4.3016,
+      "step": 276992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.545642157206379e-05,
+      "loss": 4.3114,
+      "step": 277504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.544803562455327e-05,
+      "loss": 4.292,
+      "step": 278016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.543964967704275e-05,
+      "loss": 4.2843,
+      "step": 278528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.543126372953223e-05,
+      "loss": 4.2989,
+      "step": 279040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.542287778202171e-05,
+      "loss": 4.2953,
+      "step": 279552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.541449183451119e-05,
+      "loss": 4.3083,
+      "step": 280064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.54061222658044e-05,
+      "loss": 4.3007,
+      "step": 280576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.539773631829388e-05,
+      "loss": 4.2882,
+      "step": 281088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.538935037078336e-05,
+      "loss": 4.2839,
+      "step": 281600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.538096442327284e-05,
+      "loss": 4.2984,
+      "step": 282112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5372578475762326e-05,
+      "loss": 4.279,
+      "step": 282624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5364192528251806e-05,
+      "loss": 4.2901,
+      "step": 283136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.535580658074128e-05,
+      "loss": 4.2856,
+      "step": 283648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.534742063323076e-05,
+      "loss": 4.2919,
+      "step": 284160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5339051064523975e-05,
+      "loss": 4.2791,
+      "step": 284672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5330665117013455e-05,
+      "loss": 4.2965,
+      "step": 285184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.532227916950293e-05,
+      "loss": 4.2825,
+      "step": 285696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.531389322199241e-05,
+      "loss": 4.2808,
+      "step": 286208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5305523653285624e-05,
+      "loss": 4.294,
+      "step": 286720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5297137705775104e-05,
+      "loss": 4.2807,
+      "step": 287232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.528875175826458e-05,
+      "loss": 4.304,
+      "step": 287744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5280365810754064e-05,
+      "loss": 4.2911,
+      "step": 288256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.527201262085101e-05,
+      "loss": 4.2879,
+      "step": 288768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.526362667334049e-05,
+      "loss": 4.2812,
+      "step": 289280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.525524072582997e-05,
+      "loss": 4.285,
+      "step": 289792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.524685477831945e-05,
+      "loss": 4.2848,
+      "step": 290304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.523848520961266e-05,
+      "loss": 4.2792,
+      "step": 290816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.523009926210214e-05,
+      "loss": 4.2906,
+      "step": 291328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.522171331459162e-05,
+      "loss": 4.2857,
+      "step": 291840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.52133273670811e-05,
+      "loss": 4.2799,
+      "step": 292352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.520494141957057e-05,
+      "loss": 4.2809,
+      "step": 292864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.519655547206005e-05,
+      "loss": 4.2806,
+      "step": 293376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.518816952454953e-05,
+      "loss": 4.2735,
+      "step": 293888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.517978357703902e-05,
+      "loss": 4.2887,
+      "step": 294400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.51713976295285e-05,
+      "loss": 4.269,
+      "step": 294912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.516301168201798e-05,
+      "loss": 4.2897,
+      "step": 295424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5154642113311187e-05,
+      "loss": 4.2831,
+      "step": 295936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5146256165800667e-05,
+      "loss": 4.2839,
+      "step": 296448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5137870218290147e-05,
+      "loss": 4.2732,
+      "step": 296960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5129484270779626e-05,
+      "loss": 4.2759,
+      "step": 297472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5121114702072836e-05,
+      "loss": 4.2851,
+      "step": 297984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5112728754562316e-05,
+      "loss": 4.2836,
+      "step": 298496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5104342807051795e-05,
+      "loss": 4.2739,
+      "step": 299008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5095956859541275e-05,
+      "loss": 4.2834,
+      "step": 299520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5087570912030755e-05,
+      "loss": 4.2801,
+      "step": 300032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5079184964520235e-05,
+      "loss": 4.2844,
+      "step": 300544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5070799017009715e-05,
+      "loss": 4.2721,
+      "step": 301056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5062413069499195e-05,
+      "loss": 4.2879,
+      "step": 301568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.505404350079241e-05,
+      "loss": 4.2806,
+      "step": 302080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.504565755328189e-05,
+      "loss": 4.2733,
+      "step": 302592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.503727160577137e-05,
+      "loss": 4.2737,
+      "step": 303104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.502888565826085e-05,
+      "loss": 4.2803,
+      "step": 303616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.502051608955406e-05,
+      "loss": 4.2717,
+      "step": 304128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.501213014204354e-05,
+      "loss": 4.2602,
+      "step": 304640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.500374419453302e-05,
+      "loss": 4.2765,
+      "step": 305152
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.262001037597656,
+      "eval_runtime": 291.2224,
+      "eval_samples_per_second": 1310.308,
+      "eval_steps_per_second": 40.948,
+      "step": 305280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.49953582470225e-05,
+      "loss": 4.2638,
+      "step": 305664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.498697229951198e-05,
+      "loss": 4.2596,
+      "step": 306176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.497858635200146e-05,
+      "loss": 4.282,
+      "step": 306688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.497020040449093e-05,
+      "loss": 4.2783,
+      "step": 307200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.496181445698042e-05,
+      "loss": 4.2746,
+      "step": 307712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.49534285094699e-05,
+      "loss": 4.2667,
+      "step": 308224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.494504256195938e-05,
+      "loss": 4.271,
+      "step": 308736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.493665661444886e-05,
+      "loss": 4.252,
+      "step": 309248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.492827066693834e-05,
+      "loss": 4.2762,
+      "step": 309760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.491988471942782e-05,
+      "loss": 4.2717,
+      "step": 310272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.49114987719173e-05,
+      "loss": 4.2679,
+      "step": 310784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.490311282440678e-05,
+      "loss": 4.2771,
+      "step": 311296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.489474325569999e-05,
+      "loss": 4.2592,
+      "step": 311808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.488635730818947e-05,
+      "loss": 4.2656,
+      "step": 312320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.487797136067895e-05,
+      "loss": 4.2566,
+      "step": 312832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.486958541316843e-05,
+      "loss": 4.2578,
+      "step": 313344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.486119946565791e-05,
+      "loss": 4.2598,
+      "step": 313856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.485281351814739e-05,
+      "loss": 4.2597,
+      "step": 314368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4844443949440604e-05,
+      "loss": 4.2626,
+      "step": 314880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4836058001930084e-05,
+      "loss": 4.2788,
+      "step": 315392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4827672054419564e-05,
+      "loss": 4.2659,
+      "step": 315904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.481930248571277e-05,
+      "loss": 4.2699,
+      "step": 316416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.481091653820225e-05,
+      "loss": 4.2617,
+      "step": 316928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.480254696949546e-05,
+      "loss": 4.2693,
+      "step": 317440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.479416102198494e-05,
+      "loss": 4.2574,
+      "step": 317952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.478577507447442e-05,
+      "loss": 4.2619,
+      "step": 318464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.47773891269639e-05,
+      "loss": 4.2565,
+      "step": 318976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.476900317945338e-05,
+      "loss": 4.2492,
+      "step": 319488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.476061723194286e-05,
+      "loss": 4.251,
+      "step": 320000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.475223128443234e-05,
+      "loss": 4.2576,
+      "step": 320512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.474384533692182e-05,
+      "loss": 4.2565,
+      "step": 321024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.47354593894113e-05,
+      "loss": 4.2652,
+      "step": 321536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.472707344190078e-05,
+      "loss": 4.2607,
+      "step": 322048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.471868749439026e-05,
+      "loss": 4.2613,
+      "step": 322560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.471030154687974e-05,
+      "loss": 4.2533,
+      "step": 323072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.470193197817295e-05,
+      "loss": 4.2672,
+      "step": 323584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.469354603066243e-05,
+      "loss": 4.2475,
+      "step": 324096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.468516008315191e-05,
+      "loss": 4.2533,
+      "step": 324608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.467677413564139e-05,
+      "loss": 4.2383,
+      "step": 325120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.466838818813087e-05,
+      "loss": 4.2531,
+      "step": 325632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.466000224062035e-05,
+      "loss": 4.2489,
+      "step": 326144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.465161629310983e-05,
+      "loss": 4.2597,
+      "step": 326656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.464323034559931e-05,
+      "loss": 4.2436,
+      "step": 327168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4634860776892526e-05,
+      "loss": 4.2529,
+      "step": 327680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4626474829382006e-05,
+      "loss": 4.2567,
+      "step": 328192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4618088881871486e-05,
+      "loss": 4.2583,
+      "step": 328704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4609702934360966e-05,
+      "loss": 4.2539,
+      "step": 329216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4601316986850446e-05,
+      "loss": 4.2432,
+      "step": 329728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4592931039339926e-05,
+      "loss": 4.2319,
+      "step": 330240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4584545091829406e-05,
+      "loss": 4.2592,
+      "step": 330752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4576159144318886e-05,
+      "loss": 4.2513,
+      "step": 331264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4567773196808366e-05,
+      "loss": 4.2462,
+      "step": 331776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455938724929784e-05,
+      "loss": 4.2458,
+      "step": 332288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455100130178732e-05,
+      "loss": 4.2435,
+      "step": 332800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.45426153542768e-05,
+      "loss": 4.2289,
+      "step": 333312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.453424578557001e-05,
+      "loss": 4.2443,
+      "step": 333824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.452585983805949e-05,
+      "loss": 4.23,
+      "step": 334336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4517490269352704e-05,
+      "loss": 4.24,
+      "step": 334848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.450910432184219e-05,
+      "loss": 4.2532,
+      "step": 335360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4500718374331664e-05,
+      "loss": 4.2361,
+      "step": 335872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.449234880562488e-05,
+      "loss": 4.2311,
+      "step": 336384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.448396285811436e-05,
+      "loss": 4.2444,
+      "step": 336896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.447557691060384e-05,
+      "loss": 4.223,
+      "step": 337408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.446719096309331e-05,
+      "loss": 4.2342,
+      "step": 337920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.445880501558279e-05,
+      "loss": 4.2423,
+      "step": 338432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.445043544687601e-05,
+      "loss": 4.2382,
+      "step": 338944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444204949936548e-05,
+      "loss": 4.2284,
+      "step": 339456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.443366355185496e-05,
+      "loss": 4.2296,
+      "step": 339968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.442527760434444e-05,
+      "loss": 4.2311,
+      "step": 340480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.441689165683393e-05,
+      "loss": 4.241,
+      "step": 340992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.440850570932341e-05,
+      "loss": 4.2409,
+      "step": 341504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.440011976181289e-05,
+      "loss": 4.2354,
+      "step": 342016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.439173381430237e-05,
+      "loss": 4.2422,
+      "step": 342528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.438336424559558e-05,
+      "loss": 4.2464,
+      "step": 343040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.437497829808506e-05,
+      "loss": 4.233,
+      "step": 343552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.436659235057454e-05,
+      "loss": 4.2257,
+      "step": 344064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.435820640306402e-05,
+      "loss": 4.2433,
+      "step": 344576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.43498204555535e-05,
+      "loss": 4.2348,
+      "step": 345088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.434143450804298e-05,
+      "loss": 4.2269,
+      "step": 345600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.433304856053246e-05,
+      "loss": 4.2308,
+      "step": 346112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4324678991825666e-05,
+      "loss": 4.2374,
+      "step": 346624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4316293044315146e-05,
+      "loss": 4.2441,
+      "step": 347136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4307907096804626e-05,
+      "loss": 4.2283,
+      "step": 347648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.429952114929411e-05,
+      "loss": 4.2178,
+      "step": 348160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.429113520178359e-05,
+      "loss": 4.2254,
+      "step": 348672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.42827656330768e-05,
+      "loss": 4.2213,
+      "step": 349184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.427437968556628e-05,
+      "loss": 4.2357,
+      "step": 349696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.426599373805576e-05,
+      "loss": 4.2317,
+      "step": 350208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.425760779054524e-05,
+      "loss": 4.219,
+      "step": 350720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.424922184303472e-05,
+      "loss": 4.2349,
+      "step": 351232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.424085227432793e-05,
+      "loss": 4.2267,
+      "step": 351744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.423246632681741e-05,
+      "loss": 4.2281,
+      "step": 352256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.422408037930689e-05,
+      "loss": 4.2209,
+      "step": 352768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.421569443179637e-05,
+      "loss": 4.2286,
+      "step": 353280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.420732486308958e-05,
+      "loss": 4.2391,
+      "step": 353792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.419893891557907e-05,
+      "loss": 4.2268,
+      "step": 354304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.419055296806855e-05,
+      "loss": 4.2078,
+      "step": 354816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.418216702055803e-05,
+      "loss": 4.224,
+      "step": 355328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4173797451851236e-05,
+      "loss": 4.2254,
+      "step": 355840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4165411504340716e-05,
+      "loss": 4.2316,
+      "step": 356352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4157025556830196e-05,
+      "loss": 4.2335,
+      "step": 356864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4148639609319676e-05,
+      "loss": 4.2202,
+      "step": 357376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.414025366180915e-05,
+      "loss": 4.2141,
+      "step": 357888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.413186771429863e-05,
+      "loss": 4.2263,
+      "step": 358400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4123498145591845e-05,
+      "loss": 4.2045,
+      "step": 358912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.411511219808132e-05,
+      "loss": 4.2208,
+      "step": 359424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4106726250570805e-05,
+      "loss": 4.2141,
+      "step": 359936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4098340303060284e-05,
+      "loss": 4.2219,
+      "step": 360448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4089954355549764e-05,
+      "loss": 4.2109,
+      "step": 360960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4081568408039244e-05,
+      "loss": 4.2266,
+      "step": 361472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4073182460528724e-05,
+      "loss": 4.2113,
+      "step": 361984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4064796513018204e-05,
+      "loss": 4.2108,
+      "step": 362496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4056426944311413e-05,
+      "loss": 4.225,
+      "step": 363008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.404804099680089e-05,
+      "loss": 4.2152,
+      "step": 363520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.403965504929037e-05,
+      "loss": 4.2341,
+      "step": 364032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.403126910177985e-05,
+      "loss": 4.2224,
+      "step": 364544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.402288315426933e-05,
+      "loss": 4.2161,
+      "step": 365056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.401449720675881e-05,
+      "loss": 4.2217,
+      "step": 365568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.400612763805202e-05,
+      "loss": 4.2099,
+      "step": 366080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.39977416905415e-05,
+      "loss": 4.2207,
+      "step": 366592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.398935574303098e-05,
+      "loss": 4.2103,
+      "step": 367104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.398096979552047e-05,
+      "loss": 4.2247,
+      "step": 367616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.397258384800995e-05,
+      "loss": 4.2127,
+      "step": 368128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.396419790049943e-05,
+      "loss": 4.211,
+      "step": 368640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.395581195298891e-05,
+      "loss": 4.2127,
+      "step": 369152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.394744238428212e-05,
+      "loss": 4.2162,
+      "step": 369664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.39390564367716e-05,
+      "loss": 4.2056,
+      "step": 370176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.393067048926108e-05,
+      "loss": 4.2204,
+      "step": 370688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.392228454175056e-05,
+      "loss": 4.2032,
+      "step": 371200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.391389859424003e-05,
+      "loss": 4.2233,
+      "step": 371712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.390551264672951e-05,
+      "loss": 4.2155,
+      "step": 372224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.389712669921899e-05,
+      "loss": 4.216,
+      "step": 372736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.388874075170847e-05,
+      "loss": 4.2077,
+      "step": 373248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.388035480419795e-05,
+      "loss": 4.2088,
+      "step": 373760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.387198523549117e-05,
+      "loss": 4.2219,
+      "step": 374272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3863599287980647e-05,
+      "loss": 4.2163,
+      "step": 374784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3855213340470127e-05,
+      "loss": 4.2066,
+      "step": 375296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3846827392959606e-05,
+      "loss": 4.2188,
+      "step": 375808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3838457824252816e-05,
+      "loss": 4.213,
+      "step": 376320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3830071876742296e-05,
+      "loss": 4.2216,
+      "step": 376832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3821702308035505e-05,
+      "loss": 4.2075,
+      "step": 377344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.381333273932872e-05,
+      "loss": 4.2204,
+      "step": 377856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.38049467918182e-05,
+      "loss": 4.2125,
+      "step": 378368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.379656084430768e-05,
+      "loss": 4.209,
+      "step": 378880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.378817489679716e-05,
+      "loss": 4.2108,
+      "step": 379392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.377978894928664e-05,
+      "loss": 4.2126,
+      "step": 379904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3771419380579856e-05,
+      "loss": 4.2106,
+      "step": 380416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3763033433069336e-05,
+      "loss": 4.1961,
+      "step": 380928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.375464748555881e-05,
+      "loss": 4.2091,
+      "step": 381440
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.20511531829834,
+      "eval_runtime": 292.8197,
+      "eval_samples_per_second": 1303.16,
+      "eval_steps_per_second": 40.725,
+      "step": 381600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.374626153804829e-05,
+      "loss": 4.2035,
+      "step": 381952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.373787559053777e-05,
+      "loss": 4.1945,
+      "step": 382464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.372948964302725e-05,
+      "loss": 4.2167,
+      "step": 382976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.372110369551673e-05,
+      "loss": 4.2152,
+      "step": 383488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.371273412680994e-05,
+      "loss": 4.2075,
+      "step": 384000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.370434817929942e-05,
+      "loss": 4.2036,
+      "step": 384512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.36959622317889e-05,
+      "loss": 4.2063,
+      "step": 385024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.368757628427838e-05,
+      "loss": 4.1918,
+      "step": 385536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.367919033676786e-05,
+      "loss": 4.2092,
+      "step": 386048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3670804389257345e-05,
+      "loss": 4.2118,
+      "step": 386560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3662418441746825e-05,
+      "loss": 4.2044,
+      "step": 387072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3654032494236305e-05,
+      "loss": 4.2122,
+      "step": 387584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3645662925529514e-05,
+      "loss": 4.1981,
+      "step": 388096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3637276978018994e-05,
+      "loss": 4.2033,
+      "step": 388608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3628891030508474e-05,
+      "loss": 4.194,
+      "step": 389120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3620505082997954e-05,
+      "loss": 4.1945,
+      "step": 389632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3612119135487434e-05,
+      "loss": 4.1973,
+      "step": 390144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3603733187976914e-05,
+      "loss": 4.1959,
+      "step": 390656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3595347240466394e-05,
+      "loss": 4.1999,
+      "step": 391168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.358696129295587e-05,
+      "loss": 4.2127,
+      "step": 391680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.357857534544535e-05,
+      "loss": 4.2091,
+      "step": 392192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.357020577673856e-05,
+      "loss": 4.2078,
+      "step": 392704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.356181982922804e-05,
+      "loss": 4.1972,
+      "step": 393216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.355343388171752e-05,
+      "loss": 4.2107,
+      "step": 393728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3545047934207e-05,
+      "loss": 4.1933,
+      "step": 394240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.353667836550022e-05,
+      "loss": 4.204,
+      "step": 394752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.352829241798969e-05,
+      "loss": 4.1944,
+      "step": 395264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.351992284928291e-05,
+      "loss": 4.1866,
+      "step": 395776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.351153690177239e-05,
+      "loss": 4.1936,
+      "step": 396288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.350315095426187e-05,
+      "loss": 4.193,
+      "step": 396800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.349476500675134e-05,
+      "loss": 4.1985,
+      "step": 397312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.348637905924082e-05,
+      "loss": 4.2006,
+      "step": 397824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.34779931117303e-05,
+      "loss": 4.2048,
+      "step": 398336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.346960716421978e-05,
+      "loss": 4.2028,
+      "step": 398848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.346122121670926e-05,
+      "loss": 4.1926,
+      "step": 399360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3452851648002476e-05,
+      "loss": 4.2028,
+      "step": 399872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3444465700491956e-05,
+      "loss": 4.1918,
+      "step": 400384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3436079752981436e-05,
+      "loss": 4.1874,
+      "step": 400896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3427693805470916e-05,
+      "loss": 4.1827,
+      "step": 401408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3419307857960396e-05,
+      "loss": 4.1908,
+      "step": 401920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3410921910449876e-05,
+      "loss": 4.1941,
+      "step": 402432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3402535962939356e-05,
+      "loss": 4.1983,
+      "step": 402944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3394150015428836e-05,
+      "loss": 4.1868,
+      "step": 403456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3385780446722045e-05,
+      "loss": 4.1911,
+      "step": 403968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3377394499211525e-05,
+      "loss": 4.1976,
+      "step": 404480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3369008551701005e-05,
+      "loss": 4.1966,
+      "step": 404992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3360622604190485e-05,
+      "loss": 4.1979,
+      "step": 405504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.33522530354837e-05,
+      "loss": 4.1841,
+      "step": 406016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.334386708797318e-05,
+      "loss": 4.1701,
+      "step": 406528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.333548114046266e-05,
+      "loss": 4.2019,
+      "step": 407040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.332709519295214e-05,
+      "loss": 4.1932,
+      "step": 407552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.331870924544162e-05,
+      "loss": 4.1893,
+      "step": 408064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.331033967673483e-05,
+      "loss": 4.1888,
+      "step": 408576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.330195372922431e-05,
+      "loss": 4.1799,
+      "step": 409088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.329356778171379e-05,
+      "loss": 4.1748,
+      "step": 409600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3285198213007e-05,
+      "loss": 4.1848,
+      "step": 410112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.327681226549648e-05,
+      "loss": 4.1724,
+      "step": 410624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.326842631798596e-05,
+      "loss": 4.1823,
+      "step": 411136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.326004037047544e-05,
+      "loss": 4.1946,
+      "step": 411648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.325165442296492e-05,
+      "loss": 4.1782,
+      "step": 412160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.32432684754544e-05,
+      "loss": 4.1736,
+      "step": 412672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3234882527943885e-05,
+      "loss": 4.1889,
+      "step": 413184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.322649658043336e-05,
+      "loss": 4.1682,
+      "step": 413696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.321811063292284e-05,
+      "loss": 4.1733,
+      "step": 414208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3209741064216054e-05,
+      "loss": 4.1866,
+      "step": 414720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.320135511670553e-05,
+      "loss": 4.1825,
+      "step": 415232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.319296916919501e-05,
+      "loss": 4.1717,
+      "step": 415744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.318458322168449e-05,
+      "loss": 4.1731,
+      "step": 416256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.317619727417397e-05,
+      "loss": 4.1727,
+      "step": 416768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.316781132666345e-05,
+      "loss": 4.1827,
+      "step": 417280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.315942537915293e-05,
+      "loss": 4.186,
+      "step": 417792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.315103943164241e-05,
+      "loss": 4.178,
+      "step": 418304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.314265348413189e-05,
+      "loss": 4.1844,
+      "step": 418816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.313426753662137e-05,
+      "loss": 4.1899,
+      "step": 419328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3125881589110854e-05,
+      "loss": 4.1852,
+      "step": 419840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.311751202040406e-05,
+      "loss": 4.1664,
+      "step": 420352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.310912607289354e-05,
+      "loss": 4.1847,
+      "step": 420864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.310075650418675e-05,
+      "loss": 4.1831,
+      "step": 421376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.309237055667623e-05,
+      "loss": 4.1667,
+      "step": 421888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.308398460916571e-05,
+      "loss": 4.1818,
+      "step": 422400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.307559866165519e-05,
+      "loss": 4.1764,
+      "step": 422912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.30672290929484e-05,
+      "loss": 4.1903,
+      "step": 423424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.305884314543788e-05,
+      "loss": 4.177,
+      "step": 423936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.305045719792736e-05,
+      "loss": 4.1571,
+      "step": 424448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.304207125041684e-05,
+      "loss": 4.175,
+      "step": 424960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.303368530290632e-05,
+      "loss": 4.1702,
+      "step": 425472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.30252993553958e-05,
+      "loss": 4.1778,
+      "step": 425984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3016929786689016e-05,
+      "loss": 4.1761,
+      "step": 426496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3008543839178496e-05,
+      "loss": 4.1683,
+      "step": 427008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3000157891667976e-05,
+      "loss": 4.1767,
+      "step": 427520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2991771944157456e-05,
+      "loss": 4.1746,
+      "step": 428032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2983385996646936e-05,
+      "loss": 4.1678,
+      "step": 428544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2975000049136416e-05,
+      "loss": 4.1713,
+      "step": 429056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2966630480429625e-05,
+      "loss": 4.1776,
+      "step": 429568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2958244532919105e-05,
+      "loss": 4.1864,
+      "step": 430080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2949858585408585e-05,
+      "loss": 4.1704,
+      "step": 430592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2941489016701794e-05,
+      "loss": 4.1582,
+      "step": 431104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2933103069191274e-05,
+      "loss": 4.1708,
+      "step": 431616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2924717121680754e-05,
+      "loss": 4.1708,
+      "step": 432128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.291633117417024e-05,
+      "loss": 4.1803,
+      "step": 432640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2907945226659714e-05,
+      "loss": 4.1795,
+      "step": 433152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2899559279149194e-05,
+      "loss": 4.1614,
+      "step": 433664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2891173331638674e-05,
+      "loss": 4.1627,
+      "step": 434176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2882787384128154e-05,
+      "loss": 4.1747,
+      "step": 434688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.287441781542136e-05,
+      "loss": 4.153,
+      "step": 435200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.286603186791084e-05,
+      "loss": 4.1693,
+      "step": 435712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.285764592040032e-05,
+      "loss": 4.1569,
+      "step": 436224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.28492599728898e-05,
+      "loss": 4.1693,
+      "step": 436736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.284087402537928e-05,
+      "loss": 4.1589,
+      "step": 437248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.283248807786876e-05,
+      "loss": 4.1729,
+      "step": 437760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.282410213035824e-05,
+      "loss": 4.1652,
+      "step": 438272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.281573256165146e-05,
+      "loss": 4.1557,
+      "step": 438784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.280734661414094e-05,
+      "loss": 4.172,
+      "step": 439296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.279896066663042e-05,
+      "loss": 4.1598,
+      "step": 439808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.27905747191199e-05,
+      "loss": 4.181,
+      "step": 440320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.278218877160938e-05,
+      "loss": 4.1745,
+      "step": 440832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.277380282409886e-05,
+      "loss": 4.1621,
+      "step": 441344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.276543325539207e-05,
+      "loss": 4.1719,
+      "step": 441856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.275704730788155e-05,
+      "loss": 4.157,
+      "step": 442368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.274866136037103e-05,
+      "loss": 4.1641,
+      "step": 442880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.274027541286051e-05,
+      "loss": 4.1621,
+      "step": 443392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.273188946534999e-05,
+      "loss": 4.1718,
+      "step": 443904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.272350351783947e-05,
+      "loss": 4.1628,
+      "step": 444416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.271511757032895e-05,
+      "loss": 4.1593,
+      "step": 444928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.270673162281843e-05,
+      "loss": 4.1647,
+      "step": 445440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.269836205411164e-05,
+      "loss": 4.1643,
+      "step": 445952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.268997610660112e-05,
+      "loss": 4.1528,
+      "step": 446464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.26815901590906e-05,
+      "loss": 4.1704,
+      "step": 446976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2673204211580076e-05,
+      "loss": 4.1496,
+      "step": 447488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2664818264069556e-05,
+      "loss": 4.1778,
+      "step": 448000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2656432316559036e-05,
+      "loss": 4.1583,
+      "step": 448512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2648046369048516e-05,
+      "loss": 4.1688,
+      "step": 449024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2639660421537996e-05,
+      "loss": 4.1556,
+      "step": 449536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2631290852831205e-05,
+      "loss": 4.1598,
+      "step": 450048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2622904905320685e-05,
+      "loss": 4.1691,
+      "step": 450560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.26145353366139e-05,
+      "loss": 4.1645,
+      "step": 451072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.260614938910338e-05,
+      "loss": 4.1577,
+      "step": 451584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.259776344159286e-05,
+      "loss": 4.1703,
+      "step": 452096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.258937749408234e-05,
+      "loss": 4.1652,
+      "step": 452608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.258099154657182e-05,
+      "loss": 4.1703,
+      "step": 453120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.25726055990613e-05,
+      "loss": 4.1557,
+      "step": 453632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.256423603035451e-05,
+      "loss": 4.1684,
+      "step": 454144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.255585008284399e-05,
+      "loss": 4.1631,
+      "step": 454656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.254746413533347e-05,
+      "loss": 4.1603,
+      "step": 455168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.253907818782295e-05,
+      "loss": 4.1611,
+      "step": 455680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.253070861911616e-05,
+      "loss": 4.1615,
+      "step": 456192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.252232267160564e-05,
+      "loss": 4.1682,
+      "step": 456704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.251393672409512e-05,
+      "loss": 4.1391,
+      "step": 457216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.25055507765846e-05,
+      "loss": 4.165,
+      "step": 457728
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.1645636558532715,
+      "eval_runtime": 293.7363,
+      "eval_samples_per_second": 1299.094,
+      "eval_steps_per_second": 40.598,
+      "step": 457920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2497164829074085e-05,
+      "loss": 4.1537,
+      "step": 458240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2488778881563565e-05,
+      "loss": 4.1465,
+      "step": 458752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2480392934053045e-05,
+      "loss": 4.1631,
+      "step": 459264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2472006986542525e-05,
+      "loss": 4.1691,
+      "step": 459776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2463621039032005e-05,
+      "loss": 4.1577,
+      "step": 460288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2455235091521485e-05,
+      "loss": 4.1537,
+      "step": 460800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2446849144010965e-05,
+      "loss": 4.1576,
+      "step": 461312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.243846319650044e-05,
+      "loss": 4.1457,
+      "step": 461824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.243007724898992e-05,
+      "loss": 4.1602,
+      "step": 462336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.24216913014794e-05,
+      "loss": 4.1616,
+      "step": 462848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.241330535396888e-05,
+      "loss": 4.1615,
+      "step": 463360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.240491940645836e-05,
+      "loss": 4.1609,
+      "step": 463872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.239654983775157e-05,
+      "loss": 4.155,
+      "step": 464384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.238816389024105e-05,
+      "loss": 4.1523,
+      "step": 464896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2379777942730534e-05,
+      "loss": 4.1469,
+      "step": 465408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2371391995220014e-05,
+      "loss": 4.142,
+      "step": 465920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2363006047709494e-05,
+      "loss": 4.1513,
+      "step": 466432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2354620100198974e-05,
+      "loss": 4.1474,
+      "step": 466944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2346234152688454e-05,
+      "loss": 4.1528,
+      "step": 467456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2337848205177934e-05,
+      "loss": 4.1646,
+      "step": 467968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.232947863647114e-05,
+      "loss": 4.1648,
+      "step": 468480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.232109268896062e-05,
+      "loss": 4.158,
+      "step": 468992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.23127067414501e-05,
+      "loss": 4.1506,
+      "step": 469504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.230432079393958e-05,
+      "loss": 4.1634,
+      "step": 470016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.229593484642906e-05,
+      "loss": 4.146,
+      "step": 470528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.228754889891854e-05,
+      "loss": 4.1602,
+      "step": 471040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.227916295140802e-05,
+      "loss": 4.1483,
+      "step": 471552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.22707770038975e-05,
+      "loss": 4.1375,
+      "step": 472064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.226239105638698e-05,
+      "loss": 4.1499,
+      "step": 472576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.225400510887646e-05,
+      "loss": 4.1469,
+      "step": 473088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.224563554016968e-05,
+      "loss": 4.1491,
+      "step": 473600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.223724959265915e-05,
+      "loss": 4.1567,
+      "step": 474112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.222886364514863e-05,
+      "loss": 4.1548,
+      "step": 474624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.222047769763811e-05,
+      "loss": 4.1551,
+      "step": 475136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.221210812893133e-05,
+      "loss": 4.1448,
+      "step": 475648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.22037221814208e-05,
+      "loss": 4.1626,
+      "step": 476160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.219533623391028e-05,
+      "loss": 4.1445,
+      "step": 476672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.218695028639976e-05,
+      "loss": 4.1373,
+      "step": 477184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.217856433888924e-05,
+      "loss": 4.1435,
+      "step": 477696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2170194770182456e-05,
+      "loss": 4.1408,
+      "step": 478208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2161808822671936e-05,
+      "loss": 4.1473,
+      "step": 478720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2153422875161416e-05,
+      "loss": 4.1574,
+      "step": 479232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2145036927650896e-05,
+      "loss": 4.1386,
+      "step": 479744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2136667358944105e-05,
+      "loss": 4.1454,
+      "step": 480256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2128281411433585e-05,
+      "loss": 4.1537,
+      "step": 480768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2119895463923065e-05,
+      "loss": 4.1547,
+      "step": 481280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2111509516412545e-05,
+      "loss": 4.1534,
+      "step": 481792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2103123568902025e-05,
+      "loss": 4.141,
+      "step": 482304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2094737621391505e-05,
+      "loss": 4.1165,
+      "step": 482816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2086351673880985e-05,
+      "loss": 4.16,
+      "step": 483328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2077965726370465e-05,
+      "loss": 4.1494,
+      "step": 483840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2069596157663674e-05,
+      "loss": 4.1488,
+      "step": 484352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2061210210153154e-05,
+      "loss": 4.1409,
+      "step": 484864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2052824262642634e-05,
+      "loss": 4.1354,
+      "step": 485376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.204443831513212e-05,
+      "loss": 4.131,
+      "step": 485888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.203606874642533e-05,
+      "loss": 4.1373,
+      "step": 486400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.202768279891481e-05,
+      "loss": 4.1326,
+      "step": 486912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.201929685140429e-05,
+      "loss": 4.1383,
+      "step": 487424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.201091090389377e-05,
+      "loss": 4.1521,
+      "step": 487936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.200252495638325e-05,
+      "loss": 4.1309,
+      "step": 488448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.199413900887273e-05,
+      "loss": 4.1286,
+      "step": 488960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.198575306136221e-05,
+      "loss": 4.1465,
+      "step": 489472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.197738349265542e-05,
+      "loss": 4.1296,
+      "step": 489984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.19689975451449e-05,
+      "loss": 4.1294,
+      "step": 490496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.196061159763438e-05,
+      "loss": 4.1379,
+      "step": 491008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.195222565012386e-05,
+      "loss": 4.1382,
+      "step": 491520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.194383970261334e-05,
+      "loss": 4.1268,
+      "step": 492032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.193545375510282e-05,
+      "loss": 4.1282,
+      "step": 492544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.19270678075923e-05,
+      "loss": 4.1268,
+      "step": 493056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.191868186008178e-05,
+      "loss": 4.1437,
+      "step": 493568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.191031229137499e-05,
+      "loss": 4.14,
+      "step": 494080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.190192634386447e-05,
+      "loss": 4.1365,
+      "step": 494592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.189355677515768e-05,
+      "loss": 4.1378,
+      "step": 495104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.188517082764716e-05,
+      "loss": 4.1463,
+      "step": 495616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1876784880136636e-05,
+      "loss": 4.1441,
+      "step": 496128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1868398932626116e-05,
+      "loss": 4.1269,
+      "step": 496640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1860012985115596e-05,
+      "loss": 4.1383,
+      "step": 497152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.185164341640881e-05,
+      "loss": 4.1445,
+      "step": 497664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.184325746889829e-05,
+      "loss": 4.1232,
+      "step": 498176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.183487152138777e-05,
+      "loss": 4.138,
+      "step": 498688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.182648557387725e-05,
+      "loss": 4.1361,
+      "step": 499200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.181811600517046e-05,
+      "loss": 4.1437,
+      "step": 499712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.180973005765994e-05,
+      "loss": 4.1349,
+      "step": 500224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.180134411014942e-05,
+      "loss": 4.1166,
+      "step": 500736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.17929581626389e-05,
+      "loss": 4.1287,
+      "step": 501248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.178457221512838e-05,
+      "loss": 4.1281,
+      "step": 501760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.177620264642159e-05,
+      "loss": 4.1394,
+      "step": 502272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.176781669891107e-05,
+      "loss": 4.1296,
+      "step": 502784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.175943075140055e-05,
+      "loss": 4.1277,
+      "step": 503296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.175104480389003e-05,
+      "loss": 4.1328,
+      "step": 503808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.174265885637951e-05,
+      "loss": 4.1368,
+      "step": 504320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1734289287672726e-05,
+      "loss": 4.1234,
+      "step": 504832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1725919718965935e-05,
+      "loss": 4.1273,
+      "step": 505344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1717533771455415e-05,
+      "loss": 4.1376,
+      "step": 505856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1709147823944895e-05,
+      "loss": 4.1414,
+      "step": 506368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1700761876434375e-05,
+      "loss": 4.1294,
+      "step": 506880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1692375928923855e-05,
+      "loss": 4.1186,
+      "step": 507392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1684006360217064e-05,
+      "loss": 4.1284,
+      "step": 507904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1675620412706544e-05,
+      "loss": 4.1271,
+      "step": 508416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1667234465196024e-05,
+      "loss": 4.1386,
+      "step": 508928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1658848517685503e-05,
+      "loss": 4.139,
+      "step": 509440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1650462570174983e-05,
+      "loss": 4.1239,
+      "step": 509952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.164207662266446e-05,
+      "loss": 4.122,
+      "step": 510464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.163369067515395e-05,
+      "loss": 4.1303,
+      "step": 510976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.162530472764343e-05,
+      "loss": 4.1112,
+      "step": 511488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.161691878013291e-05,
+      "loss": 4.1311,
+      "step": 512000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.160854921142612e-05,
+      "loss": 4.1112,
+      "step": 512512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16001632639156e-05,
+      "loss": 4.1302,
+      "step": 513024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.159179369520881e-05,
+      "loss": 4.1195,
+      "step": 513536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.158340774769829e-05,
+      "loss": 4.1293,
+      "step": 514048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.157502180018777e-05,
+      "loss": 4.1235,
+      "step": 514560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.156663585267725e-05,
+      "loss": 4.1213,
+      "step": 515072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.155826628397046e-05,
+      "loss": 4.1292,
+      "step": 515584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.154988033645994e-05,
+      "loss": 4.119,
+      "step": 516096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.154149438894942e-05,
+      "loss": 4.1444,
+      "step": 516608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1533108441438904e-05,
+      "loss": 4.1321,
+      "step": 517120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1524722493928384e-05,
+      "loss": 4.1242,
+      "step": 517632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1516336546417864e-05,
+      "loss": 4.1315,
+      "step": 518144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1507950598907344e-05,
+      "loss": 4.1159,
+      "step": 518656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1499564651396824e-05,
+      "loss": 4.126,
+      "step": 519168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.149121146149376e-05,
+      "loss": 4.1246,
+      "step": 519680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.148282551398324e-05,
+      "loss": 4.1266,
+      "step": 520192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.147443956647272e-05,
+      "loss": 4.1247,
+      "step": 520704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14660536189622e-05,
+      "loss": 4.1211,
+      "step": 521216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.145766767145168e-05,
+      "loss": 4.1249,
+      "step": 521728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.144928172394116e-05,
+      "loss": 4.1231,
+      "step": 522240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.144089577643064e-05,
+      "loss": 4.1166,
+      "step": 522752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.143252620772386e-05,
+      "loss": 4.1293,
+      "step": 523264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.142414026021334e-05,
+      "loss": 4.1092,
+      "step": 523776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.141575431270282e-05,
+      "loss": 4.1358,
+      "step": 524288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14073683651923e-05,
+      "loss": 4.1238,
+      "step": 524800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.139898241768177e-05,
+      "loss": 4.1292,
+      "step": 525312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.139059647017125e-05,
+      "loss": 4.1121,
+      "step": 525824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.138221052266073e-05,
+      "loss": 4.1261,
+      "step": 526336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.137382457515021e-05,
+      "loss": 4.1304,
+      "step": 526848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.136543862763969e-05,
+      "loss": 4.1285,
+      "step": 527360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.135705268012917e-05,
+      "loss": 4.1174,
+      "step": 527872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.134866673261865e-05,
+      "loss": 4.1292,
+      "step": 528384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.134028078510813e-05,
+      "loss": 4.1236,
+      "step": 528896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.133191121640134e-05,
+      "loss": 4.1327,
+      "step": 529408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.132352526889082e-05,
+      "loss": 4.1149,
+      "step": 529920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1315155700184035e-05,
+      "loss": 4.1309,
+      "step": 530432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1306769752673515e-05,
+      "loss": 4.1233,
+      "step": 530944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1298383805162995e-05,
+      "loss": 4.1215,
+      "step": 531456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1289997857652475e-05,
+      "loss": 4.1233,
+      "step": 531968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1281611910141955e-05,
+      "loss": 4.1238,
+      "step": 532480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1273225962631435e-05,
+      "loss": 4.1259,
+      "step": 532992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1264856393924644e-05,
+      "loss": 4.1059,
+      "step": 533504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1256470446414124e-05,
+      "loss": 4.1287,
+      "step": 534016
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.133726596832275,
+      "eval_runtime": 299.6081,
+      "eval_samples_per_second": 1273.634,
+      "eval_steps_per_second": 39.802,
+      "step": 534240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1248084498903604e-05,
+      "loss": 4.1155,
+      "step": 534528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1239698551393084e-05,
+      "loss": 4.1081,
+      "step": 535040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1231312603882564e-05,
+      "loss": 4.1222,
+      "step": 535552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1222926656372044e-05,
+      "loss": 4.1301,
+      "step": 536064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1214540708861524e-05,
+      "loss": 4.1188,
+      "step": 536576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1206154761351004e-05,
+      "loss": 4.1169,
+      "step": 537088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.119780157144795e-05,
+      "loss": 4.1221,
+      "step": 537600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.118941562393743e-05,
+      "loss": 4.107,
+      "step": 538112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.118102967642691e-05,
+      "loss": 4.1224,
+      "step": 538624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.117264372891639e-05,
+      "loss": 4.123,
+      "step": 539136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.116425778140587e-05,
+      "loss": 4.1222,
+      "step": 539648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.115587183389535e-05,
+      "loss": 4.1246,
+      "step": 540160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.114750226518856e-05,
+      "loss": 4.113,
+      "step": 540672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.113911631767804e-05,
+      "loss": 4.1139,
+      "step": 541184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1130746748971247e-05,
+      "loss": 4.1126,
+      "step": 541696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1122360801460727e-05,
+      "loss": 4.1039,
+      "step": 542208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.111397485395021e-05,
+      "loss": 4.1133,
+      "step": 542720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.110558890643969e-05,
+      "loss": 4.109,
+      "step": 543232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.109720295892917e-05,
+      "loss": 4.1154,
+      "step": 543744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.108881701141865e-05,
+      "loss": 4.1244,
+      "step": 544256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.108043106390813e-05,
+      "loss": 4.1293,
+      "step": 544768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1072045116397606e-05,
+      "loss": 4.126,
+      "step": 545280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1063659168887086e-05,
+      "loss": 4.1114,
+      "step": 545792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1055273221376566e-05,
+      "loss": 4.1226,
+      "step": 546304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1046887273866046e-05,
+      "loss": 4.1123,
+      "step": 546816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1038501326355526e-05,
+      "loss": 4.1195,
+      "step": 547328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1030131757648735e-05,
+      "loss": 4.1131,
+      "step": 547840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.102176218894195e-05,
+      "loss": 4.1009,
+      "step": 548352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.101337624143143e-05,
+      "loss": 4.1151,
+      "step": 548864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.100499029392091e-05,
+      "loss": 4.1052,
+      "step": 549376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.099662072521413e-05,
+      "loss": 4.1112,
+      "step": 549888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.098823477770361e-05,
+      "loss": 4.1201,
+      "step": 550400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.097984883019308e-05,
+      "loss": 4.1151,
+      "step": 550912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.097146288268256e-05,
+      "loss": 4.1166,
+      "step": 551424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.096307693517204e-05,
+      "loss": 4.1078,
+      "step": 551936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.095469098766152e-05,
+      "loss": 4.1261,
+      "step": 552448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0946305040151e-05,
+      "loss": 4.1124,
+      "step": 552960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.093791909264048e-05,
+      "loss": 4.0977,
+      "step": 553472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.092953314512996e-05,
+      "loss": 4.1106,
+      "step": 553984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.092114719761944e-05,
+      "loss": 4.1028,
+      "step": 554496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.091276125010892e-05,
+      "loss": 4.1082,
+      "step": 555008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.09043753025984e-05,
+      "loss": 4.1146,
+      "step": 555520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.089598935508788e-05,
+      "loss": 4.1087,
+      "step": 556032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0887619786381095e-05,
+      "loss": 4.1042,
+      "step": 556544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0879233838870575e-05,
+      "loss": 4.1182,
+      "step": 557056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0870864270163784e-05,
+      "loss": 4.1166,
+      "step": 557568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0862478322653264e-05,
+      "loss": 4.1194,
+      "step": 558080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0854092375142744e-05,
+      "loss": 4.1029,
+      "step": 558592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0845706427632224e-05,
+      "loss": 4.0779,
+      "step": 559104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0837320480121704e-05,
+      "loss": 4.1276,
+      "step": 559616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0828934532611184e-05,
+      "loss": 4.1079,
+      "step": 560128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.082056496390439e-05,
+      "loss": 4.1143,
+      "step": 560640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.081217901639387e-05,
+      "loss": 4.1069,
+      "step": 561152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.080379306888335e-05,
+      "loss": 4.0964,
+      "step": 561664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.079540712137283e-05,
+      "loss": 4.0983,
+      "step": 562176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.078702117386231e-05,
+      "loss": 4.1,
+      "step": 562688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.077863522635179e-05,
+      "loss": 4.0957,
+      "step": 563200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.077024927884127e-05,
+      "loss": 4.1012,
+      "step": 563712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.076186333133075e-05,
+      "loss": 4.1212,
+      "step": 564224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.075347738382023e-05,
+      "loss": 4.0932,
+      "step": 564736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.074510781511344e-05,
+      "loss": 4.0908,
+      "step": 565248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.073673824640666e-05,
+      "loss": 4.1138,
+      "step": 565760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.072835229889614e-05,
+      "loss": 4.0902,
+      "step": 566272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.071996635138561e-05,
+      "loss": 4.094,
+      "step": 566784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.071158040387509e-05,
+      "loss": 4.1029,
+      "step": 567296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.070319445636457e-05,
+      "loss": 4.1032,
+      "step": 567808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.069480850885405e-05,
+      "loss": 4.0912,
+      "step": 568320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.068642256134354e-05,
+      "loss": 4.1004,
+      "step": 568832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.067803661383302e-05,
+      "loss": 4.0882,
+      "step": 569344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.066966704512623e-05,
+      "loss": 4.1079,
+      "step": 569856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.066128109761571e-05,
+      "loss": 4.1055,
+      "step": 570368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.065289515010519e-05,
+      "loss": 4.1008,
+      "step": 570880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0644525581398396e-05,
+      "loss": 4.1037,
+      "step": 571392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.063615601269161e-05,
+      "loss": 4.1083,
+      "step": 571904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0627770065181085e-05,
+      "loss": 4.114,
+      "step": 572416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0619384117670565e-05,
+      "loss": 4.0906,
+      "step": 572928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0610998170160045e-05,
+      "loss": 4.1032,
+      "step": 573440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.060262860145326e-05,
+      "loss": 4.1089,
+      "step": 573952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.059424265394274e-05,
+      "loss": 4.086,
+      "step": 574464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.058585670643222e-05,
+      "loss": 4.1064,
+      "step": 574976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.05774707589217e-05,
+      "loss": 4.0983,
+      "step": 575488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.056908481141118e-05,
+      "loss": 4.1103,
+      "step": 576000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.056069886390066e-05,
+      "loss": 4.1032,
+      "step": 576512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.055231291639014e-05,
+      "loss": 4.0837,
+      "step": 577024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.054392696887962e-05,
+      "loss": 4.0884,
+      "step": 577536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.053555740017283e-05,
+      "loss": 4.0956,
+      "step": 578048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.052718783146604e-05,
+      "loss": 4.106,
+      "step": 578560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.051880188395552e-05,
+      "loss": 4.0911,
+      "step": 579072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0510415936445e-05,
+      "loss": 4.0941,
+      "step": 579584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.050202998893448e-05,
+      "loss": 4.0983,
+      "step": 580096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.049364404142396e-05,
+      "loss": 4.1018,
+      "step": 580608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0485258093913445e-05,
+      "loss": 4.0929,
+      "step": 581120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0476872146402925e-05,
+      "loss": 4.0916,
+      "step": 581632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0468486198892405e-05,
+      "loss": 4.1022,
+      "step": 582144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0460116630185614e-05,
+      "loss": 4.1062,
+      "step": 582656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0451730682675094e-05,
+      "loss": 4.0967,
+      "step": 583168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0443344735164574e-05,
+      "loss": 4.0852,
+      "step": 583680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.043497516645778e-05,
+      "loss": 4.0936,
+      "step": 584192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.042658921894726e-05,
+      "loss": 4.0913,
+      "step": 584704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.041820327143674e-05,
+      "loss": 4.1061,
+      "step": 585216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.040981732392622e-05,
+      "loss": 4.0994,
+      "step": 585728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.04014313764157e-05,
+      "loss": 4.0944,
+      "step": 586240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.039304542890518e-05,
+      "loss": 4.0898,
+      "step": 586752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.03846758601984e-05,
+      "loss": 4.0901,
+      "step": 587264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.037628991268788e-05,
+      "loss": 4.0819,
+      "step": 587776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.036790396517736e-05,
+      "loss": 4.0935,
+      "step": 588288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.035951801766684e-05,
+      "loss": 4.0807,
+      "step": 588800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.035113207015632e-05,
+      "loss": 4.099,
+      "step": 589312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.03427461226458e-05,
+      "loss": 4.0887,
+      "step": 589824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.033437655393901e-05,
+      "loss": 4.0939,
+      "step": 590336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.032599060642849e-05,
+      "loss": 4.0877,
+      "step": 590848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.03176210377217e-05,
+      "loss": 4.0868,
+      "step": 591360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0309235090211177e-05,
+      "loss": 4.0975,
+      "step": 591872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0300849142700657e-05,
+      "loss": 4.0869,
+      "step": 592384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0292463195190137e-05,
+      "loss": 4.11,
+      "step": 592896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0284077247679616e-05,
+      "loss": 4.0996,
+      "step": 593408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0275691300169096e-05,
+      "loss": 4.0909,
+      "step": 593920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0267305352658576e-05,
+      "loss": 4.0956,
+      "step": 594432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0258919405148056e-05,
+      "loss": 4.0843,
+      "step": 594944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.025054983644127e-05,
+      "loss": 4.0898,
+      "step": 595456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0242163888930745e-05,
+      "loss": 4.0978,
+      "step": 595968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0233777941420225e-05,
+      "loss": 4.0912,
+      "step": 596480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0225391993909705e-05,
+      "loss": 4.0914,
+      "step": 596992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0217006046399185e-05,
+      "loss": 4.0901,
+      "step": 597504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0208620098888665e-05,
+      "loss": 4.0924,
+      "step": 598016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0200234151378145e-05,
+      "loss": 4.0929,
+      "step": 598528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0191848203867625e-05,
+      "loss": 4.0816,
+      "step": 599040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0183478635160834e-05,
+      "loss": 4.0954,
+      "step": 599552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0175092687650314e-05,
+      "loss": 4.0813,
+      "step": 600064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.01667067401398e-05,
+      "loss": 4.1027,
+      "step": 600576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.015832079262928e-05,
+      "loss": 4.0901,
+      "step": 601088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.014993484511876e-05,
+      "loss": 4.0933,
+      "step": 601600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.014154889760824e-05,
+      "loss": 4.0828,
+      "step": 602112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.013316295009772e-05,
+      "loss": 4.0937,
+      "step": 602624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.012479338139093e-05,
+      "loss": 4.096,
+      "step": 603136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.011640743388041e-05,
+      "loss": 4.0956,
+      "step": 603648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.010802148636989e-05,
+      "loss": 4.0879,
+      "step": 604160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.009963553885937e-05,
+      "loss": 4.0954,
+      "step": 604672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.009124959134885e-05,
+      "loss": 4.0912,
+      "step": 605184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.008288002264206e-05,
+      "loss": 4.1049,
+      "step": 605696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.007449407513154e-05,
+      "loss": 4.0814,
+      "step": 606208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.006610812762102e-05,
+      "loss": 4.0981,
+      "step": 606720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.00577221801105e-05,
+      "loss": 4.0905,
+      "step": 607232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0049352611403714e-05,
+      "loss": 4.0905,
+      "step": 607744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0040966663893194e-05,
+      "loss": 4.0905,
+      "step": 608256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0032580716382674e-05,
+      "loss": 4.0915,
+      "step": 608768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0024194768872154e-05,
+      "loss": 4.0929,
+      "step": 609280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0015825200165363e-05,
+      "loss": 4.0741,
+      "step": 609792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.000745563145857e-05,
+      "loss": 4.0943,
+      "step": 610304
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.10930061340332,
+      "eval_runtime": 294.0693,
+      "eval_samples_per_second": 1297.623,
+      "eval_steps_per_second": 40.552,
+      "step": 610560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.999906968394805e-05,
+      "loss": 4.0787,
+      "step": 610816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.999068373643753e-05,
+      "loss": 4.0787,
+      "step": 611328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.998229778892701e-05,
+      "loss": 4.0884,
+      "step": 611840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.997391184141649e-05,
+      "loss": 4.1001,
+      "step": 612352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.996552589390597e-05,
+      "loss": 4.0909,
+      "step": 612864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.995713994639545e-05,
+      "loss": 4.0824,
+      "step": 613376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.994875399888493e-05,
+      "loss": 4.0958,
+      "step": 613888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.994036805137441e-05,
+      "loss": 4.0724,
+      "step": 614400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.993198210386389e-05,
+      "loss": 4.0875,
+      "step": 614912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.992359615635337e-05,
+      "loss": 4.0954,
+      "step": 615424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.991521020884285e-05,
+      "loss": 4.0879,
+      "step": 615936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.990682426133233e-05,
+      "loss": 4.0903,
+      "step": 616448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.989845469262554e-05,
+      "loss": 4.0914,
+      "step": 616960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.989006874511502e-05,
+      "loss": 4.0774,
+      "step": 617472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.98816827976045e-05,
+      "loss": 4.0765,
+      "step": 617984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.987329685009398e-05,
+      "loss": 4.0748,
+      "step": 618496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.986491090258346e-05,
+      "loss": 4.0806,
+      "step": 619008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.985652495507294e-05,
+      "loss": 4.0824,
+      "step": 619520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.984813900756242e-05,
+      "loss": 4.08,
+      "step": 620032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.983975306005191e-05,
+      "loss": 4.0909,
+      "step": 620544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.983136711254139e-05,
+      "loss": 4.0987,
+      "step": 621056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.982298116503087e-05,
+      "loss": 4.0947,
+      "step": 621568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9814611596324077e-05,
+      "loss": 4.0842,
+      "step": 622080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9806225648813557e-05,
+      "loss": 4.087,
+      "step": 622592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9797839701303036e-05,
+      "loss": 4.0876,
+      "step": 623104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9789453753792516e-05,
+      "loss": 4.0838,
+      "step": 623616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9781067806281996e-05,
+      "loss": 4.0836,
+      "step": 624128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9772698237575205e-05,
+      "loss": 4.0743,
+      "step": 624640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9764312290064685e-05,
+      "loss": 4.0813,
+      "step": 625152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9755926342554165e-05,
+      "loss": 4.0757,
+      "step": 625664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9747540395043645e-05,
+      "loss": 4.0785,
+      "step": 626176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9739154447533125e-05,
+      "loss": 4.0886,
+      "step": 626688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9730768500022605e-05,
+      "loss": 4.0877,
+      "step": 627200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9722382552512085e-05,
+      "loss": 4.0854,
+      "step": 627712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9713996605001565e-05,
+      "loss": 4.0804,
+      "step": 628224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.970565979390224e-05,
+      "loss": 4.0879,
+      "step": 628736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.969727384639172e-05,
+      "loss": 4.0837,
+      "step": 629248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.96888878988812e-05,
+      "loss": 4.07,
+      "step": 629760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.968050195137068e-05,
+      "loss": 4.0789,
+      "step": 630272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.967211600386016e-05,
+      "loss": 4.0712,
+      "step": 630784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.966373005634964e-05,
+      "loss": 4.0767,
+      "step": 631296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.965534410883912e-05,
+      "loss": 4.0851,
+      "step": 631808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.96469581613286e-05,
+      "loss": 4.0758,
+      "step": 632320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.963858859262181e-05,
+      "loss": 4.0783,
+      "step": 632832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9630202645111295e-05,
+      "loss": 4.0901,
+      "step": 633344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.962181669760077e-05,
+      "loss": 4.0831,
+      "step": 633856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.961343075009025e-05,
+      "loss": 4.0861,
+      "step": 634368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.960504480257973e-05,
+      "loss": 4.0786,
+      "step": 634880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9596675233872944e-05,
+      "loss": 4.0512,
+      "step": 635392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.958828928636242e-05,
+      "loss": 4.0928,
+      "step": 635904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.95799033388519e-05,
+      "loss": 4.078,
+      "step": 636416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.957151739134138e-05,
+      "loss": 4.084,
+      "step": 636928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.956313144383086e-05,
+      "loss": 4.0773,
+      "step": 637440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9554761875124066e-05,
+      "loss": 4.0675,
+      "step": 637952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9546375927613546e-05,
+      "loss": 4.0652,
+      "step": 638464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.953800635890676e-05,
+      "loss": 4.0721,
+      "step": 638976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.952962041139624e-05,
+      "loss": 4.0664,
+      "step": 639488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.952123446388572e-05,
+      "loss": 4.0676,
+      "step": 640000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.95128485163752e-05,
+      "loss": 4.0894,
+      "step": 640512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.950446256886468e-05,
+      "loss": 4.0683,
+      "step": 641024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.949607662135416e-05,
+      "loss": 4.0616,
+      "step": 641536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.948769067384364e-05,
+      "loss": 4.0861,
+      "step": 642048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.947930472633312e-05,
+      "loss": 4.057,
+      "step": 642560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.94709187788226e-05,
+      "loss": 4.0636,
+      "step": 643072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.946253283131208e-05,
+      "loss": 4.0784,
+      "step": 643584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.945414688380156e-05,
+      "loss": 4.0756,
+      "step": 644096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.944577731509477e-05,
+      "loss": 4.0621,
+      "step": 644608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.943739136758425e-05,
+      "loss": 4.0679,
+      "step": 645120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.942900542007373e-05,
+      "loss": 4.0604,
+      "step": 645632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.942061947256322e-05,
+      "loss": 4.0784,
+      "step": 646144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9412249903856426e-05,
+      "loss": 4.0739,
+      "step": 646656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9403863956345906e-05,
+      "loss": 4.0733,
+      "step": 647168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9395478008835386e-05,
+      "loss": 4.0751,
+      "step": 647680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9387092061324866e-05,
+      "loss": 4.079,
+      "step": 648192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9378706113814346e-05,
+      "loss": 4.0855,
+      "step": 648704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9370320166303826e-05,
+      "loss": 4.0627,
+      "step": 649216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9361934218793306e-05,
+      "loss": 4.0745,
+      "step": 649728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.935354827128278e-05,
+      "loss": 4.0807,
+      "step": 650240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.934516232377226e-05,
+      "loss": 4.0552,
+      "step": 650752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9336809133869204e-05,
+      "loss": 4.0756,
+      "step": 651264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9328423186358684e-05,
+      "loss": 4.0742,
+      "step": 651776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.932003723884817e-05,
+      "loss": 4.0817,
+      "step": 652288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.931165129133765e-05,
+      "loss": 4.0744,
+      "step": 652800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.930326534382713e-05,
+      "loss": 4.0588,
+      "step": 653312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9294879396316604e-05,
+      "loss": 4.055,
+      "step": 653824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9286493448806084e-05,
+      "loss": 4.0673,
+      "step": 654336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9278107501295564e-05,
+      "loss": 4.0763,
+      "step": 654848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.926973793258878e-05,
+      "loss": 4.0657,
+      "step": 655360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.926135198507825e-05,
+      "loss": 4.0641,
+      "step": 655872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.925296603756773e-05,
+      "loss": 4.0676,
+      "step": 656384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.924458009005721e-05,
+      "loss": 4.0753,
+      "step": 656896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.923619414254669e-05,
+      "loss": 4.0655,
+      "step": 657408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.922782457383991e-05,
+      "loss": 4.0636,
+      "step": 657920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.921943862632939e-05,
+      "loss": 4.0707,
+      "step": 658432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.921105267881887e-05,
+      "loss": 4.0783,
+      "step": 658944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.920266673130835e-05,
+      "loss": 4.0681,
+      "step": 659456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.919429716260156e-05,
+      "loss": 4.0606,
+      "step": 659968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.918591121509104e-05,
+      "loss": 4.0609,
+      "step": 660480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.917754164638425e-05,
+      "loss": 4.066,
+      "step": 660992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9169155698873727e-05,
+      "loss": 4.0757,
+      "step": 661504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9160769751363206e-05,
+      "loss": 4.0748,
+      "step": 662016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9152383803852686e-05,
+      "loss": 4.0644,
+      "step": 662528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9143997856342166e-05,
+      "loss": 4.0627,
+      "step": 663040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9135611908831646e-05,
+      "loss": 4.0577,
+      "step": 663552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9127225961321126e-05,
+      "loss": 4.0609,
+      "step": 664064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.911885639261434e-05,
+      "loss": 4.0629,
+      "step": 664576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.911047044510382e-05,
+      "loss": 4.0522,
+      "step": 665088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.91020844975933e-05,
+      "loss": 4.0725,
+      "step": 665600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.909369855008278e-05,
+      "loss": 4.0611,
+      "step": 666112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.908531260257226e-05,
+      "loss": 4.0633,
+      "step": 666624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.907694303386547e-05,
+      "loss": 4.0576,
+      "step": 667136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.906855708635495e-05,
+      "loss": 4.0585,
+      "step": 667648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.906017113884443e-05,
+      "loss": 4.0723,
+      "step": 668160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.905178519133391e-05,
+      "loss": 4.0548,
+      "step": 668672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.904339924382339e-05,
+      "loss": 4.0847,
+      "step": 669184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.903501329631287e-05,
+      "loss": 4.0714,
+      "step": 669696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.902664372760608e-05,
+      "loss": 4.0652,
+      "step": 670208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.901825778009556e-05,
+      "loss": 4.0668,
+      "step": 670720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900987183258504e-05,
+      "loss": 4.0604,
+      "step": 671232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9001485885074527e-05,
+      "loss": 4.062,
+      "step": 671744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8993099937564007e-05,
+      "loss": 4.0699,
+      "step": 672256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8984713990053487e-05,
+      "loss": 4.0621,
+      "step": 672768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8976328042542966e-05,
+      "loss": 4.0675,
+      "step": 673280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8967958473836176e-05,
+      "loss": 4.059,
+      "step": 673792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8959572526325656e-05,
+      "loss": 4.0643,
+      "step": 674304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8951186578815135e-05,
+      "loss": 4.0635,
+      "step": 674816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8942800631304615e-05,
+      "loss": 4.0534,
+      "step": 675328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.893441468379409e-05,
+      "loss": 4.0685,
+      "step": 675840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8926045115087304e-05,
+      "loss": 4.055,
+      "step": 676352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8917659167576784e-05,
+      "loss": 4.0753,
+      "step": 676864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8909273220066264e-05,
+      "loss": 4.0635,
+      "step": 677376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8900887272555744e-05,
+      "loss": 4.0665,
+      "step": 677888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8892501325045224e-05,
+      "loss": 4.0525,
+      "step": 678400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8884115377534704e-05,
+      "loss": 4.0725,
+      "step": 678912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8875729430024184e-05,
+      "loss": 4.0631,
+      "step": 679424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8867343482513664e-05,
+      "loss": 4.069,
+      "step": 679936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8858957535003144e-05,
+      "loss": 4.0613,
+      "step": 680448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.885058796629635e-05,
+      "loss": 4.0694,
+      "step": 680960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.884220201878583e-05,
+      "loss": 4.0654,
+      "step": 681472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.883381607127531e-05,
+      "loss": 4.0807,
+      "step": 681984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.882543012376479e-05,
+      "loss": 4.0532,
+      "step": 682496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8817060555058e-05,
+      "loss": 4.072,
+      "step": 683008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.880867460754748e-05,
+      "loss": 4.0645,
+      "step": 683520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.880028866003696e-05,
+      "loss": 4.0678,
+      "step": 684032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.879191909133018e-05,
+      "loss": 4.0552,
+      "step": 684544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.878353314381966e-05,
+      "loss": 4.0676,
+      "step": 685056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.877514719630914e-05,
+      "loss": 4.0644,
+      "step": 685568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.876676124879862e-05,
+      "loss": 4.0555,
+      "step": 686080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.87583753012881e-05,
+      "loss": 4.0646,
+      "step": 686592
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.090025424957275,
+      "eval_runtime": 298.162,
+      "eval_samples_per_second": 1279.811,
+      "eval_steps_per_second": 39.995,
+      "step": 686880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.874998935377758e-05,
+      "loss": 4.0529,
+      "step": 687104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.874160340626706e-05,
+      "loss": 4.0494,
+      "step": 687616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.873321745875654e-05,
+      "loss": 4.0605,
+      "step": 688128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.872484789004975e-05,
+      "loss": 4.0742,
+      "step": 688640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.871646194253923e-05,
+      "loss": 4.0659,
+      "step": 689152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.870807599502871e-05,
+      "loss": 4.053,
+      "step": 689664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.869969004751819e-05,
+      "loss": 4.0705,
+      "step": 690176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.86913204788114e-05,
+      "loss": 4.0485,
+      "step": 690688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.868293453130088e-05,
+      "loss": 4.0589,
+      "step": 691200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.867456496259409e-05,
+      "loss": 4.0728,
+      "step": 691712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.866617901508357e-05,
+      "loss": 4.0599,
+      "step": 692224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.865779306757305e-05,
+      "loss": 4.0633,
+      "step": 692736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.864942349886626e-05,
+      "loss": 4.0656,
+      "step": 693248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.864103755135574e-05,
+      "loss": 4.0531,
+      "step": 693760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.863265160384522e-05,
+      "loss": 4.0499,
+      "step": 694272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.86242656563347e-05,
+      "loss": 4.0489,
+      "step": 694784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.861587970882418e-05,
+      "loss": 4.0523,
+      "step": 695296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.860749376131366e-05,
+      "loss": 4.0579,
+      "step": 695808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.859910781380314e-05,
+      "loss": 4.0547,
+      "step": 696320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.859072186629262e-05,
+      "loss": 4.0657,
+      "step": 696832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.85823359187821e-05,
+      "loss": 4.0714,
+      "step": 697344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.857394997127158e-05,
+      "loss": 4.0705,
+      "step": 697856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.856556402376106e-05,
+      "loss": 4.062,
+      "step": 698368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.855717807625054e-05,
+      "loss": 4.0551,
+      "step": 698880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.854879212874002e-05,
+      "loss": 4.0645,
+      "step": 699392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.854042256003323e-05,
+      "loss": 4.0577,
+      "step": 699904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.853203661252271e-05,
+      "loss": 4.0547,
+      "step": 700416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.852365066501219e-05,
+      "loss": 4.0529,
+      "step": 700928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.851526471750167e-05,
+      "loss": 4.0539,
+      "step": 701440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.850689514879488e-05,
+      "loss": 4.0516,
+      "step": 701952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.849850920128436e-05,
+      "loss": 4.0526,
+      "step": 702464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.849012325377384e-05,
+      "loss": 4.0598,
+      "step": 702976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.848173730626332e-05,
+      "loss": 4.0664,
+      "step": 703488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8473351358752805e-05,
+      "loss": 4.0578,
+      "step": 704000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8464965411242285e-05,
+      "loss": 4.0584,
+      "step": 704512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8456595842535494e-05,
+      "loss": 4.0556,
+      "step": 705024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8448209895024974e-05,
+      "loss": 4.0638,
+      "step": 705536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.843984032631818e-05,
+      "loss": 4.0467,
+      "step": 706048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.843145437880766e-05,
+      "loss": 4.054,
+      "step": 706560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.842306843129714e-05,
+      "loss": 4.0446,
+      "step": 707072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.841468248378662e-05,
+      "loss": 4.0544,
+      "step": 707584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.84062965362761e-05,
+      "loss": 4.0561,
+      "step": 708096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.839791058876558e-05,
+      "loss": 4.0523,
+      "step": 708608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.838952464125506e-05,
+      "loss": 4.0559,
+      "step": 709120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.838113869374454e-05,
+      "loss": 4.0608,
+      "step": 709632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.837275274623402e-05,
+      "loss": 4.0589,
+      "step": 710144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.83643667987235e-05,
+      "loss": 4.0611,
+      "step": 710656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.835598085121299e-05,
+      "loss": 4.0556,
+      "step": 711168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.834762766130993e-05,
+      "loss": 4.0248,
+      "step": 711680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.833924171379941e-05,
+      "loss": 4.0647,
+      "step": 712192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.833085576628889e-05,
+      "loss": 4.0528,
+      "step": 712704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.832246981877837e-05,
+      "loss": 4.0603,
+      "step": 713216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.831408387126785e-05,
+      "loss": 4.0541,
+      "step": 713728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.830569792375733e-05,
+      "loss": 4.0425,
+      "step": 714240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.829731197624681e-05,
+      "loss": 4.039,
+      "step": 714752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8288942407540016e-05,
+      "loss": 4.0463,
+      "step": 715264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8280556460029496e-05,
+      "loss": 4.0463,
+      "step": 715776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8272170512518976e-05,
+      "loss": 4.04,
+      "step": 716288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8263784565008456e-05,
+      "loss": 4.0642,
+      "step": 716800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8255398617497936e-05,
+      "loss": 4.046,
+      "step": 717312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8247012669987416e-05,
+      "loss": 4.0317,
+      "step": 717824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.823864310128063e-05,
+      "loss": 4.0635,
+      "step": 718336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8230257153770105e-05,
+      "loss": 4.0338,
+      "step": 718848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8221871206259585e-05,
+      "loss": 4.037,
+      "step": 719360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8213485258749065e-05,
+      "loss": 4.0547,
+      "step": 719872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8205099311238545e-05,
+      "loss": 4.0454,
+      "step": 720384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8196713363728025e-05,
+      "loss": 4.0419,
+      "step": 720896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8188327416217505e-05,
+      "loss": 4.0436,
+      "step": 721408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8179941468706985e-05,
+      "loss": 4.037,
+      "step": 721920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8171555521196465e-05,
+      "loss": 4.0529,
+      "step": 722432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.816318595248968e-05,
+      "loss": 4.0477,
+      "step": 722944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.815481638378289e-05,
+      "loss": 4.0502,
+      "step": 723456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.814643043627237e-05,
+      "loss": 4.048,
+      "step": 723968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.813804448876185e-05,
+      "loss": 4.053,
+      "step": 724480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.812965854125133e-05,
+      "loss": 4.0627,
+      "step": 724992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.812127259374081e-05,
+      "loss": 4.0338,
+      "step": 725504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.811290302503402e-05,
+      "loss": 4.0491,
+      "step": 726016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.81045170775235e-05,
+      "loss": 4.0595,
+      "step": 726528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.809613113001298e-05,
+      "loss": 4.0319,
+      "step": 727040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.808774518250246e-05,
+      "loss": 4.0464,
+      "step": 727552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.807937561379567e-05,
+      "loss": 4.0518,
+      "step": 728064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.807098966628515e-05,
+      "loss": 4.0554,
+      "step": 728576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8062603718774634e-05,
+      "loss": 4.0562,
+      "step": 729088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8054217771264114e-05,
+      "loss": 4.0325,
+      "step": 729600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8045831823753594e-05,
+      "loss": 4.03,
+      "step": 730112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.80374622550468e-05,
+      "loss": 4.0478,
+      "step": 730624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.802907630753628e-05,
+      "loss": 4.0509,
+      "step": 731136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.802069036002576e-05,
+      "loss": 4.0384,
+      "step": 731648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.801230441251524e-05,
+      "loss": 4.0442,
+      "step": 732160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.800391846500472e-05,
+      "loss": 4.0407,
+      "step": 732672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.79955325174942e-05,
+      "loss": 4.0498,
+      "step": 733184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.798714656998368e-05,
+      "loss": 4.0442,
+      "step": 733696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.797876062247316e-05,
+      "loss": 4.0405,
+      "step": 734208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.797037467496264e-05,
+      "loss": 4.0424,
+      "step": 734720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.796200510625585e-05,
+      "loss": 4.0527,
+      "step": 735232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.795361915874533e-05,
+      "loss": 4.0471,
+      "step": 735744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.794523321123481e-05,
+      "loss": 4.0383,
+      "step": 736256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.79368472637243e-05,
+      "loss": 4.0355,
+      "step": 736768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.792849407382124e-05,
+      "loss": 4.0426,
+      "step": 737280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.792010812631072e-05,
+      "loss": 4.0518,
+      "step": 737792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.79117221788002e-05,
+      "loss": 4.0482,
+      "step": 738304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7903352610093406e-05,
+      "loss": 4.0403,
+      "step": 738816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7894966662582886e-05,
+      "loss": 4.04,
+      "step": 739328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7886580715072366e-05,
+      "loss": 4.034,
+      "step": 739840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7878194767561846e-05,
+      "loss": 4.0399,
+      "step": 740352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7869808820051326e-05,
+      "loss": 4.0341,
+      "step": 740864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7861422872540806e-05,
+      "loss": 4.0315,
+      "step": 741376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7853036925030286e-05,
+      "loss": 4.0489,
+      "step": 741888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7844650977519766e-05,
+      "loss": 4.0351,
+      "step": 742400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7836265030009246e-05,
+      "loss": 4.04,
+      "step": 742912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7827879082498725e-05,
+      "loss": 4.0353,
+      "step": 743424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7819493134988205e-05,
+      "loss": 4.0333,
+      "step": 743936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7811107187477685e-05,
+      "loss": 4.0489,
+      "step": 744448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7802737618770894e-05,
+      "loss": 4.0335,
+      "step": 744960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7794351671260374e-05,
+      "loss": 4.0574,
+      "step": 745472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7785965723749854e-05,
+      "loss": 4.047,
+      "step": 745984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7777579776239334e-05,
+      "loss": 4.0427,
+      "step": 746496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7769210207532543e-05,
+      "loss": 4.0443,
+      "step": 747008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7760824260022023e-05,
+      "loss": 4.0388,
+      "step": 747520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.77524383125115e-05,
+      "loss": 4.035,
+      "step": 748032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.774405236500099e-05,
+      "loss": 4.0484,
+      "step": 748544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.773566641749047e-05,
+      "loss": 4.0437,
+      "step": 749056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.772729684878368e-05,
+      "loss": 4.0403,
+      "step": 749568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.771891090127316e-05,
+      "loss": 4.0352,
+      "step": 750080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.771052495376264e-05,
+      "loss": 4.0407,
+      "step": 750592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.770213900625212e-05,
+      "loss": 4.04,
+      "step": 751104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.769376943754533e-05,
+      "loss": 4.0307,
+      "step": 751616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.768538349003481e-05,
+      "loss": 4.0464,
+      "step": 752128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.767699754252429e-05,
+      "loss": 4.0247,
+      "step": 752640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.76686279738175e-05,
+      "loss": 4.0579,
+      "step": 753152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.766024202630698e-05,
+      "loss": 4.0374,
+      "step": 753664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.765185607879646e-05,
+      "loss": 4.0447,
+      "step": 754176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7643470131285944e-05,
+      "loss": 4.031,
+      "step": 754688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7635084183775424e-05,
+      "loss": 4.0494,
+      "step": 755200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.762671461506863e-05,
+      "loss": 4.0392,
+      "step": 755712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.761832866755811e-05,
+      "loss": 4.0455,
+      "step": 756224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.760994272004759e-05,
+      "loss": 4.0401,
+      "step": 756736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.760155677253707e-05,
+      "loss": 4.0418,
+      "step": 757248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.759317082502655e-05,
+      "loss": 4.0423,
+      "step": 757760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.758478487751603e-05,
+      "loss": 4.0587,
+      "step": 758272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.757639893000551e-05,
+      "loss": 4.0306,
+      "step": 758784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.756802936129872e-05,
+      "loss": 4.0488,
+      "step": 759296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.75596434137882e-05,
+      "loss": 4.0455,
+      "step": 759808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.755125746627768e-05,
+      "loss": 4.0436,
+      "step": 760320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.754287151876716e-05,
+      "loss": 4.0323,
+      "step": 760832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.753448557125664e-05,
+      "loss": 4.0443,
+      "step": 761344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.752609962374613e-05,
+      "loss": 4.0394,
+      "step": 761856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.75177136762356e-05,
+      "loss": 4.035,
+      "step": 762368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.750934410752882e-05,
+      "loss": 4.0398,
+      "step": 762880
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.074465274810791,
+      "eval_runtime": 296.8734,
+      "eval_samples_per_second": 1285.366,
+      "eval_steps_per_second": 40.169,
+      "step": 763200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.75009581600183e-05,
+      "loss": 4.0307,
+      "step": 763392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.749257221250778e-05,
+      "loss": 4.0244,
+      "step": 763904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.748418626499725e-05,
+      "loss": 4.0393,
+      "step": 764416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.747580031748673e-05,
+      "loss": 4.0493,
+      "step": 764928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.746741436997621e-05,
+      "loss": 4.0449,
+      "step": 765440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.745902842246569e-05,
+      "loss": 4.0318,
+      "step": 765952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.745064247495517e-05,
+      "loss": 4.0446,
+      "step": 766464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.744225652744465e-05,
+      "loss": 4.0305,
+      "step": 766976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.743387057993413e-05,
+      "loss": 4.0333,
+      "step": 767488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.742548463242361e-05,
+      "loss": 4.0473,
+      "step": 768000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.741709868491309e-05,
+      "loss": 4.0374,
+      "step": 768512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.740871273740258e-05,
+      "loss": 4.0458,
+      "step": 769024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7400343168695786e-05,
+      "loss": 4.045,
+      "step": 769536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7391957221185266e-05,
+      "loss": 4.0289,
+      "step": 770048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7383571273674746e-05,
+      "loss": 4.0281,
+      "step": 770560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7375185326164226e-05,
+      "loss": 4.0237,
+      "step": 771072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7366799378653706e-05,
+      "loss": 4.0337,
+      "step": 771584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7358413431143186e-05,
+      "loss": 4.0381,
+      "step": 772096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7350027483632666e-05,
+      "loss": 4.0272,
+      "step": 772608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.734164153612214e-05,
+      "loss": 4.0401,
+      "step": 773120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.733325558861162e-05,
+      "loss": 4.0485,
+      "step": 773632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.73248696411011e-05,
+      "loss": 4.0442,
+      "step": 774144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.731648369359058e-05,
+      "loss": 4.0408,
+      "step": 774656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7308114124883794e-05,
+      "loss": 4.0353,
+      "step": 775168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7299728177373274e-05,
+      "loss": 4.0406,
+      "step": 775680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7291342229862754e-05,
+      "loss": 4.0374,
+      "step": 776192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7282956282352234e-05,
+      "loss": 4.0281,
+      "step": 776704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7274586713645443e-05,
+      "loss": 4.0275,
+      "step": 777216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.726620076613492e-05,
+      "loss": 4.0367,
+      "step": 777728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.72578148186244e-05,
+      "loss": 4.0286,
+      "step": 778240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.724942887111388e-05,
+      "loss": 4.0299,
+      "step": 778752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.724104292360336e-05,
+      "loss": 4.0375,
+      "step": 779264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.723267335489657e-05,
+      "loss": 4.0397,
+      "step": 779776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.722428740738605e-05,
+      "loss": 4.0425,
+      "step": 780288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.721590145987553e-05,
+      "loss": 4.0291,
+      "step": 780800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.720753189116875e-05,
+      "loss": 4.0374,
+      "step": 781312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.719914594365823e-05,
+      "loss": 4.0432,
+      "step": 781824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.719075999614771e-05,
+      "loss": 4.0196,
+      "step": 782336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.718237404863719e-05,
+      "loss": 4.0316,
+      "step": 782848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.717398810112667e-05,
+      "loss": 4.0236,
+      "step": 783360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.716560215361615e-05,
+      "loss": 4.0329,
+      "step": 783872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.715721620610563e-05,
+      "loss": 4.0307,
+      "step": 784384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.714883025859511e-05,
+      "loss": 4.0298,
+      "step": 784896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.714046068988832e-05,
+      "loss": 4.0348,
+      "step": 785408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.71320747423778e-05,
+      "loss": 4.0389,
+      "step": 785920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.712368879486728e-05,
+      "loss": 4.0388,
+      "step": 786432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.711530284735676e-05,
+      "loss": 4.0405,
+      "step": 786944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.710691689984624e-05,
+      "loss": 4.0286,
+      "step": 787456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.709853095233572e-05,
+      "loss": 4.0054,
+      "step": 787968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.70901450048252e-05,
+      "loss": 4.0414,
+      "step": 788480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.708177543611841e-05,
+      "loss": 4.0319,
+      "step": 788992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.707338948860789e-05,
+      "loss": 4.0378,
+      "step": 789504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.706500354109737e-05,
+      "loss": 4.029,
+      "step": 790016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.705661759358685e-05,
+      "loss": 4.0225,
+      "step": 790528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7048231646076326e-05,
+      "loss": 4.0176,
+      "step": 791040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.703986207736954e-05,
+      "loss": 4.023,
+      "step": 791552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.703147612985902e-05,
+      "loss": 4.028,
+      "step": 792064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.70230901823485e-05,
+      "loss": 4.0124,
+      "step": 792576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7014704234837975e-05,
+      "loss": 4.0433,
+      "step": 793088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7006318287327454e-05,
+      "loss": 4.0261,
+      "step": 793600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6997932339816934e-05,
+      "loss": 4.0095,
+      "step": 794112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.698956277111015e-05,
+      "loss": 4.0366,
+      "step": 794624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.698117682359963e-05,
+      "loss": 4.014,
+      "step": 795136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.697279087608911e-05,
+      "loss": 4.018,
+      "step": 795648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.696440492857859e-05,
+      "loss": 4.0315,
+      "step": 796160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.695601898106807e-05,
+      "loss": 4.0254,
+      "step": 796672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6947665791165015e-05,
+      "loss": 4.0178,
+      "step": 797184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6939279843654495e-05,
+      "loss": 4.0203,
+      "step": 797696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6930893896143975e-05,
+      "loss": 4.017,
+      "step": 798208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.692250794863345e-05,
+      "loss": 4.026,
+      "step": 798720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.691412200112293e-05,
+      "loss": 4.033,
+      "step": 799232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.690573605361241e-05,
+      "loss": 4.0246,
+      "step": 799744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.689735010610189e-05,
+      "loss": 4.0267,
+      "step": 800256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.688896415859137e-05,
+      "loss": 4.0271,
+      "step": 800768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6880578211080855e-05,
+      "loss": 4.0431,
+      "step": 801280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6872192263570335e-05,
+      "loss": 4.0152,
+      "step": 801792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6863806316059815e-05,
+      "loss": 4.0276,
+      "step": 802304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6855420368549295e-05,
+      "loss": 4.0312,
+      "step": 802816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.684706717864623e-05,
+      "loss": 4.0166,
+      "step": 803328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.683869760993945e-05,
+      "loss": 4.0228,
+      "step": 803840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.683031166242892e-05,
+      "loss": 4.0278,
+      "step": 804352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.68219257149184e-05,
+      "loss": 4.0325,
+      "step": 804864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.681353976740788e-05,
+      "loss": 4.0368,
+      "step": 805376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.680515381989736e-05,
+      "loss": 4.0063,
+      "step": 805888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.679676787238684e-05,
+      "loss": 4.0132,
+      "step": 806400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.678838192487632e-05,
+      "loss": 4.0225,
+      "step": 806912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.677999597736581e-05,
+      "loss": 4.0284,
+      "step": 807424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.677161002985529e-05,
+      "loss": 4.0145,
+      "step": 807936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.676322408234477e-05,
+      "loss": 4.0244,
+      "step": 808448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.675483813483425e-05,
+      "loss": 4.0213,
+      "step": 808960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.674645218732373e-05,
+      "loss": 4.0293,
+      "step": 809472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.673808261861694e-05,
+      "loss": 4.0225,
+      "step": 809984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.672969667110642e-05,
+      "loss": 4.0168,
+      "step": 810496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6721327102399626e-05,
+      "loss": 4.0247,
+      "step": 811008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6712941154889106e-05,
+      "loss": 4.0313,
+      "step": 811520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6704555207378586e-05,
+      "loss": 4.0268,
+      "step": 812032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6696169259868066e-05,
+      "loss": 4.0118,
+      "step": 812544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6687783312357546e-05,
+      "loss": 4.0144,
+      "step": 813056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6679397364847026e-05,
+      "loss": 4.0233,
+      "step": 813568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6671011417336506e-05,
+      "loss": 4.0286,
+      "step": 814080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6662625469825986e-05,
+      "loss": 4.0287,
+      "step": 814592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6654239522315466e-05,
+      "loss": 4.0185,
+      "step": 815104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6645853574804946e-05,
+      "loss": 4.0183,
+      "step": 815616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6637467627294426e-05,
+      "loss": 4.0128,
+      "step": 816128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6629081679783906e-05,
+      "loss": 4.0183,
+      "step": 816640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6620712111077115e-05,
+      "loss": 4.0143,
+      "step": 817152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6612326163566595e-05,
+      "loss": 4.0121,
+      "step": 817664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.660395659485981e-05,
+      "loss": 4.0258,
+      "step": 818176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6595570647349284e-05,
+      "loss": 4.0139,
+      "step": 818688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6587184699838764e-05,
+      "loss": 4.0171,
+      "step": 819200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6578798752328244e-05,
+      "loss": 4.0129,
+      "step": 819712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.657041280481773e-05,
+      "loss": 4.0117,
+      "step": 820224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.656202685730721e-05,
+      "loss": 4.0254,
+      "step": 820736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.655364090979669e-05,
+      "loss": 4.0204,
+      "step": 821248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.654525496228617e-05,
+      "loss": 4.0301,
+      "step": 821760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.653688539357938e-05,
+      "loss": 4.0248,
+      "step": 822272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.652851582487259e-05,
+      "loss": 4.024,
+      "step": 822784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6520146256165805e-05,
+      "loss": 4.0226,
+      "step": 823296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6511760308655285e-05,
+      "loss": 4.0178,
+      "step": 823808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.650337436114476e-05,
+      "loss": 4.0191,
+      "step": 824320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.649498841363424e-05,
+      "loss": 4.0256,
+      "step": 824832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.648660246612372e-05,
+      "loss": 4.0219,
+      "step": 825344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.64782165186132e-05,
+      "loss": 4.0241,
+      "step": 825856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6469830571102684e-05,
+      "loss": 4.011,
+      "step": 826368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6461444623592164e-05,
+      "loss": 4.0234,
+      "step": 826880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6453075054885373e-05,
+      "loss": 4.0156,
+      "step": 827392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.644468910737485e-05,
+      "loss": 4.0127,
+      "step": 827904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.643630315986433e-05,
+      "loss": 4.0215,
+      "step": 828416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.642791721235381e-05,
+      "loss": 4.004,
+      "step": 828928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.641953126484329e-05,
+      "loss": 4.0394,
+      "step": 829440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.641114531733277e-05,
+      "loss": 4.0161,
+      "step": 829952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.640275936982225e-05,
+      "loss": 4.0257,
+      "step": 830464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.639437342231173e-05,
+      "loss": 4.0106,
+      "step": 830976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.638600385360494e-05,
+      "loss": 4.0255,
+      "step": 831488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.637761790609442e-05,
+      "loss": 4.0184,
+      "step": 832000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.63692319585839e-05,
+      "loss": 4.0265,
+      "step": 832512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.636084601107338e-05,
+      "loss": 4.0175,
+      "step": 833024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.635246006356286e-05,
+      "loss": 4.0191,
+      "step": 833536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.634409049485608e-05,
+      "loss": 4.0229,
+      "step": 834048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.633570454734556e-05,
+      "loss": 4.042,
+      "step": 834560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.632731859983504e-05,
+      "loss": 4.0051,
+      "step": 835072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.631893265232452e-05,
+      "loss": 4.0321,
+      "step": 835584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.631056308361773e-05,
+      "loss": 4.0267,
+      "step": 836096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6302193514910936e-05,
+      "loss": 4.0197,
+      "step": 836608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6293807567400416e-05,
+      "loss": 4.0123,
+      "step": 837120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6285421619889896e-05,
+      "loss": 4.0225,
+      "step": 837632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6277035672379376e-05,
+      "loss": 4.0192,
+      "step": 838144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6268649724868856e-05,
+      "loss": 4.0185,
+      "step": 838656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6260263777358336e-05,
+      "loss": 4.0153,
+      "step": 839168
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.060510635375977,
+      "eval_runtime": 304.7531,
+      "eval_samples_per_second": 1252.132,
+      "eval_steps_per_second": 39.13,
+      "step": 839520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6251877829847816e-05,
+      "loss": 4.012,
+      "step": 839680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6243491882337296e-05,
+      "loss": 4.0065,
+      "step": 840192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.623513869243424e-05,
+      "loss": 4.0162,
+      "step": 840704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.622675274492372e-05,
+      "loss": 4.0277,
+      "step": 841216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.62183667974132e-05,
+      "loss": 4.0269,
+      "step": 841728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.620998084990268e-05,
+      "loss": 4.0135,
+      "step": 842240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.620159490239216e-05,
+      "loss": 4.0258,
+      "step": 842752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.619320895488164e-05,
+      "loss": 4.0114,
+      "step": 843264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6184823007371114e-05,
+      "loss": 4.0105,
+      "step": 843776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6176437059860594e-05,
+      "loss": 4.0236,
+      "step": 844288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.616806749115381e-05,
+      "loss": 4.025,
+      "step": 844800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.615968154364329e-05,
+      "loss": 4.0244,
+      "step": 845312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6151328353740234e-05,
+      "loss": 4.0212,
+      "step": 845824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6142942406229714e-05,
+      "loss": 4.01,
+      "step": 846336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6134556458719194e-05,
+      "loss": 4.0109,
+      "step": 846848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6126170511208674e-05,
+      "loss": 3.9988,
+      "step": 847360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6117784563698154e-05,
+      "loss": 4.0167,
+      "step": 847872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6109398616187634e-05,
+      "loss": 4.017,
+      "step": 848384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6101012668677114e-05,
+      "loss": 4.0065,
+      "step": 848896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.609262672116659e-05,
+      "loss": 4.0189,
+      "step": 849408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.60842571524598e-05,
+      "loss": 4.0286,
+      "step": 849920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.607587120494928e-05,
+      "loss": 4.0233,
+      "step": 850432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.606748525743876e-05,
+      "loss": 4.022,
+      "step": 850944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.605909930992824e-05,
+      "loss": 4.0157,
+      "step": 851456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.605071336241772e-05,
+      "loss": 4.0197,
+      "step": 851968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.60423274149072e-05,
+      "loss": 4.019,
+      "step": 852480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.603395784620042e-05,
+      "loss": 4.0085,
+      "step": 852992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.602557189868989e-05,
+      "loss": 4.0048,
+      "step": 853504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.601718595117937e-05,
+      "loss": 4.0159,
+      "step": 854016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.600880000366885e-05,
+      "loss": 4.0098,
+      "step": 854528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.600041405615833e-05,
+      "loss": 4.0142,
+      "step": 855040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.599202810864781e-05,
+      "loss": 4.0149,
+      "step": 855552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.598364216113729e-05,
+      "loss": 4.0163,
+      "step": 856064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.597525621362677e-05,
+      "loss": 4.0257,
+      "step": 856576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.596687026611625e-05,
+      "loss": 4.0073,
+      "step": 857088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.595850069740946e-05,
+      "loss": 4.0196,
+      "step": 857600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.595011474989895e-05,
+      "loss": 4.021,
+      "step": 858112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.594172880238843e-05,
+      "loss": 3.9986,
+      "step": 858624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.593334285487791e-05,
+      "loss": 4.0147,
+      "step": 859136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.592495690736739e-05,
+      "loss": 4.0056,
+      "step": 859648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.591657095985687e-05,
+      "loss": 4.0086,
+      "step": 860160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5908201391150076e-05,
+      "loss": 4.0117,
+      "step": 860672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5899815443639556e-05,
+      "loss": 4.0122,
+      "step": 861184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5891445874932766e-05,
+      "loss": 4.0138,
+      "step": 861696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5883059927422245e-05,
+      "loss": 4.0187,
+      "step": 862208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5874673979911725e-05,
+      "loss": 4.0198,
+      "step": 862720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5866288032401205e-05,
+      "loss": 4.0203,
+      "step": 863232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5857902084890685e-05,
+      "loss": 4.0115,
+      "step": 863744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5849516137380165e-05,
+      "loss": 3.9875,
+      "step": 864256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.584114656867338e-05,
+      "loss": 4.0195,
+      "step": 864768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.583276062116286e-05,
+      "loss": 4.0122,
+      "step": 865280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.582437467365234e-05,
+      "loss": 4.0216,
+      "step": 865792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.581598872614182e-05,
+      "loss": 4.0112,
+      "step": 866304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.58076027786313e-05,
+      "loss": 4.0005,
+      "step": 866816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5799216831120774e-05,
+      "loss": 4.0043,
+      "step": 867328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.579084726241399e-05,
+      "loss": 3.9961,
+      "step": 867840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.578246131490347e-05,
+      "loss": 4.0104,
+      "step": 868352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.577407536739295e-05,
+      "loss": 3.9888,
+      "step": 868864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.576568941988242e-05,
+      "loss": 4.027,
+      "step": 869376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.57573034723719e-05,
+      "loss": 4.0083,
+      "step": 869888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.574891752486138e-05,
+      "loss": 3.99,
+      "step": 870400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.57405479561546e-05,
+      "loss": 4.0176,
+      "step": 870912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.573216200864408e-05,
+      "loss": 3.9955,
+      "step": 871424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.572377606113356e-05,
+      "loss": 3.9996,
+      "step": 871936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.571539011362304e-05,
+      "loss": 4.0083,
+      "step": 872448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.570700416611252e-05,
+      "loss": 4.0048,
+      "step": 872960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5698618218602e-05,
+      "loss": 3.9998,
+      "step": 873472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.569024864989521e-05,
+      "loss": 4.0063,
+      "step": 873984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.568186270238469e-05,
+      "loss": 3.9958,
+      "step": 874496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.567347675487417e-05,
+      "loss": 4.0079,
+      "step": 875008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.566509080736365e-05,
+      "loss": 4.0148,
+      "step": 875520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.565670485985313e-05,
+      "loss": 4.0062,
+      "step": 876032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.564831891234261e-05,
+      "loss": 4.0109,
+      "step": 876544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.563993296483209e-05,
+      "loss": 4.0041,
+      "step": 877056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.563154701732157e-05,
+      "loss": 4.0298,
+      "step": 877568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.562317744861478e-05,
+      "loss": 3.9939,
+      "step": 878080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.561479150110426e-05,
+      "loss": 4.0121,
+      "step": 878592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.560642193239747e-05,
+      "loss": 4.002,
+      "step": 879104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.559803598488695e-05,
+      "loss": 4.007,
+      "step": 879616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.558965003737643e-05,
+      "loss": 4.0027,
+      "step": 880128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.558126408986591e-05,
+      "loss": 4.0114,
+      "step": 880640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.557287814235539e-05,
+      "loss": 4.0154,
+      "step": 881152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.556449219484487e-05,
+      "loss": 4.0173,
+      "step": 881664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.555610624733435e-05,
+      "loss": 3.9884,
+      "step": 882176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.554772029982383e-05,
+      "loss": 3.9919,
+      "step": 882688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.553935073111704e-05,
+      "loss": 4.0048,
+      "step": 883200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.553096478360652e-05,
+      "loss": 4.0119,
+      "step": 883712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5522578836096e-05,
+      "loss": 3.9972,
+      "step": 884224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.551419288858549e-05,
+      "loss": 4.0031,
+      "step": 884736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.550580694107497e-05,
+      "loss": 4.0037,
+      "step": 885248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.549743737236818e-05,
+      "loss": 4.0071,
+      "step": 885760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.548905142485766e-05,
+      "loss": 4.0052,
+      "step": 886272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.548066547734714e-05,
+      "loss": 4.0003,
+      "step": 886784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.547227952983661e-05,
+      "loss": 4.0055,
+      "step": 887296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5463909961129826e-05,
+      "loss": 4.0107,
+      "step": 887808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5455524013619306e-05,
+      "loss": 4.0138,
+      "step": 888320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5447138066108786e-05,
+      "loss": 3.9966,
+      "step": 888832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.543875211859826e-05,
+      "loss": 3.9952,
+      "step": 889344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5430382549891475e-05,
+      "loss": 4.0045,
+      "step": 889856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5421996602380955e-05,
+      "loss": 4.0094,
+      "step": 890368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.541361065487044e-05,
+      "loss": 4.0113,
+      "step": 890880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5405224707359915e-05,
+      "loss": 4.0003,
+      "step": 891392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5396838759849395e-05,
+      "loss": 4.0034,
+      "step": 891904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5388452812338875e-05,
+      "loss": 3.9953,
+      "step": 892416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5380083243632084e-05,
+      "loss": 3.9994,
+      "step": 892928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.53717136749253e-05,
+      "loss": 3.9994,
+      "step": 893440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.536332772741478e-05,
+      "loss": 3.9956,
+      "step": 893952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.535494177990426e-05,
+      "loss": 4.0045,
+      "step": 894464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.534655583239373e-05,
+      "loss": 3.9986,
+      "step": 894976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.533816988488321e-05,
+      "loss": 3.9984,
+      "step": 895488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.532978393737269e-05,
+      "loss": 3.9975,
+      "step": 896000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.532139798986218e-05,
+      "loss": 3.9927,
+      "step": 896512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.531301204235166e-05,
+      "loss": 4.0083,
+      "step": 897024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.530462609484114e-05,
+      "loss": 4.0013,
+      "step": 897536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.529625652613435e-05,
+      "loss": 4.0123,
+      "step": 898048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.528787057862383e-05,
+      "loss": 4.0047,
+      "step": 898560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.527948463111331e-05,
+      "loss": 4.01,
+      "step": 899072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.527109868360279e-05,
+      "loss": 4.006,
+      "step": 899584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.526271273609227e-05,
+      "loss": 3.9946,
+      "step": 900096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.525432678858175e-05,
+      "loss": 4.0036,
+      "step": 900608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.524594084107123e-05,
+      "loss": 4.0033,
+      "step": 901120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.523755489356071e-05,
+      "loss": 4.0025,
+      "step": 901632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.522918532485392e-05,
+      "loss": 4.0091,
+      "step": 902144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.52207993773434e-05,
+      "loss": 3.9973,
+      "step": 902656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.521241342983288e-05,
+      "loss": 4.0027,
+      "step": 903168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5204027482322364e-05,
+      "loss": 3.9962,
+      "step": 903680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.519565791361557e-05,
+      "loss": 3.9986,
+      "step": 904192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.518727196610505e-05,
+      "loss": 4.0022,
+      "step": 904704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.517888601859453e-05,
+      "loss": 3.9911,
+      "step": 905216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.517050007108401e-05,
+      "loss": 4.0183,
+      "step": 905728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.516213050237722e-05,
+      "loss": 4.0,
+      "step": 906240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.51537445548667e-05,
+      "loss": 4.0083,
+      "step": 906752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.514537498615991e-05,
+      "loss": 3.9946,
+      "step": 907264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.513698903864939e-05,
+      "loss": 4.006,
+      "step": 907776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.512860309113887e-05,
+      "loss": 4.0048,
+      "step": 908288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.512021714362835e-05,
+      "loss": 4.0102,
+      "step": 908800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.511183119611783e-05,
+      "loss": 4.0006,
+      "step": 909312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.510344524860731e-05,
+      "loss": 4.0005,
+      "step": 909824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.50950593010968e-05,
+      "loss": 4.0031,
+      "step": 910336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.508667335358627e-05,
+      "loss": 4.0236,
+      "step": 910848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5078303784879486e-05,
+      "loss": 3.9939,
+      "step": 911360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5069917837368966e-05,
+      "loss": 4.0117,
+      "step": 911872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5061531889858446e-05,
+      "loss": 4.0093,
+      "step": 912384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.505314594234792e-05,
+      "loss": 3.9996,
+      "step": 912896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5044776373641135e-05,
+      "loss": 4.0038,
+      "step": 913408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5036390426130615e-05,
+      "loss": 4.0014,
+      "step": 913920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5028020857423824e-05,
+      "loss": 4.0017,
+      "step": 914432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5019634909913304e-05,
+      "loss": 4.0009,
+      "step": 914944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5011248962402784e-05,
+      "loss": 4.0,
+      "step": 915456
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.049605369567871,
+      "eval_runtime": 305.6035,
+      "eval_samples_per_second": 1248.647,
+      "eval_steps_per_second": 39.021,
+      "step": 915840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5002863014892264e-05,
+      "loss": 4.0058,
+      "step": 915968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4994477067381744e-05,
+      "loss": 3.9846,
+      "step": 916480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4986091119871224e-05,
+      "loss": 3.9985,
+      "step": 916992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4977705172360704e-05,
+      "loss": 4.0148,
+      "step": 917504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4969319224850184e-05,
+      "loss": 4.0046,
+      "step": 918016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4960933277339664e-05,
+      "loss": 3.998,
+      "step": 918528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4952547329829144e-05,
+      "loss": 4.0149,
+      "step": 919040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4944161382318624e-05,
+      "loss": 3.9935,
+      "step": 919552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4935775434808104e-05,
+      "loss": 3.9913,
+      "step": 920064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4927389487297584e-05,
+      "loss": 4.0095,
+      "step": 920576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4919003539787064e-05,
+      "loss": 4.0075,
+      "step": 921088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4910617592276544e-05,
+      "loss": 4.0034,
+      "step": 921600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.490224802356975e-05,
+      "loss": 4.0094,
+      "step": 922112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.489386207605923e-05,
+      "loss": 3.9893,
+      "step": 922624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.488547612854872e-05,
+      "loss": 3.9898,
+      "step": 923136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.48770901810382e-05,
+      "loss": 3.9838,
+      "step": 923648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.486870423352768e-05,
+      "loss": 4.0001,
+      "step": 924160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.486031828601716e-05,
+      "loss": 4.0008,
+      "step": 924672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.485193233850663e-05,
+      "loss": 3.9932,
+      "step": 925184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.484354639099611e-05,
+      "loss": 3.9969,
+      "step": 925696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.483516044348559e-05,
+      "loss": 4.02,
+      "step": 926208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.482677449597507e-05,
+      "loss": 3.9998,
+      "step": 926720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.481838854846455e-05,
+      "loss": 4.005,
+      "step": 927232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.481000260095403e-05,
+      "loss": 4.0004,
+      "step": 927744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.480161665344351e-05,
+      "loss": 4.0053,
+      "step": 928256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.479326346354046e-05,
+      "loss": 3.999,
+      "step": 928768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.478487751602994e-05,
+      "loss": 3.995,
+      "step": 929280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.477649156851942e-05,
+      "loss": 3.9866,
+      "step": 929792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.47681056210089e-05,
+      "loss": 4.0009,
+      "step": 930304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.475971967349838e-05,
+      "loss": 3.9899,
+      "step": 930816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4751350104791586e-05,
+      "loss": 3.9977,
+      "step": 931328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4742964157281066e-05,
+      "loss": 3.9964,
+      "step": 931840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4734578209770546e-05,
+      "loss": 4.0039,
+      "step": 932352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4726192262260026e-05,
+      "loss": 4.0059,
+      "step": 932864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4717806314749506e-05,
+      "loss": 3.9967,
+      "step": 933376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4709436746042715e-05,
+      "loss": 4.0011,
+      "step": 933888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4701050798532195e-05,
+      "loss": 4.005,
+      "step": 934400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4692664851021675e-05,
+      "loss": 3.9792,
+      "step": 934912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4684278903511155e-05,
+      "loss": 3.9966,
+      "step": 935424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.467589295600064e-05,
+      "loss": 3.9899,
+      "step": 935936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.466750700849012e-05,
+      "loss": 3.9908,
+      "step": 936448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.465913743978333e-05,
+      "loss": 4.0012,
+      "step": 936960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.465075149227281e-05,
+      "loss": 3.9966,
+      "step": 937472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.464236554476229e-05,
+      "loss": 3.9968,
+      "step": 937984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.463397959725177e-05,
+      "loss": 3.9989,
+      "step": 938496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.462559364974125e-05,
+      "loss": 4.0041,
+      "step": 939008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.461720770223073e-05,
+      "loss": 4.0038,
+      "step": 939520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.460882175472021e-05,
+      "loss": 3.9973,
+      "step": 940032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.460045218601342e-05,
+      "loss": 3.9703,
+      "step": 940544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.459208261730663e-05,
+      "loss": 3.9998,
+      "step": 941056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.458369666979611e-05,
+      "loss": 3.9976,
+      "step": 941568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4575310722285595e-05,
+      "loss": 4.0035,
+      "step": 942080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4566924774775075e-05,
+      "loss": 3.996,
+      "step": 942592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4558538827264555e-05,
+      "loss": 3.9876,
+      "step": 943104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4550152879754035e-05,
+      "loss": 3.9823,
+      "step": 943616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4541766932243515e-05,
+      "loss": 3.9843,
+      "step": 944128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4533380984732995e-05,
+      "loss": 3.9913,
+      "step": 944640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.452499503722247e-05,
+      "loss": 3.9749,
+      "step": 945152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.451660908971195e-05,
+      "loss": 4.0114,
+      "step": 945664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.450822314220143e-05,
+      "loss": 3.9942,
+      "step": 946176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.449983719469091e-05,
+      "loss": 3.9749,
+      "step": 946688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.449145124718039e-05,
+      "loss": 4.0004,
+      "step": 947200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.44830816784736e-05,
+      "loss": 3.9803,
+      "step": 947712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.447471210976681e-05,
+      "loss": 3.9815,
+      "step": 948224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.446632616225629e-05,
+      "loss": 3.9901,
+      "step": 948736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.445794021474577e-05,
+      "loss": 3.9912,
+      "step": 949248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.444955426723525e-05,
+      "loss": 3.9787,
+      "step": 949760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.444118469852847e-05,
+      "loss": 3.9941,
+      "step": 950272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.443279875101794e-05,
+      "loss": 3.9773,
+      "step": 950784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.442441280350742e-05,
+      "loss": 3.988,
+      "step": 951296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.44160268559969e-05,
+      "loss": 3.9985,
+      "step": 951808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.440764090848638e-05,
+      "loss": 3.9897,
+      "step": 952320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.439925496097586e-05,
+      "loss": 3.9897,
+      "step": 952832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.439088539226907e-05,
+      "loss": 3.99,
+      "step": 953344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.438249944475855e-05,
+      "loss": 4.0143,
+      "step": 953856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.437412987605177e-05,
+      "loss": 3.9763,
+      "step": 954368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.436574392854125e-05,
+      "loss": 3.9976,
+      "step": 954880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.435735798103073e-05,
+      "loss": 3.9859,
+      "step": 955392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.434897203352021e-05,
+      "loss": 3.9882,
+      "step": 955904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.434058608600969e-05,
+      "loss": 3.9888,
+      "step": 956416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.433220013849917e-05,
+      "loss": 3.9874,
+      "step": 956928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.432381419098865e-05,
+      "loss": 4.0029,
+      "step": 957440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4315444622281856e-05,
+      "loss": 4.001,
+      "step": 957952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4307058674771336e-05,
+      "loss": 3.9725,
+      "step": 958464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4298672727260816e-05,
+      "loss": 3.977,
+      "step": 958976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4290286779750296e-05,
+      "loss": 3.9828,
+      "step": 959488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4281900832239776e-05,
+      "loss": 3.9983,
+      "step": 960000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4273514884729256e-05,
+      "loss": 3.9828,
+      "step": 960512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4265128937218735e-05,
+      "loss": 3.9855,
+      "step": 961024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4256742989708215e-05,
+      "loss": 3.9948,
+      "step": 961536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4248357042197695e-05,
+      "loss": 3.9893,
+      "step": 962048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.423997109468718e-05,
+      "loss": 3.9908,
+      "step": 962560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4231585147176655e-05,
+      "loss": 3.981,
+      "step": 963072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4223199199666135e-05,
+      "loss": 3.9922,
+      "step": 963584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.421484600976308e-05,
+      "loss": 3.9929,
+      "step": 964096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.420646006225256e-05,
+      "loss": 3.9973,
+      "step": 964608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.419807411474204e-05,
+      "loss": 3.9796,
+      "step": 965120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.418968816723152e-05,
+      "loss": 3.9803,
+      "step": 965632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4181302219721e-05,
+      "loss": 3.9891,
+      "step": 966144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.417291627221048e-05,
+      "loss": 3.9903,
+      "step": 966656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.416453032469995e-05,
+      "loss": 3.9998,
+      "step": 967168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.415614437718943e-05,
+      "loss": 3.9811,
+      "step": 967680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.414777480848265e-05,
+      "loss": 3.9868,
+      "step": 968192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.413938886097213e-05,
+      "loss": 3.9835,
+      "step": 968704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4131019292265345e-05,
+      "loss": 3.9788,
+      "step": 969216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4122633344754825e-05,
+      "loss": 3.9843,
+      "step": 969728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4114247397244305e-05,
+      "loss": 3.9849,
+      "step": 970240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.410586144973378e-05,
+      "loss": 3.9882,
+      "step": 970752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.409747550222326e-05,
+      "loss": 3.9822,
+      "step": 971264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4089105933516474e-05,
+      "loss": 3.9832,
+      "step": 971776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.408071998600595e-05,
+      "loss": 3.9838,
+      "step": 972288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.407233403849543e-05,
+      "loss": 3.974,
+      "step": 972800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.406394809098491e-05,
+      "loss": 3.9912,
+      "step": 973312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.405556214347439e-05,
+      "loss": 3.9886,
+      "step": 973824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4047176195963874e-05,
+      "loss": 3.9926,
+      "step": 974336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4038790248453354e-05,
+      "loss": 3.9929,
+      "step": 974848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4030404300942833e-05,
+      "loss": 3.9928,
+      "step": 975360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.402203473223604e-05,
+      "loss": 3.9884,
+      "step": 975872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.401364878472552e-05,
+      "loss": 3.9807,
+      "step": 976384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4005262837215e-05,
+      "loss": 3.9919,
+      "step": 976896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.399687688970448e-05,
+      "loss": 3.9858,
+      "step": 977408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.398850732099769e-05,
+      "loss": 3.9881,
+      "step": 977920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.39801377522909e-05,
+      "loss": 3.9918,
+      "step": 978432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.397175180478038e-05,
+      "loss": 3.9818,
+      "step": 978944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.396336585726986e-05,
+      "loss": 3.9868,
+      "step": 979456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.395497990975934e-05,
+      "loss": 3.9782,
+      "step": 979968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.394659396224882e-05,
+      "loss": 3.9826,
+      "step": 980480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.393820801473831e-05,
+      "loss": 3.987,
+      "step": 980992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.392982206722779e-05,
+      "loss": 3.9794,
+      "step": 981504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3921452498520996e-05,
+      "loss": 3.998,
+      "step": 982016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3913082929814205e-05,
+      "loss": 3.9883,
+      "step": 982528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3904696982303685e-05,
+      "loss": 3.9905,
+      "step": 983040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3896311034793165e-05,
+      "loss": 3.9817,
+      "step": 983552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3887925087282645e-05,
+      "loss": 3.9883,
+      "step": 984064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3879539139772125e-05,
+      "loss": 3.9932,
+      "step": 984576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3871153192261605e-05,
+      "loss": 3.9938,
+      "step": 985088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3862767244751085e-05,
+      "loss": 3.987,
+      "step": 985600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3854381297240565e-05,
+      "loss": 3.9828,
+      "step": 986112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3845995349730045e-05,
+      "loss": 3.9887,
+      "step": 986624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3837609402219525e-05,
+      "loss": 4.0029,
+      "step": 987136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3829223454709005e-05,
+      "loss": 3.9841,
+      "step": 987648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.382083750719849e-05,
+      "loss": 3.9923,
+      "step": 988160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3812451559687965e-05,
+      "loss": 3.995,
+      "step": 988672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.380408199098118e-05,
+      "loss": 3.9857,
+      "step": 989184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.379569604347066e-05,
+      "loss": 3.9879,
+      "step": 989696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.378731009596014e-05,
+      "loss": 3.9844,
+      "step": 990208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3778924148449614e-05,
+      "loss": 3.9862,
+      "step": 990720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.377055457974283e-05,
+      "loss": 3.9879,
+      "step": 991232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.376216863223231e-05,
+      "loss": 3.9807,
+      "step": 991744
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.040164470672607,
+      "eval_runtime": 294.7289,
+      "eval_samples_per_second": 1294.719,
+      "eval_steps_per_second": 40.461,
+      "step": 992160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.375378268472178e-05,
+      "loss": 3.984,
+      "step": 992256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.374539673721126e-05,
+      "loss": 3.9689,
+      "step": 992768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.373701078970074e-05,
+      "loss": 3.9845,
+      "step": 993280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.372862484219023e-05,
+      "loss": 4.0004,
+      "step": 993792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.372023889467971e-05,
+      "loss": 3.9885,
+      "step": 994304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.371186932597292e-05,
+      "loss": 3.985,
+      "step": 994816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.37034833784624e-05,
+      "loss": 3.9916,
+      "step": 995328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3695113809755614e-05,
+      "loss": 3.9829,
+      "step": 995840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.368672786224509e-05,
+      "loss": 3.9768,
+      "step": 996352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.367834191473457e-05,
+      "loss": 3.9959,
+      "step": 996864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.366995596722405e-05,
+      "loss": 3.9872,
+      "step": 997376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.366157001971353e-05,
+      "loss": 3.9922,
+      "step": 997888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3653200451006736e-05,
+      "loss": 3.9938,
+      "step": 998400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3644814503496216e-05,
+      "loss": 3.9741,
+      "step": 998912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3636428555985696e-05,
+      "loss": 3.976,
+      "step": 999424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.362804260847518e-05,
+      "loss": 3.968,
+      "step": 999936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.361965666096466e-05,
+      "loss": 3.9843,
+      "step": 1000448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.361127071345414e-05,
+      "loss": 3.9836,
+      "step": 1000960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.360290114474735e-05,
+      "loss": 3.9794,
+      "step": 1001472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.359451519723683e-05,
+      "loss": 3.9793,
+      "step": 1001984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.358612924972631e-05,
+      "loss": 4.0055,
+      "step": 1002496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.357774330221579e-05,
+      "loss": 3.9844,
+      "step": 1003008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.356935735470527e-05,
+      "loss": 3.9889,
+      "step": 1003520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.356097140719475e-05,
+      "loss": 3.987,
+      "step": 1004032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.355258545968423e-05,
+      "loss": 3.9914,
+      "step": 1004544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.354419951217371e-05,
+      "loss": 3.9828,
+      "step": 1005056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.353582994346692e-05,
+      "loss": 3.9824,
+      "step": 1005568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.35274439959564e-05,
+      "loss": 3.9782,
+      "step": 1006080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.351905804844588e-05,
+      "loss": 3.9811,
+      "step": 1006592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.351067210093537e-05,
+      "loss": 3.9755,
+      "step": 1007104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.350228615342485e-05,
+      "loss": 3.9835,
+      "step": 1007616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3493916584718057e-05,
+      "loss": 3.9782,
+      "step": 1008128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3485530637207537e-05,
+      "loss": 3.9914,
+      "step": 1008640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3477144689697016e-05,
+      "loss": 3.9866,
+      "step": 1009152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3468758742186496e-05,
+      "loss": 3.9829,
+      "step": 1009664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3460389173479706e-05,
+      "loss": 3.9931,
+      "step": 1010176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3452003225969185e-05,
+      "loss": 3.9883,
+      "step": 1010688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3443617278458665e-05,
+      "loss": 3.9634,
+      "step": 1011200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3435231330948145e-05,
+      "loss": 3.9838,
+      "step": 1011712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.342684538343762e-05,
+      "loss": 3.9753,
+      "step": 1012224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3418459435927105e-05,
+      "loss": 3.9759,
+      "step": 1012736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3410073488416585e-05,
+      "loss": 3.9845,
+      "step": 1013248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.34017039197098e-05,
+      "loss": 3.9862,
+      "step": 1013760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3393317972199274e-05,
+      "loss": 3.98,
+      "step": 1014272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3384932024688754e-05,
+      "loss": 3.9876,
+      "step": 1014784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3376546077178234e-05,
+      "loss": 3.9861,
+      "step": 1015296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3368160129667714e-05,
+      "loss": 3.9867,
+      "step": 1015808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3359774182157194e-05,
+      "loss": 3.9845,
+      "step": 1016320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.33514046134504e-05,
+      "loss": 3.9607,
+      "step": 1016832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.334301866593988e-05,
+      "loss": 3.9824,
+      "step": 1017344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.333463271842936e-05,
+      "loss": 3.987,
+      "step": 1017856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.332624677091884e-05,
+      "loss": 3.9862,
+      "step": 1018368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.331786082340832e-05,
+      "loss": 3.9791,
+      "step": 1018880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.33094748758978e-05,
+      "loss": 3.9748,
+      "step": 1019392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.330110530719102e-05,
+      "loss": 3.969,
+      "step": 1019904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.32927193596805e-05,
+      "loss": 3.9651,
+      "step": 1020416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.328433341216998e-05,
+      "loss": 3.9815,
+      "step": 1020928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.327594746465946e-05,
+      "loss": 3.9616,
+      "step": 1021440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.326756151714894e-05,
+      "loss": 3.9889,
+      "step": 1021952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.325917556963842e-05,
+      "loss": 3.9833,
+      "step": 1022464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.32507896221279e-05,
+      "loss": 3.9632,
+      "step": 1022976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.324240367461738e-05,
+      "loss": 3.9834,
+      "step": 1023488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.323403410591059e-05,
+      "loss": 3.9654,
+      "step": 1024000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.32256645372038e-05,
+      "loss": 3.9719,
+      "step": 1024512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.321727858969328e-05,
+      "loss": 3.9709,
+      "step": 1025024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320889264218276e-05,
+      "loss": 3.9812,
+      "step": 1025536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320050669467224e-05,
+      "loss": 3.9661,
+      "step": 1026048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.319212074716172e-05,
+      "loss": 3.9788,
+      "step": 1026560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.31837347996512e-05,
+      "loss": 3.9628,
+      "step": 1027072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.317534885214068e-05,
+      "loss": 3.9751,
+      "step": 1027584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.316697928343389e-05,
+      "loss": 3.9823,
+      "step": 1028096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.315859333592337e-05,
+      "loss": 3.9804,
+      "step": 1028608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.315020738841285e-05,
+      "loss": 3.9739,
+      "step": 1029120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.314182144090233e-05,
+      "loss": 3.9758,
+      "step": 1029632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3133435493391805e-05,
+      "loss": 3.996,
+      "step": 1030144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3125049545881285e-05,
+      "loss": 3.9639,
+      "step": 1030656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3116663598370765e-05,
+      "loss": 3.9838,
+      "step": 1031168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3108277650860245e-05,
+      "loss": 3.971,
+      "step": 1031680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.309990808215346e-05,
+      "loss": 3.9711,
+      "step": 1032192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.309153851344668e-05,
+      "loss": 3.9734,
+      "step": 1032704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.308315256593616e-05,
+      "loss": 3.9757,
+      "step": 1033216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.307476661842564e-05,
+      "loss": 3.9863,
+      "step": 1033728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.306638067091511e-05,
+      "loss": 3.9889,
+      "step": 1034240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.305799472340459e-05,
+      "loss": 3.9637,
+      "step": 1034752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.304960877589407e-05,
+      "loss": 3.9633,
+      "step": 1035264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.304122282838355e-05,
+      "loss": 3.9686,
+      "step": 1035776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.303283688087303e-05,
+      "loss": 3.9819,
+      "step": 1036288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.302446731216624e-05,
+      "loss": 3.969,
+      "step": 1036800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3016097743459455e-05,
+      "loss": 3.9713,
+      "step": 1037312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.300771179594893e-05,
+      "loss": 3.9784,
+      "step": 1037824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2999342227242144e-05,
+      "loss": 3.9715,
+      "step": 1038336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.299095627973163e-05,
+      "loss": 3.9831,
+      "step": 1038848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.298257033222111e-05,
+      "loss": 3.9652,
+      "step": 1039360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2974184384710584e-05,
+      "loss": 3.9786,
+      "step": 1039872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2965798437200064e-05,
+      "loss": 3.9728,
+      "step": 1040384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2957412489689544e-05,
+      "loss": 3.9893,
+      "step": 1040896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2949026542179024e-05,
+      "loss": 3.9639,
+      "step": 1041408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2940640594668504e-05,
+      "loss": 3.9653,
+      "step": 1041920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2932254647157984e-05,
+      "loss": 3.9796,
+      "step": 1042432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2923868699647464e-05,
+      "loss": 3.9751,
+      "step": 1042944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2915482752136944e-05,
+      "loss": 3.9849,
+      "step": 1043456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2907096804626423e-05,
+      "loss": 3.9698,
+      "step": 1043968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.289872723591963e-05,
+      "loss": 3.9693,
+      "step": 1044480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.289034128840911e-05,
+      "loss": 3.9684,
+      "step": 1044992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.288197171970233e-05,
+      "loss": 3.9705,
+      "step": 1045504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.287360215099554e-05,
+      "loss": 3.9682,
+      "step": 1046016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.286521620348502e-05,
+      "loss": 3.9688,
+      "step": 1046528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.28568302559745e-05,
+      "loss": 3.9737,
+      "step": 1047040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.284844430846398e-05,
+      "loss": 3.9677,
+      "step": 1047552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.284005836095346e-05,
+      "loss": 3.9691,
+      "step": 1048064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.283167241344294e-05,
+      "loss": 3.9709,
+      "step": 1048576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.282328646593242e-05,
+      "loss": 3.9608,
+      "step": 1049088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.28149005184219e-05,
+      "loss": 3.9801,
+      "step": 1049600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.280651457091138e-05,
+      "loss": 3.9717,
+      "step": 1050112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.279812862340086e-05,
+      "loss": 3.9774,
+      "step": 1050624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.278974267589034e-05,
+      "loss": 3.9821,
+      "step": 1051136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.278135672837982e-05,
+      "loss": 3.9787,
+      "step": 1051648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.277300353847676e-05,
+      "loss": 3.9707,
+      "step": 1052160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.276461759096624e-05,
+      "loss": 3.9664,
+      "step": 1052672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.275623164345572e-05,
+      "loss": 3.9767,
+      "step": 1053184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.27478456959452e-05,
+      "loss": 3.9718,
+      "step": 1053696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.273945974843468e-05,
+      "loss": 3.9707,
+      "step": 1054208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.273107380092416e-05,
+      "loss": 3.9787,
+      "step": 1054720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.272268785341364e-05,
+      "loss": 3.9712,
+      "step": 1055232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.271431828470685e-05,
+      "loss": 3.9736,
+      "step": 1055744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.270593233719633e-05,
+      "loss": 3.9663,
+      "step": 1056256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.269754638968581e-05,
+      "loss": 3.967,
+      "step": 1056768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.268916044217529e-05,
+      "loss": 3.9753,
+      "step": 1057280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.268077449466477e-05,
+      "loss": 3.9667,
+      "step": 1057792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.267238854715425e-05,
+      "loss": 3.9787,
+      "step": 1058304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.266400259964373e-05,
+      "loss": 3.982,
+      "step": 1058816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.265561665213321e-05,
+      "loss": 3.9723,
+      "step": 1059328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.264723070462269e-05,
+      "loss": 3.966,
+      "step": 1059840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.26388611359159e-05,
+      "loss": 3.9791,
+      "step": 1060352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2630491567209115e-05,
+      "loss": 3.9748,
+      "step": 1060864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.262210561969859e-05,
+      "loss": 3.9797,
+      "step": 1061376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.261371967218807e-05,
+      "loss": 3.9711,
+      "step": 1061888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.260533372467755e-05,
+      "loss": 3.9704,
+      "step": 1062400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.259694777716703e-05,
+      "loss": 3.9822,
+      "step": 1062912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.258857820846024e-05,
+      "loss": 3.9895,
+      "step": 1063424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2580192260949724e-05,
+      "loss": 3.9666,
+      "step": 1063936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2571806313439204e-05,
+      "loss": 3.9792,
+      "step": 1064448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2563420365928684e-05,
+      "loss": 3.9808,
+      "step": 1064960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2555034418418164e-05,
+      "loss": 3.973,
+      "step": 1065472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2546648470907644e-05,
+      "loss": 3.9735,
+      "step": 1065984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2538262523397124e-05,
+      "loss": 3.971,
+      "step": 1066496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2529876575886604e-05,
+      "loss": 3.9715,
+      "step": 1067008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2521490628376084e-05,
+      "loss": 3.9821,
+      "step": 1067520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.251313743847302e-05,
+      "loss": 3.9611,
+      "step": 1068032
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.0328216552734375,
+      "eval_runtime": 298.7715,
+      "eval_samples_per_second": 1277.2,
+      "eval_steps_per_second": 39.913,
+      "step": 1068480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.25047514909625e-05,
+      "loss": 3.9753,
+      "step": 1068544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.249636554345198e-05,
+      "loss": 3.9591,
+      "step": 1069056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.248797959594146e-05,
+      "loss": 3.9689,
+      "step": 1069568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.247959364843094e-05,
+      "loss": 3.9856,
+      "step": 1070080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.247120770092042e-05,
+      "loss": 3.9798,
+      "step": 1070592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.246282175340991e-05,
+      "loss": 3.9684,
+      "step": 1071104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.245443580589939e-05,
+      "loss": 3.9785,
+      "step": 1071616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.244604985838887e-05,
+      "loss": 3.9661,
+      "step": 1072128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.243766391087835e-05,
+      "loss": 3.9641,
+      "step": 1072640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.242927796336783e-05,
+      "loss": 3.9795,
+      "step": 1073152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.24208920158573e-05,
+      "loss": 3.9753,
+      "step": 1073664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.241250606834678e-05,
+      "loss": 3.9789,
+      "step": 1074176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.240413649964e-05,
+      "loss": 3.9823,
+      "step": 1074688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.239575055212948e-05,
+      "loss": 3.9607,
+      "step": 1075200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.238736460461895e-05,
+      "loss": 3.9617,
+      "step": 1075712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.237897865710843e-05,
+      "loss": 3.9558,
+      "step": 1076224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.237059270959791e-05,
+      "loss": 3.9716,
+      "step": 1076736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.236220676208739e-05,
+      "loss": 3.9684,
+      "step": 1077248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.235382081457688e-05,
+      "loss": 3.9668,
+      "step": 1077760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.234543486706636e-05,
+      "loss": 3.9697,
+      "step": 1078272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.233704891955584e-05,
+      "loss": 3.9874,
+      "step": 1078784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.232866297204532e-05,
+      "loss": 3.9771,
+      "step": 1079296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2320293403338526e-05,
+      "loss": 3.9719,
+      "step": 1079808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2311907455828006e-05,
+      "loss": 3.9739,
+      "step": 1080320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2303521508317486e-05,
+      "loss": 3.9782,
+      "step": 1080832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2295135560806966e-05,
+      "loss": 3.9718,
+      "step": 1081344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2286749613296446e-05,
+      "loss": 3.9637,
+      "step": 1081856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2278380044589655e-05,
+      "loss": 3.966,
+      "step": 1082368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2269994097079135e-05,
+      "loss": 3.9645,
+      "step": 1082880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2261608149568615e-05,
+      "loss": 3.9676,
+      "step": 1083392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2253222202058095e-05,
+      "loss": 3.9675,
+      "step": 1083904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2244836254547575e-05,
+      "loss": 3.9695,
+      "step": 1084416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2236450307037055e-05,
+      "loss": 3.9723,
+      "step": 1084928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.222806435952654e-05,
+      "loss": 3.9756,
+      "step": 1085440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2219678412016015e-05,
+      "loss": 3.9697,
+      "step": 1085952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.221132522211296e-05,
+      "loss": 3.9768,
+      "step": 1086464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.220293927460244e-05,
+      "loss": 3.9771,
+      "step": 1086976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.219455332709192e-05,
+      "loss": 3.9509,
+      "step": 1087488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.21861673795814e-05,
+      "loss": 3.9751,
+      "step": 1088000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.217778143207088e-05,
+      "loss": 3.9555,
+      "step": 1088512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.216939548456036e-05,
+      "loss": 3.9659,
+      "step": 1089024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.216100953704984e-05,
+      "loss": 3.9671,
+      "step": 1089536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.215263996834305e-05,
+      "loss": 3.9774,
+      "step": 1090048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.214425402083253e-05,
+      "loss": 3.9655,
+      "step": 1090560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.213586807332201e-05,
+      "loss": 3.9691,
+      "step": 1091072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.212748212581149e-05,
+      "loss": 3.9782,
+      "step": 1091584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.211909617830097e-05,
+      "loss": 3.971,
+      "step": 1092096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2110726609594184e-05,
+      "loss": 3.9744,
+      "step": 1092608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2102340662083664e-05,
+      "loss": 3.9466,
+      "step": 1093120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.209395471457314e-05,
+      "loss": 3.9669,
+      "step": 1093632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.208556876706262e-05,
+      "loss": 3.9759,
+      "step": 1094144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.20771828195521e-05,
+      "loss": 3.9746,
+      "step": 1094656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.206879687204158e-05,
+      "loss": 3.9685,
+      "step": 1095168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.206041092453106e-05,
+      "loss": 3.959,
+      "step": 1095680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2052041355824267e-05,
+      "loss": 3.9613,
+      "step": 1096192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.204367178711748e-05,
+      "loss": 3.9478,
+      "step": 1096704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.203528583960696e-05,
+      "loss": 3.975,
+      "step": 1097216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.202689989209644e-05,
+      "loss": 3.9474,
+      "step": 1097728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.201851394458592e-05,
+      "loss": 3.9746,
+      "step": 1098240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.20101279970754e-05,
+      "loss": 3.9689,
+      "step": 1098752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.200174204956488e-05,
+      "loss": 3.9551,
+      "step": 1099264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.199335610205436e-05,
+      "loss": 3.9671,
+      "step": 1099776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.198497015454384e-05,
+      "loss": 3.9537,
+      "step": 1100288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.197658420703332e-05,
+      "loss": 3.9585,
+      "step": 1100800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.19681982595228e-05,
+      "loss": 3.957,
+      "step": 1101312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.195981231201228e-05,
+      "loss": 3.9673,
+      "step": 1101824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.195142636450176e-05,
+      "loss": 3.9527,
+      "step": 1102336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.194304041699124e-05,
+      "loss": 3.9637,
+      "step": 1102848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.193467084828445e-05,
+      "loss": 3.9514,
+      "step": 1103360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.192628490077393e-05,
+      "loss": 3.9591,
+      "step": 1103872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.191789895326342e-05,
+      "loss": 3.9731,
+      "step": 1104384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.19095130057529e-05,
+      "loss": 3.9688,
+      "step": 1104896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.190114343704611e-05,
+      "loss": 3.9614,
+      "step": 1105408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.189275748953559e-05,
+      "loss": 3.9662,
+      "step": 1105920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1884371542025067e-05,
+      "loss": 3.9799,
+      "step": 1106432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1876001973318276e-05,
+      "loss": 3.9534,
+      "step": 1106944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1867616025807756e-05,
+      "loss": 3.9709,
+      "step": 1107456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1859230078297236e-05,
+      "loss": 3.9591,
+      "step": 1107968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1850844130786716e-05,
+      "loss": 3.9636,
+      "step": 1108480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1842458183276196e-05,
+      "loss": 3.9585,
+      "step": 1108992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1834088614569405e-05,
+      "loss": 3.9616,
+      "step": 1109504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1825702667058885e-05,
+      "loss": 3.9747,
+      "step": 1110016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1817316719548365e-05,
+      "loss": 3.9749,
+      "step": 1110528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.180893077203785e-05,
+      "loss": 3.9587,
+      "step": 1111040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.180056120333106e-05,
+      "loss": 3.9444,
+      "step": 1111552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.179217525582054e-05,
+      "loss": 3.9542,
+      "step": 1112064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.178378930831002e-05,
+      "loss": 3.9679,
+      "step": 1112576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.17754033607995e-05,
+      "loss": 3.957,
+      "step": 1113088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.176703379209271e-05,
+      "loss": 3.9579,
+      "step": 1113600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.175866422338592e-05,
+      "loss": 3.9673,
+      "step": 1114112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.17502782758754e-05,
+      "loss": 3.9617,
+      "step": 1114624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.174189232836488e-05,
+      "loss": 3.9642,
+      "step": 1115136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.173350638085436e-05,
+      "loss": 3.9556,
+      "step": 1115648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.172512043334384e-05,
+      "loss": 3.9676,
+      "step": 1116160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.171673448583332e-05,
+      "loss": 3.9562,
+      "step": 1116672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.17083485383228e-05,
+      "loss": 3.975,
+      "step": 1117184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.169996259081228e-05,
+      "loss": 3.9543,
+      "step": 1117696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1691593022105494e-05,
+      "loss": 3.9534,
+      "step": 1118208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.16832234533987e-05,
+      "loss": 3.9688,
+      "step": 1118720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.167483750588818e-05,
+      "loss": 3.9615,
+      "step": 1119232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.166645155837766e-05,
+      "loss": 3.971,
+      "step": 1119744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.165806561086714e-05,
+      "loss": 3.9597,
+      "step": 1120256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1649679663356616e-05,
+      "loss": 3.9592,
+      "step": 1120768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1641293715846096e-05,
+      "loss": 3.9557,
+      "step": 1121280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.163292414713931e-05,
+      "loss": 3.9604,
+      "step": 1121792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.162453819962879e-05,
+      "loss": 3.9516,
+      "step": 1122304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.161615225211827e-05,
+      "loss": 3.9578,
+      "step": 1122816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.160776630460775e-05,
+      "loss": 3.961,
+      "step": 1123328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.159938035709723e-05,
+      "loss": 3.9546,
+      "step": 1123840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.159099440958671e-05,
+      "loss": 3.9562,
+      "step": 1124352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.158260846207619e-05,
+      "loss": 3.9642,
+      "step": 1124864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.157422251456567e-05,
+      "loss": 3.947,
+      "step": 1125376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.156583656705515e-05,
+      "loss": 3.9663,
+      "step": 1125888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.155748337715209e-05,
+      "loss": 3.9586,
+      "step": 1126400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.154909742964157e-05,
+      "loss": 3.9646,
+      "step": 1126912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.154071148213105e-05,
+      "loss": 3.9683,
+      "step": 1127424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.153232553462053e-05,
+      "loss": 3.9697,
+      "step": 1127936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1523955965913746e-05,
+      "loss": 3.9605,
+      "step": 1128448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1515570018403226e-05,
+      "loss": 3.9573,
+      "step": 1128960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1507184070892705e-05,
+      "loss": 3.962,
+      "step": 1129472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1498798123382185e-05,
+      "loss": 3.9618,
+      "step": 1129984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1490428554675395e-05,
+      "loss": 3.9548,
+      "step": 1130496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1482042607164874e-05,
+      "loss": 3.9702,
+      "step": 1131008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1473656659654354e-05,
+      "loss": 3.9559,
+      "step": 1131520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1465270712143834e-05,
+      "loss": 3.9635,
+      "step": 1132032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1456884764633314e-05,
+      "loss": 3.9549,
+      "step": 1132544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1448498817122794e-05,
+      "loss": 3.951,
+      "step": 1133056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1440112869612274e-05,
+      "loss": 3.9672,
+      "step": 1133568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.143174330090548e-05,
+      "loss": 3.9557,
+      "step": 1134080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.14233737321987e-05,
+      "loss": 3.9593,
+      "step": 1134592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.141498778468818e-05,
+      "loss": 3.9712,
+      "step": 1135104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.140660183717766e-05,
+      "loss": 3.9581,
+      "step": 1135616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.139821588966714e-05,
+      "loss": 3.9572,
+      "step": 1136128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.138982994215662e-05,
+      "loss": 3.9667,
+      "step": 1136640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.13814439946461e-05,
+      "loss": 3.9658,
+      "step": 1137152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.137305804713558e-05,
+      "loss": 3.967,
+      "step": 1137664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.136467209962506e-05,
+      "loss": 3.9541,
+      "step": 1138176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.135628615211454e-05,
+      "loss": 3.9596,
+      "step": 1138688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.134790020460402e-05,
+      "loss": 3.9732,
+      "step": 1139200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.13395142570935e-05,
+      "loss": 3.9756,
+      "step": 1139712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.133114468838671e-05,
+      "loss": 3.9595,
+      "step": 1140224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.132275874087619e-05,
+      "loss": 3.9627,
+      "step": 1140736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1314389172169404e-05,
+      "loss": 3.9703,
+      "step": 1141248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1306003224658884e-05,
+      "loss": 3.9614,
+      "step": 1141760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1297617277148364e-05,
+      "loss": 3.9608,
+      "step": 1142272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1289231329637844e-05,
+      "loss": 3.9597,
+      "step": 1142784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1280845382127324e-05,
+      "loss": 3.958,
+      "step": 1143296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1272459434616803e-05,
+      "loss": 3.9707,
+      "step": 1143808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1264073487106283e-05,
+      "loss": 3.95,
+      "step": 1144320
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.026025772094727,
+      "eval_runtime": 290.3356,
+      "eval_samples_per_second": 1314.31,
+      "eval_steps_per_second": 41.073,
+      "step": 1144800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125568753959576e-05,
+      "loss": 3.93,
+      "step": 1144832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1247301592085237e-05,
+      "loss": 3.9512,
+      "step": 1145344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1238915644574717e-05,
+      "loss": 3.9532,
+      "step": 1145856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1230529697064196e-05,
+      "loss": 3.9754,
+      "step": 1146368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1222143749553676e-05,
+      "loss": 3.968,
+      "step": 1146880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1213757802043156e-05,
+      "loss": 3.9567,
+      "step": 1147392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1205371854532636e-05,
+      "loss": 3.965,
+      "step": 1147904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1196985907022116e-05,
+      "loss": 3.9558,
+      "step": 1148416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1188599959511596e-05,
+      "loss": 3.9554,
+      "step": 1148928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.118021401200108e-05,
+      "loss": 3.9657,
+      "step": 1149440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.117182806449056e-05,
+      "loss": 3.9602,
+      "step": 1149952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.116344211698004e-05,
+      "loss": 3.9666,
+      "step": 1150464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.115507254827325e-05,
+      "loss": 3.9711,
+      "step": 1150976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.114668660076273e-05,
+      "loss": 3.952,
+      "step": 1151488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.113830065325221e-05,
+      "loss": 3.9486,
+      "step": 1152000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.112991470574169e-05,
+      "loss": 3.9434,
+      "step": 1152512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1121528758231165e-05,
+      "loss": 3.9575,
+      "step": 1153024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1113142810720645e-05,
+      "loss": 3.9604,
+      "step": 1153536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1104756863210125e-05,
+      "loss": 3.9539,
+      "step": 1154048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1096370915699605e-05,
+      "loss": 3.9568,
+      "step": 1154560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1087984968189085e-05,
+      "loss": 3.9739,
+      "step": 1155072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1079599020678565e-05,
+      "loss": 3.9606,
+      "step": 1155584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.107121307316805e-05,
+      "loss": 3.9638,
+      "step": 1156096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.106282712565753e-05,
+      "loss": 3.9576,
+      "step": 1156608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.105444117814701e-05,
+      "loss": 3.9702,
+      "step": 1157120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.104607160944022e-05,
+      "loss": 3.9566,
+      "step": 1157632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.103770204073343e-05,
+      "loss": 3.9571,
+      "step": 1158144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.102931609322291e-05,
+      "loss": 3.9528,
+      "step": 1158656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.102093014571239e-05,
+      "loss": 3.9535,
+      "step": 1159168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.101254419820187e-05,
+      "loss": 3.9543,
+      "step": 1159680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.100415825069135e-05,
+      "loss": 3.9532,
+      "step": 1160192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.099577230318083e-05,
+      "loss": 3.9581,
+      "step": 1160704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.098740273447404e-05,
+      "loss": 3.9611,
+      "step": 1161216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.097901678696352e-05,
+      "loss": 3.9624,
+      "step": 1161728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0970630839453005e-05,
+      "loss": 3.9628,
+      "step": 1162240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0962261270746214e-05,
+      "loss": 3.9613,
+      "step": 1162752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0953875323235694e-05,
+      "loss": 3.9653,
+      "step": 1163264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0945489375725174e-05,
+      "loss": 3.9424,
+      "step": 1163776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0937103428214654e-05,
+      "loss": 3.9638,
+      "step": 1164288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0928717480704134e-05,
+      "loss": 3.94,
+      "step": 1164800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0920331533193614e-05,
+      "loss": 3.9568,
+      "step": 1165312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0911945585683094e-05,
+      "loss": 3.956,
+      "step": 1165824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0903559638172574e-05,
+      "loss": 3.9653,
+      "step": 1166336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0895173690662054e-05,
+      "loss": 3.9523,
+      "step": 1166848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.088680412195526e-05,
+      "loss": 3.9621,
+      "step": 1167360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.087841817444474e-05,
+      "loss": 3.9635,
+      "step": 1167872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.087003222693422e-05,
+      "loss": 3.9616,
+      "step": 1168384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.086166265822744e-05,
+      "loss": 3.959,
+      "step": 1168896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.085327671071692e-05,
+      "loss": 3.9371,
+      "step": 1169408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.08448907632064e-05,
+      "loss": 3.9523,
+      "step": 1169920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.083650481569588e-05,
+      "loss": 3.9646,
+      "step": 1170432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.082811886818536e-05,
+      "loss": 3.9608,
+      "step": 1170944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.081973292067483e-05,
+      "loss": 3.9562,
+      "step": 1171456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.081134697316431e-05,
+      "loss": 3.9471,
+      "step": 1171968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.080296102565379e-05,
+      "loss": 3.9486,
+      "step": 1172480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.079457507814327e-05,
+      "loss": 3.9356,
+      "step": 1172992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.078618913063275e-05,
+      "loss": 3.9651,
+      "step": 1173504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.077781956192596e-05,
+      "loss": 3.9338,
+      "step": 1174016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.076943361441544e-05,
+      "loss": 3.9597,
+      "step": 1174528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.076104766690493e-05,
+      "loss": 3.9561,
+      "step": 1175040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.075266171939441e-05,
+      "loss": 3.9463,
+      "step": 1175552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.074427577188389e-05,
+      "loss": 3.9526,
+      "step": 1176064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.073588982437337e-05,
+      "loss": 3.941,
+      "step": 1176576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0727520255666576e-05,
+      "loss": 3.9472,
+      "step": 1177088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0719134308156056e-05,
+      "loss": 3.9385,
+      "step": 1177600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0710748360645536e-05,
+      "loss": 3.9613,
+      "step": 1178112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0702362413135016e-05,
+      "loss": 3.9469,
+      "step": 1178624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0693976465624496e-05,
+      "loss": 3.9454,
+      "step": 1179136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0685590518113976e-05,
+      "loss": 3.941,
+      "step": 1179648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0677204570603456e-05,
+      "loss": 3.9487,
+      "step": 1180160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0668835001896665e-05,
+      "loss": 3.9642,
+      "step": 1180672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0660449054386145e-05,
+      "loss": 3.9553,
+      "step": 1181184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0652063106875625e-05,
+      "loss": 3.9452,
+      "step": 1181696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0643677159365105e-05,
+      "loss": 3.9579,
+      "step": 1182208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.063529121185459e-05,
+      "loss": 3.9665,
+      "step": 1182720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.062690526434407e-05,
+      "loss": 3.9408,
+      "step": 1183232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.061855207444101e-05,
+      "loss": 3.959,
+      "step": 1183744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.061016612693049e-05,
+      "loss": 3.9477,
+      "step": 1184256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.060178017941997e-05,
+      "loss": 3.9512,
+      "step": 1184768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.059339423190945e-05,
+      "loss": 3.9463,
+      "step": 1185280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.058500828439893e-05,
+      "loss": 3.9513,
+      "step": 1185792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.057662233688841e-05,
+      "loss": 3.9619,
+      "step": 1186304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.056823638937789e-05,
+      "loss": 3.9609,
+      "step": 1186816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.055985044186736e-05,
+      "loss": 3.9492,
+      "step": 1187328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.055146449435684e-05,
+      "loss": 3.9276,
+      "step": 1187840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.054307854684633e-05,
+      "loss": 3.948,
+      "step": 1188352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.053469259933581e-05,
+      "loss": 3.9524,
+      "step": 1188864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.052630665182529e-05,
+      "loss": 3.9515,
+      "step": 1189376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.051792070431477e-05,
+      "loss": 3.9414,
+      "step": 1189888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.050955113560798e-05,
+      "loss": 3.9579,
+      "step": 1190400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0501181566901188e-05,
+      "loss": 3.9515,
+      "step": 1190912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0492795619390668e-05,
+      "loss": 3.9535,
+      "step": 1191424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0484409671880148e-05,
+      "loss": 3.9414,
+      "step": 1191936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0476023724369628e-05,
+      "loss": 3.957,
+      "step": 1192448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.046763777685911e-05,
+      "loss": 3.9427,
+      "step": 1192960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.045926820815232e-05,
+      "loss": 3.9639,
+      "step": 1193472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.04508822606418e-05,
+      "loss": 3.9485,
+      "step": 1193984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.044249631313128e-05,
+      "loss": 3.9394,
+      "step": 1194496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.043411036562076e-05,
+      "loss": 3.9524,
+      "step": 1195008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0425740796913972e-05,
+      "loss": 3.9523,
+      "step": 1195520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0417354849403452e-05,
+      "loss": 3.9575,
+      "step": 1196032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0408968901892932e-05,
+      "loss": 3.9479,
+      "step": 1196544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0400582954382412e-05,
+      "loss": 3.9463,
+      "step": 1197056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0392197006871892e-05,
+      "loss": 3.9434,
+      "step": 1197568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0383811059361372e-05,
+      "loss": 3.9515,
+      "step": 1198080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0375425111850852e-05,
+      "loss": 3.9367,
+      "step": 1198592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0367039164340332e-05,
+      "loss": 3.9487,
+      "step": 1199104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0358669595633545e-05,
+      "loss": 3.9494,
+      "step": 1199616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0350283648123024e-05,
+      "loss": 3.9472,
+      "step": 1200128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0341897700612504e-05,
+      "loss": 3.9406,
+      "step": 1200640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0333511753101984e-05,
+      "loss": 3.95,
+      "step": 1201152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0325125805591464e-05,
+      "loss": 3.9389,
+      "step": 1201664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0316739858080944e-05,
+      "loss": 3.9535,
+      "step": 1202176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0308370289374153e-05,
+      "loss": 3.9472,
+      "step": 1202688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0299984341863637e-05,
+      "loss": 3.95,
+      "step": 1203200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0291598394353117e-05,
+      "loss": 3.957,
+      "step": 1203712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0283212446842597e-05,
+      "loss": 3.9565,
+      "step": 1204224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0274859256939535e-05,
+      "loss": 3.952,
+      "step": 1204736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0266473309429015e-05,
+      "loss": 3.9423,
+      "step": 1205248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0258087361918498e-05,
+      "loss": 3.9505,
+      "step": 1205760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0249701414407978e-05,
+      "loss": 3.9512,
+      "step": 1206272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0241315466897458e-05,
+      "loss": 3.9447,
+      "step": 1206784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0232929519386938e-05,
+      "loss": 3.9556,
+      "step": 1207296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0224543571876418e-05,
+      "loss": 3.9471,
+      "step": 1207808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0216157624365898e-05,
+      "loss": 3.9531,
+      "step": 1208320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0207771676855378e-05,
+      "loss": 3.9412,
+      "step": 1208832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.019940210814859e-05,
+      "loss": 3.9434,
+      "step": 1209344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.019101616063807e-05,
+      "loss": 3.9503,
+      "step": 1209856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.018263021312755e-05,
+      "loss": 3.9439,
+      "step": 1210368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0174244265617024e-05,
+      "loss": 3.9434,
+      "step": 1210880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.016587469691024e-05,
+      "loss": 3.9635,
+      "step": 1211392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.015748874939972e-05,
+      "loss": 3.9507,
+      "step": 1211904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.01491028018892e-05,
+      "loss": 3.9447,
+      "step": 1212416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0140716854378676e-05,
+      "loss": 3.9556,
+      "step": 1212928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0132347285671892e-05,
+      "loss": 3.9519,
+      "step": 1213440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0123961338161372e-05,
+      "loss": 3.9527,
+      "step": 1213952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.011557539065085e-05,
+      "loss": 3.9493,
+      "step": 1214464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0107189443140328e-05,
+      "loss": 3.945,
+      "step": 1214976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0098803495629808e-05,
+      "loss": 3.9607,
+      "step": 1215488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0090433926923024e-05,
+      "loss": 3.9624,
+      "step": 1216000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0082064358216233e-05,
+      "loss": 3.949,
+      "step": 1216512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0073678410705713e-05,
+      "loss": 3.9487,
+      "step": 1217024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0065292463195193e-05,
+      "loss": 3.9593,
+      "step": 1217536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0056906515684673e-05,
+      "loss": 3.9457,
+      "step": 1218048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.004852056817415e-05,
+      "loss": 3.9509,
+      "step": 1218560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0040150999467365e-05,
+      "loss": 3.9494,
+      "step": 1219072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0031765051956845e-05,
+      "loss": 3.9463,
+      "step": 1219584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0023379104446325e-05,
+      "loss": 3.9593,
+      "step": 1220096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.00149931569358e-05,
+      "loss": 3.9401,
+      "step": 1220608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0006607209425282e-05,
+      "loss": 3.944,
+      "step": 1221120
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.020059108734131,
+      "eval_runtime": 289.8585,
+      "eval_samples_per_second": 1316.473,
+      "eval_steps_per_second": 41.141,
+      "step": 1221120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9998237640718498e-05,
+      "loss": 3.9374,
+      "step": 1221632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.998985169320797e-05,
+      "loss": 3.9411,
+      "step": 1222144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.998146574569745e-05,
+      "loss": 3.9627,
+      "step": 1222656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.997307979818693e-05,
+      "loss": 3.9582,
+      "step": 1223168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.996469385067641e-05,
+      "loss": 3.9448,
+      "step": 1223680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.995630790316589e-05,
+      "loss": 3.9547,
+      "step": 1224192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9947921955655374e-05,
+      "loss": 3.9452,
+      "step": 1224704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9939536008144854e-05,
+      "loss": 3.9411,
+      "step": 1225216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9931166439438063e-05,
+      "loss": 3.9564,
+      "step": 1225728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9922780491927543e-05,
+      "loss": 3.947,
+      "step": 1226240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9914394544417023e-05,
+      "loss": 3.9579,
+      "step": 1226752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9906008596906503e-05,
+      "loss": 3.9549,
+      "step": 1227264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9897639028199716e-05,
+      "loss": 3.9483,
+      "step": 1227776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9889253080689195e-05,
+      "loss": 3.9384,
+      "step": 1228288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9880867133178675e-05,
+      "loss": 3.9302,
+      "step": 1228800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9872481185668155e-05,
+      "loss": 3.9436,
+      "step": 1229312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9864095238157635e-05,
+      "loss": 3.9494,
+      "step": 1229824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9855709290647115e-05,
+      "loss": 3.9437,
+      "step": 1230336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9847323343136595e-05,
+      "loss": 3.9446,
+      "step": 1230848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9838970153233537e-05,
+      "loss": 3.9675,
+      "step": 1231360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9830584205723017e-05,
+      "loss": 3.9506,
+      "step": 1231872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9822198258212497e-05,
+      "loss": 3.9517,
+      "step": 1232384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9813812310701977e-05,
+      "loss": 3.9483,
+      "step": 1232896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9805426363191457e-05,
+      "loss": 3.9622,
+      "step": 1233408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.979705679448467e-05,
+      "loss": 3.9402,
+      "step": 1233920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.978867084697415e-05,
+      "loss": 3.947,
+      "step": 1234432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.978028489946363e-05,
+      "loss": 3.9392,
+      "step": 1234944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.977189895195311e-05,
+      "loss": 3.9436,
+      "step": 1235456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9763529383246318e-05,
+      "loss": 3.945,
+      "step": 1235968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9755143435735798e-05,
+      "loss": 3.942,
+      "step": 1236480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.974675748822528e-05,
+      "loss": 3.9468,
+      "step": 1236992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.973837154071476e-05,
+      "loss": 3.9515,
+      "step": 1237504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.972998559320424e-05,
+      "loss": 3.9464,
+      "step": 1238016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.972159964569372e-05,
+      "loss": 3.9557,
+      "step": 1238528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.971323007698693e-05,
+      "loss": 3.9512,
+      "step": 1239040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.970484412947641e-05,
+      "loss": 3.9555,
+      "step": 1239552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.969645818196589e-05,
+      "loss": 3.9315,
+      "step": 1240064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.968807223445537e-05,
+      "loss": 3.9536,
+      "step": 1240576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9679686286944854e-05,
+      "loss": 3.9297,
+      "step": 1241088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9671300339434334e-05,
+      "loss": 3.9467,
+      "step": 1241600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9662914391923807e-05,
+      "loss": 3.9426,
+      "step": 1242112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9654528444413287e-05,
+      "loss": 3.9527,
+      "step": 1242624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9646158875706503e-05,
+      "loss": 3.9378,
+      "step": 1243136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9637789306999715e-05,
+      "loss": 3.9531,
+      "step": 1243648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9629403359489195e-05,
+      "loss": 3.9487,
+      "step": 1244160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9621017411978675e-05,
+      "loss": 3.9514,
+      "step": 1244672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9612631464468155e-05,
+      "loss": 3.9494,
+      "step": 1245184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9604245516957628e-05,
+      "loss": 3.9262,
+      "step": 1245696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9595859569447108e-05,
+      "loss": 3.9445,
+      "step": 1246208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.958747362193659e-05,
+      "loss": 3.9529,
+      "step": 1246720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9579104053229807e-05,
+      "loss": 3.9468,
+      "step": 1247232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.957071810571928e-05,
+      "loss": 3.9476,
+      "step": 1247744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.956233215820876e-05,
+      "loss": 3.9349,
+      "step": 1248256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.955394621069824e-05,
+      "loss": 3.9416,
+      "step": 1248768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.954556026318772e-05,
+      "loss": 3.9236,
+      "step": 1249280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.95371743156772e-05,
+      "loss": 3.9553,
+      "step": 1249792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9528788368166684e-05,
+      "loss": 3.9231,
+      "step": 1250304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9520402420656164e-05,
+      "loss": 3.9511,
+      "step": 1250816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9512016473145644e-05,
+      "loss": 3.9458,
+      "step": 1251328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9503646904438853e-05,
+      "loss": 3.9335,
+      "step": 1251840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9495260956928333e-05,
+      "loss": 3.9393,
+      "step": 1252352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9486875009417813e-05,
+      "loss": 3.9376,
+      "step": 1252864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9478489061907292e-05,
+      "loss": 3.9366,
+      "step": 1253376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9470103114396776e-05,
+      "loss": 3.9286,
+      "step": 1253888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9461717166886256e-05,
+      "loss": 3.9499,
+      "step": 1254400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9453331219375736e-05,
+      "loss": 3.9365,
+      "step": 1254912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9444961650668945e-05,
+      "loss": 3.9337,
+      "step": 1255424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9436575703158425e-05,
+      "loss": 3.9322,
+      "step": 1255936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9428189755647905e-05,
+      "loss": 3.9378,
+      "step": 1256448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9419820186941117e-05,
+      "loss": 3.9516,
+      "step": 1256960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9411450618234326e-05,
+      "loss": 3.9471,
+      "step": 1257472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9403064670723806e-05,
+      "loss": 3.934,
+      "step": 1257984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9394678723213286e-05,
+      "loss": 3.9479,
+      "step": 1258496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9386292775702766e-05,
+      "loss": 3.9526,
+      "step": 1259008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9377906828192246e-05,
+      "loss": 3.9318,
+      "step": 1259520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.936953725948546e-05,
+      "loss": 3.9458,
+      "step": 1260032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.936115131197494e-05,
+      "loss": 3.9398,
+      "step": 1260544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.935276536446442e-05,
+      "loss": 3.9419,
+      "step": 1261056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.93443794169539e-05,
+      "loss": 3.9354,
+      "step": 1261568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.933599346944338e-05,
+      "loss": 3.938,
+      "step": 1262080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.932760752193286e-05,
+      "loss": 3.9488,
+      "step": 1262592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.931922157442234e-05,
+      "loss": 3.9573,
+      "step": 1263104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9310835626911815e-05,
+      "loss": 3.9344,
+      "step": 1263616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9302449679401295e-05,
+      "loss": 3.9176,
+      "step": 1264128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.929408011069451e-05,
+      "loss": 3.9389,
+      "step": 1264640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.928569416318399e-05,
+      "loss": 3.9412,
+      "step": 1265152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9277308215673467e-05,
+      "loss": 3.9402,
+      "step": 1265664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9268922268162947e-05,
+      "loss": 3.9355,
+      "step": 1266176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9260552699456163e-05,
+      "loss": 3.9449,
+      "step": 1266688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9252166751945643e-05,
+      "loss": 3.941,
+      "step": 1267200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9243780804435116e-05,
+      "loss": 3.9398,
+      "step": 1267712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9235394856924596e-05,
+      "loss": 3.9309,
+      "step": 1268224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9227008909414076e-05,
+      "loss": 3.9428,
+      "step": 1268736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.921862296190356e-05,
+      "loss": 3.934,
+      "step": 1269248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.921023701439304e-05,
+      "loss": 3.9505,
+      "step": 1269760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.920185106688252e-05,
+      "loss": 3.9411,
+      "step": 1270272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9193465119372e-05,
+      "loss": 3.9266,
+      "step": 1270784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.918509555066521e-05,
+      "loss": 3.9401,
+      "step": 1271296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.917670960315469e-05,
+      "loss": 3.9432,
+      "step": 1271808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.916832365564417e-05,
+      "loss": 3.95,
+      "step": 1272320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.915995408693738e-05,
+      "loss": 3.9368,
+      "step": 1272832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.915156813942686e-05,
+      "loss": 3.9368,
+      "step": 1273344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.914319857072007e-05,
+      "loss": 3.9298,
+      "step": 1273856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.913481262320955e-05,
+      "loss": 3.9438,
+      "step": 1274368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.912642667569903e-05,
+      "loss": 3.9272,
+      "step": 1274880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.911804072818851e-05,
+      "loss": 3.9368,
+      "step": 1275392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9109654780677993e-05,
+      "loss": 3.9389,
+      "step": 1275904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9101268833167473e-05,
+      "loss": 3.9389,
+      "step": 1276416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9092882885656953e-05,
+      "loss": 3.9297,
+      "step": 1276928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9084496938146433e-05,
+      "loss": 3.9384,
+      "step": 1277440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9076110990635913e-05,
+      "loss": 3.934,
+      "step": 1277952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9067725043125393e-05,
+      "loss": 3.9428,
+      "step": 1278464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9059339095614873e-05,
+      "loss": 3.934,
+      "step": 1278976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9050953148104353e-05,
+      "loss": 3.9362,
+      "step": 1279488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9042583579397565e-05,
+      "loss": 3.9504,
+      "step": 1280000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9034197631887045e-05,
+      "loss": 3.947,
+      "step": 1280512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9025828063180254e-05,
+      "loss": 3.9416,
+      "step": 1281024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9017442115669734e-05,
+      "loss": 3.9341,
+      "step": 1281536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9009056168159214e-05,
+      "loss": 3.9397,
+      "step": 1282048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9000670220648694e-05,
+      "loss": 3.936,
+      "step": 1282560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8992284273138178e-05,
+      "loss": 3.936,
+      "step": 1283072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.898389832562765e-05,
+      "loss": 3.9481,
+      "step": 1283584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.897551237811713e-05,
+      "loss": 3.9328,
+      "step": 1284096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.896712643060661e-05,
+      "loss": 3.9437,
+      "step": 1284608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.895874048309609e-05,
+      "loss": 3.9286,
+      "step": 1285120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.895035453558557e-05,
+      "loss": 3.9324,
+      "step": 1285632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8941968588075054e-05,
+      "loss": 3.9409,
+      "step": 1286144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8933599019368263e-05,
+      "loss": 3.934,
+      "step": 1286656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8925213071857743e-05,
+      "loss": 3.9311,
+      "step": 1287168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8916827124347223e-05,
+      "loss": 3.955,
+      "step": 1287680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8908441176836703e-05,
+      "loss": 3.9424,
+      "step": 1288192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8900055229326183e-05,
+      "loss": 3.9369,
+      "step": 1288704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8891669281815663e-05,
+      "loss": 3.9378,
+      "step": 1289216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8883299713108875e-05,
+      "loss": 3.9402,
+      "step": 1289728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8874913765598355e-05,
+      "loss": 3.9438,
+      "step": 1290240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8866527818087835e-05,
+      "loss": 3.9413,
+      "step": 1290752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8858141870577315e-05,
+      "loss": 3.9377,
+      "step": 1291264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8849755923066795e-05,
+      "loss": 3.9521,
+      "step": 1291776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8841369975556275e-05,
+      "loss": 3.9458,
+      "step": 1292288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8832984028045755e-05,
+      "loss": 3.9427,
+      "step": 1292800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8824614459338968e-05,
+      "loss": 3.9388,
+      "step": 1293312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8816244890632177e-05,
+      "loss": 3.9458,
+      "step": 1293824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8807858943121657e-05,
+      "loss": 3.9385,
+      "step": 1294336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8799472995611137e-05,
+      "loss": 3.9409,
+      "step": 1294848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8791087048100616e-05,
+      "loss": 3.9394,
+      "step": 1295360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.87827011005901e-05,
+      "loss": 3.9391,
+      "step": 1295872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.877431515307958e-05,
+      "loss": 3.946,
+      "step": 1296384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.876592920556906e-05,
+      "loss": 3.9302,
+      "step": 1296896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.875754325805854e-05,
+      "loss": 3.9343,
+      "step": 1297408
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.015102863311768,
+      "eval_runtime": 289.246,
+      "eval_samples_per_second": 1319.261,
+      "eval_steps_per_second": 41.228,
+      "step": 1297440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8749157310548013e-05,
+      "loss": 3.9284,
+      "step": 1297920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8740771363037493e-05,
+      "loss": 3.9324,
+      "step": 1298432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8732385415526976e-05,
+      "loss": 3.9548,
+      "step": 1298944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8723999468016456e-05,
+      "loss": 3.9451,
+      "step": 1299456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8715613520505936e-05,
+      "loss": 3.9396,
+      "step": 1299968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8707227572995416e-05,
+      "loss": 3.9401,
+      "step": 1300480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8698841625484896e-05,
+      "loss": 3.9367,
+      "step": 1300992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8690455677974376e-05,
+      "loss": 3.9299,
+      "step": 1301504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8682069730463856e-05,
+      "loss": 3.9432,
+      "step": 1302016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8673683782953336e-05,
+      "loss": 3.9422,
+      "step": 1302528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8665297835442816e-05,
+      "loss": 3.9448,
+      "step": 1303040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8656911887932296e-05,
+      "loss": 3.9472,
+      "step": 1303552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8648542319225508e-05,
+      "loss": 3.9354,
+      "step": 1304064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8640156371714988e-05,
+      "loss": 3.9301,
+      "step": 1304576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8631770424204468e-05,
+      "loss": 3.9258,
+      "step": 1305088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8623384476693948e-05,
+      "loss": 3.9262,
+      "step": 1305600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8614998529183428e-05,
+      "loss": 3.9395,
+      "step": 1306112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8606612581672905e-05,
+      "loss": 3.9315,
+      "step": 1306624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8598226634162385e-05,
+      "loss": 3.9371,
+      "step": 1307136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8589840686651865e-05,
+      "loss": 3.9591,
+      "step": 1307648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8581454739141344e-05,
+      "loss": 3.937,
+      "step": 1308160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8573068791630824e-05,
+      "loss": 3.9431,
+      "step": 1308672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8564682844120304e-05,
+      "loss": 3.9357,
+      "step": 1309184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8556313275413517e-05,
+      "loss": 3.9484,
+      "step": 1309696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8547927327902997e-05,
+      "loss": 3.9326,
+      "step": 1310208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8539541380392477e-05,
+      "loss": 3.9348,
+      "step": 1310720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8531155432881957e-05,
+      "loss": 3.9285,
+      "step": 1311232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8522785864175166e-05,
+      "loss": 3.9374,
+      "step": 1311744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8514399916664646e-05,
+      "loss": 3.9319,
+      "step": 1312256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8506013969154126e-05,
+      "loss": 3.9312,
+      "step": 1312768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.849762802164361e-05,
+      "loss": 3.9357,
+      "step": 1313280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8489258452936818e-05,
+      "loss": 3.9436,
+      "step": 1313792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8480872505426298e-05,
+      "loss": 3.938,
+      "step": 1314304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8472486557915778e-05,
+      "loss": 3.9426,
+      "step": 1314816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8464116989208987e-05,
+      "loss": 3.9376,
+      "step": 1315328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.845573104169847e-05,
+      "loss": 3.9458,
+      "step": 1315840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.844734509418795e-05,
+      "loss": 3.9231,
+      "step": 1316352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.843895914667743e-05,
+      "loss": 3.9408,
+      "step": 1316864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.843057319916691e-05,
+      "loss": 3.9202,
+      "step": 1317376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.842218725165639e-05,
+      "loss": 3.94,
+      "step": 1317888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.841380130414587e-05,
+      "loss": 3.9295,
+      "step": 1318400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.840543173543908e-05,
+      "loss": 3.9445,
+      "step": 1318912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8397045787928563e-05,
+      "loss": 3.9281,
+      "step": 1319424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8388659840418043e-05,
+      "loss": 3.9421,
+      "step": 1319936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8380273892907523e-05,
+      "loss": 3.9419,
+      "step": 1320448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8371887945397003e-05,
+      "loss": 3.9402,
+      "step": 1320960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8363518376690212e-05,
+      "loss": 3.9435,
+      "step": 1321472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.835513242917969e-05,
+      "loss": 3.9189,
+      "step": 1321984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.834674648166917e-05,
+      "loss": 3.9262,
+      "step": 1322496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8338360534158655e-05,
+      "loss": 3.9412,
+      "step": 1323008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8329974586648135e-05,
+      "loss": 3.938,
+      "step": 1323520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8321588639137615e-05,
+      "loss": 3.9342,
+      "step": 1324032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8313202691627088e-05,
+      "loss": 3.9301,
+      "step": 1324544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8304816744116568e-05,
+      "loss": 3.9292,
+      "step": 1325056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8296447175409784e-05,
+      "loss": 3.9191,
+      "step": 1325568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8288061227899264e-05,
+      "loss": 3.9371,
+      "step": 1326080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.827967528038874e-05,
+      "loss": 3.9192,
+      "step": 1326592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.827128933287822e-05,
+      "loss": 3.9327,
+      "step": 1327104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.82629033853677e-05,
+      "loss": 3.9394,
+      "step": 1327616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.825451743785718e-05,
+      "loss": 3.9254,
+      "step": 1328128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8246147869150393e-05,
+      "loss": 3.9252,
+      "step": 1328640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8237761921639873e-05,
+      "loss": 3.9293,
+      "step": 1329152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8229375974129353e-05,
+      "loss": 3.9203,
+      "step": 1329664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8220990026618833e-05,
+      "loss": 3.9201,
+      "step": 1330176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8212604079108313e-05,
+      "loss": 3.9416,
+      "step": 1330688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8204218131597793e-05,
+      "loss": 3.9282,
+      "step": 1331200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8195832184087273e-05,
+      "loss": 3.9241,
+      "step": 1331712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8187446236576752e-05,
+      "loss": 3.9219,
+      "step": 1332224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8179060289066232e-05,
+      "loss": 3.9262,
+      "step": 1332736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8170690720359445e-05,
+      "loss": 3.9403,
+      "step": 1333248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8162304772848925e-05,
+      "loss": 3.935,
+      "step": 1333760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8153918825338405e-05,
+      "loss": 3.9276,
+      "step": 1334272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8145549256631614e-05,
+      "loss": 3.939,
+      "step": 1334784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8137179687924826e-05,
+      "loss": 3.9414,
+      "step": 1335296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8128793740414306e-05,
+      "loss": 3.9255,
+      "step": 1335808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8120407792903786e-05,
+      "loss": 3.9322,
+      "step": 1336320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8112021845393266e-05,
+      "loss": 3.9317,
+      "step": 1336832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8103635897882746e-05,
+      "loss": 3.9337,
+      "step": 1337344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8095249950372226e-05,
+      "loss": 3.9277,
+      "step": 1337856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8086864002861706e-05,
+      "loss": 3.9275,
+      "step": 1338368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8078478055351186e-05,
+      "loss": 3.9393,
+      "step": 1338880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8070092107840666e-05,
+      "loss": 3.9387,
+      "step": 1339392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.806172253913388e-05,
+      "loss": 3.9298,
+      "step": 1339904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.805333659162336e-05,
+      "loss": 3.9071,
+      "step": 1340416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.804495064411284e-05,
+      "loss": 3.9261,
+      "step": 1340928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.803656469660232e-05,
+      "loss": 3.9322,
+      "step": 1341440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.80281787490918e-05,
+      "loss": 3.9319,
+      "step": 1341952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8019792801581275e-05,
+      "loss": 3.93,
+      "step": 1342464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.801142323287449e-05,
+      "loss": 3.9296,
+      "step": 1342976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.800303728536397e-05,
+      "loss": 3.9325,
+      "step": 1343488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.799465133785345e-05,
+      "loss": 3.9269,
+      "step": 1344000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7986265390342924e-05,
+      "loss": 3.9217,
+      "step": 1344512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7977879442832404e-05,
+      "loss": 3.9359,
+      "step": 1345024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7969493495321887e-05,
+      "loss": 3.919,
+      "step": 1345536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7961123926615103e-05,
+      "loss": 3.9398,
+      "step": 1346048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7952737979104576e-05,
+      "loss": 3.9351,
+      "step": 1346560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7944352031594056e-05,
+      "loss": 3.9148,
+      "step": 1347072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7935966084083536e-05,
+      "loss": 3.9327,
+      "step": 1347584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7927580136573016e-05,
+      "loss": 3.9292,
+      "step": 1348096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7919194189062496e-05,
+      "loss": 3.9392,
+      "step": 1348608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.791080824155198e-05,
+      "loss": 3.928,
+      "step": 1349120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.790242229404146e-05,
+      "loss": 3.926,
+      "step": 1349632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.789405272533467e-05,
+      "loss": 3.9207,
+      "step": 1350144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.788566677782415e-05,
+      "loss": 3.9351,
+      "step": 1350656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.787728083031363e-05,
+      "loss": 3.9132,
+      "step": 1351168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.786889488280311e-05,
+      "loss": 3.9255,
+      "step": 1351680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7860508935292588e-05,
+      "loss": 3.9304,
+      "step": 1352192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.78521393665858e-05,
+      "loss": 3.9282,
+      "step": 1352704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.784375341907528e-05,
+      "loss": 3.9185,
+      "step": 1353216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.783536747156476e-05,
+      "loss": 3.9304,
+      "step": 1353728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.782699790285797e-05,
+      "loss": 3.9231,
+      "step": 1354240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.781861195534745e-05,
+      "loss": 3.9285,
+      "step": 1354752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7810226007836933e-05,
+      "loss": 3.9261,
+      "step": 1355264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7801856439130142e-05,
+      "loss": 3.9228,
+      "step": 1355776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7793470491619622e-05,
+      "loss": 3.9465,
+      "step": 1356288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7785084544109102e-05,
+      "loss": 3.9369,
+      "step": 1356800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7776698596598582e-05,
+      "loss": 3.9311,
+      "step": 1357312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7768312649088062e-05,
+      "loss": 3.9218,
+      "step": 1357824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7759926701577542e-05,
+      "loss": 3.9278,
+      "step": 1358336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7751540754067022e-05,
+      "loss": 3.9286,
+      "step": 1358848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7743154806556505e-05,
+      "loss": 3.9224,
+      "step": 1359360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7734768859045985e-05,
+      "loss": 3.9383,
+      "step": 1359872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7726382911535465e-05,
+      "loss": 3.9283,
+      "step": 1360384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.771799696402494e-05,
+      "loss": 3.9284,
+      "step": 1360896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7709611016514418e-05,
+      "loss": 3.9238,
+      "step": 1361408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7701241447807634e-05,
+      "loss": 3.923,
+      "step": 1361920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7692871879100847e-05,
+      "loss": 3.9276,
+      "step": 1362432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7684502310394056e-05,
+      "loss": 3.9265,
+      "step": 1362944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7676116362883536e-05,
+      "loss": 3.9157,
+      "step": 1363456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7667746794176748e-05,
+      "loss": 3.9458,
+      "step": 1363968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7659360846666228e-05,
+      "loss": 3.9367,
+      "step": 1364480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7650974899155708e-05,
+      "loss": 3.928,
+      "step": 1364992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7642588951645188e-05,
+      "loss": 3.9278,
+      "step": 1365504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7634203004134668e-05,
+      "loss": 3.9244,
+      "step": 1366016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7625817056624148e-05,
+      "loss": 3.9379,
+      "step": 1366528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7617431109113628e-05,
+      "loss": 3.9323,
+      "step": 1367040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7609045161603108e-05,
+      "loss": 3.9275,
+      "step": 1367552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7600659214092584e-05,
+      "loss": 3.9414,
+      "step": 1368064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7592273266582064e-05,
+      "loss": 3.9327,
+      "step": 1368576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7583887319071544e-05,
+      "loss": 3.9355,
+      "step": 1369088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7575501371561024e-05,
+      "loss": 3.9279,
+      "step": 1369600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7567131802854233e-05,
+      "loss": 3.9374,
+      "step": 1370112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7558745855343713e-05,
+      "loss": 3.9315,
+      "step": 1370624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7550359907833197e-05,
+      "loss": 3.9285,
+      "step": 1371136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7541973960322677e-05,
+      "loss": 3.9289,
+      "step": 1371648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7533588012812157e-05,
+      "loss": 3.9312,
+      "step": 1372160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7525218444105366e-05,
+      "loss": 3.9343,
+      "step": 1372672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7516832496594846e-05,
+      "loss": 3.9221,
+      "step": 1373184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7508446549084326e-05,
+      "loss": 3.9249,
+      "step": 1373696
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.010009765625,
+      "eval_runtime": 293.1177,
+      "eval_samples_per_second": 1301.835,
+      "eval_steps_per_second": 40.683,
+      "step": 1373760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7500060601573806e-05,
+      "loss": 3.9199,
+      "step": 1374208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7491691032867018e-05,
+      "loss": 3.9228,
+      "step": 1374720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7483305085356498e-05,
+      "loss": 3.9421,
+      "step": 1375232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7474919137845978e-05,
+      "loss": 3.9369,
+      "step": 1375744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7466533190335458e-05,
+      "loss": 3.9291,
+      "step": 1376256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7458147242824938e-05,
+      "loss": 3.9286,
+      "step": 1376768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7449761295314418e-05,
+      "loss": 3.9285,
+      "step": 1377280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7441375347803898e-05,
+      "loss": 3.9167,
+      "step": 1377792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.743298940029338e-05,
+      "loss": 3.9373,
+      "step": 1378304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.742461983158659e-05,
+      "loss": 3.932,
+      "step": 1378816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.741623388407607e-05,
+      "loss": 3.9327,
+      "step": 1379328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.740786431536928e-05,
+      "loss": 3.9344,
+      "step": 1379840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7399494746662492e-05,
+      "loss": 3.9245,
+      "step": 1380352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7391108799151972e-05,
+      "loss": 3.9204,
+      "step": 1380864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.738272285164145e-05,
+      "loss": 3.9197,
+      "step": 1381376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.737433690413093e-05,
+      "loss": 3.92,
+      "step": 1381888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.736595095662041e-05,
+      "loss": 3.9267,
+      "step": 1382400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.735756500910989e-05,
+      "loss": 3.9198,
+      "step": 1382912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.734917906159937e-05,
+      "loss": 3.928,
+      "step": 1383424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.734079311408885e-05,
+      "loss": 3.9518,
+      "step": 1383936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7332407166578335e-05,
+      "loss": 3.927,
+      "step": 1384448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7324021219067815e-05,
+      "loss": 3.9311,
+      "step": 1384960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7315635271557295e-05,
+      "loss": 3.9289,
+      "step": 1385472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7307249324046768e-05,
+      "loss": 3.9355,
+      "step": 1385984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7298863376536248e-05,
+      "loss": 3.9207,
+      "step": 1386496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7290477429025728e-05,
+      "loss": 3.9298,
+      "step": 1387008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7282107860318944e-05,
+      "loss": 3.9184,
+      "step": 1387520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.727372191280842e-05,
+      "loss": 3.9242,
+      "step": 1388032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.72653359652979e-05,
+      "loss": 3.9198,
+      "step": 1388544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.725695001778738e-05,
+      "loss": 3.9205,
+      "step": 1389056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.724856407027686e-05,
+      "loss": 3.9255,
+      "step": 1389568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7240194501570073e-05,
+      "loss": 3.935,
+      "step": 1390080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7231808554059553e-05,
+      "loss": 3.9281,
+      "step": 1390592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7223422606549033e-05,
+      "loss": 3.9365,
+      "step": 1391104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.721505303784224e-05,
+      "loss": 3.9215,
+      "step": 1391616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.720666709033172e-05,
+      "loss": 3.9385,
+      "step": 1392128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.71982811428212e-05,
+      "loss": 3.9113,
+      "step": 1392640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.718989519531068e-05,
+      "loss": 3.9327,
+      "step": 1393152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7181509247800165e-05,
+      "loss": 3.911,
+      "step": 1393664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7173123300289645e-05,
+      "loss": 3.9295,
+      "step": 1394176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7164737352779125e-05,
+      "loss": 3.9195,
+      "step": 1394688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7156351405268605e-05,
+      "loss": 3.9381,
+      "step": 1395200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7147981836561814e-05,
+      "loss": 3.9192,
+      "step": 1395712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7139612267855026e-05,
+      "loss": 3.929,
+      "step": 1396224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7131226320344506e-05,
+      "loss": 3.93,
+      "step": 1396736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7122840372833986e-05,
+      "loss": 3.9316,
+      "step": 1397248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7114454425323466e-05,
+      "loss": 3.9325,
+      "step": 1397760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7106068477812946e-05,
+      "loss": 3.9145,
+      "step": 1398272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7097682530302426e-05,
+      "loss": 3.9131,
+      "step": 1398784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7089296582791906e-05,
+      "loss": 3.933,
+      "step": 1399296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7080910635281386e-05,
+      "loss": 3.9289,
+      "step": 1399808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7072524687770866e-05,
+      "loss": 3.9229,
+      "step": 1400320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.706415511906408e-05,
+      "loss": 3.9217,
+      "step": 1400832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.705576917155356e-05,
+      "loss": 3.9191,
+      "step": 1401344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.704738322404304e-05,
+      "loss": 3.9077,
+      "step": 1401856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7038997276532518e-05,
+      "loss": 3.9251,
+      "step": 1402368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7030611329021998e-05,
+      "loss": 3.9095,
+      "step": 1402880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7022225381511478e-05,
+      "loss": 3.9238,
+      "step": 1403392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.701385581280469e-05,
+      "loss": 3.9309,
+      "step": 1403904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.700546986529417e-05,
+      "loss": 3.9131,
+      "step": 1404416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.699708391778365e-05,
+      "loss": 3.9164,
+      "step": 1404928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.698869797027313e-05,
+      "loss": 3.9206,
+      "step": 1405440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6980312022762604e-05,
+      "loss": 3.9107,
+      "step": 1405952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.697194245405582e-05,
+      "loss": 3.9096,
+      "step": 1406464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.69635565065453e-05,
+      "loss": 3.9313,
+      "step": 1406976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6955170559034783e-05,
+      "loss": 3.9192,
+      "step": 1407488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6946800990327992e-05,
+      "loss": 3.9142,
+      "step": 1408000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6938415042817472e-05,
+      "loss": 3.9118,
+      "step": 1408512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6930029095306952e-05,
+      "loss": 3.9173,
+      "step": 1409024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.692164314779643e-05,
+      "loss": 3.9294,
+      "step": 1409536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.691325720028591e-05,
+      "loss": 3.927,
+      "step": 1410048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.690487125277539e-05,
+      "loss": 3.9163,
+      "step": 1410560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.689648530526487e-05,
+      "loss": 3.9286,
+      "step": 1411072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6888099357754348e-05,
+      "loss": 3.9288,
+      "step": 1411584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6879729789047557e-05,
+      "loss": 3.9181,
+      "step": 1412096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6871343841537037e-05,
+      "loss": 3.9206,
+      "step": 1412608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.686295789402652e-05,
+      "loss": 3.9253,
+      "step": 1413120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6854571946516e-05,
+      "loss": 3.9238,
+      "step": 1413632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.684618599900548e-05,
+      "loss": 3.9186,
+      "step": 1414144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.683780005149496e-05,
+      "loss": 3.9166,
+      "step": 1414656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.682941410398444e-05,
+      "loss": 3.9283,
+      "step": 1415168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.682104453527765e-05,
+      "loss": 3.9319,
+      "step": 1415680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.681265858776713e-05,
+      "loss": 3.9185,
+      "step": 1416192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6804272640256613e-05,
+      "loss": 3.9045,
+      "step": 1416704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6795886692746093e-05,
+      "loss": 3.9156,
+      "step": 1417216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6787500745235573e-05,
+      "loss": 3.9211,
+      "step": 1417728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6779114797725053e-05,
+      "loss": 3.9248,
+      "step": 1418240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6770728850214533e-05,
+      "loss": 3.9176,
+      "step": 1418752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6762375660311474e-05,
+      "loss": 3.9196,
+      "step": 1419264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6753989712800954e-05,
+      "loss": 3.928,
+      "step": 1419776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6745603765290434e-05,
+      "loss": 3.9127,
+      "step": 1420288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6737217817779914e-05,
+      "loss": 3.9139,
+      "step": 1420800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6728831870269394e-05,
+      "loss": 3.9241,
+      "step": 1421312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6720445922758874e-05,
+      "loss": 3.9135,
+      "step": 1421824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6712059975248354e-05,
+      "loss": 3.9331,
+      "step": 1422336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6703674027737834e-05,
+      "loss": 3.9223,
+      "step": 1422848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6695304459031047e-05,
+      "loss": 3.9076,
+      "step": 1423360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6686918511520526e-05,
+      "loss": 3.9205,
+      "step": 1423872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6678532564010006e-05,
+      "loss": 3.9186,
+      "step": 1424384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6670146616499486e-05,
+      "loss": 3.9342,
+      "step": 1424896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6661760668988966e-05,
+      "loss": 3.9223,
+      "step": 1425408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6653374721478443e-05,
+      "loss": 3.9155,
+      "step": 1425920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6644988773967923e-05,
+      "loss": 3.9103,
+      "step": 1426432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6636602826457403e-05,
+      "loss": 3.9244,
+      "step": 1426944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6628216878946883e-05,
+      "loss": 3.9069,
+      "step": 1427456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6619847310240092e-05,
+      "loss": 3.9177,
+      "step": 1427968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6611461362729572e-05,
+      "loss": 3.9187,
+      "step": 1428480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6603091794022788e-05,
+      "loss": 3.9179,
+      "step": 1428992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6594705846512264e-05,
+      "loss": 3.9088,
+      "step": 1429504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.658633627780548e-05,
+      "loss": 3.9263,
+      "step": 1430016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.657795033029496e-05,
+      "loss": 3.9134,
+      "step": 1430528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.656956438278444e-05,
+      "loss": 3.9166,
+      "step": 1431040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6561178435273913e-05,
+      "loss": 3.9207,
+      "step": 1431552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6552792487763397e-05,
+      "loss": 3.9131,
+      "step": 1432064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6544406540252877e-05,
+      "loss": 3.9344,
+      "step": 1432576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6536020592742356e-05,
+      "loss": 3.925,
+      "step": 1433088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6527634645231836e-05,
+      "loss": 3.9237,
+      "step": 1433600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6519248697721316e-05,
+      "loss": 3.912,
+      "step": 1434112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6510862750210796e-05,
+      "loss": 3.9176,
+      "step": 1434624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6502493181504005e-05,
+      "loss": 3.9219,
+      "step": 1435136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6494107233993485e-05,
+      "loss": 3.91,
+      "step": 1435648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.648572128648297e-05,
+      "loss": 3.927,
+      "step": 1436160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.647733533897245e-05,
+      "loss": 3.9199,
+      "step": 1436672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.646894939146193e-05,
+      "loss": 3.9193,
+      "step": 1437184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.646056344395141e-05,
+      "loss": 3.9188,
+      "step": 1437696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.645217749644089e-05,
+      "loss": 3.9184,
+      "step": 1438208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.644379154893037e-05,
+      "loss": 3.9118,
+      "step": 1438720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6435421980223578e-05,
+      "loss": 3.9233,
+      "step": 1439232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.642703603271306e-05,
+      "loss": 3.9087,
+      "step": 1439744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.641866646400627e-05,
+      "loss": 3.9333,
+      "step": 1440256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.641028051649575e-05,
+      "loss": 3.9314,
+      "step": 1440768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.640189456898523e-05,
+      "loss": 3.916,
+      "step": 1441280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.639350862147471e-05,
+      "loss": 3.9178,
+      "step": 1441792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.638512267396419e-05,
+      "loss": 3.9168,
+      "step": 1442304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.637673672645367e-05,
+      "loss": 3.9261,
+      "step": 1442816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6368367157746882e-05,
+      "loss": 3.9258,
+      "step": 1443328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6359981210236362e-05,
+      "loss": 3.9198,
+      "step": 1443840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6351595262725842e-05,
+      "loss": 3.9294,
+      "step": 1444352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6343209315215322e-05,
+      "loss": 3.9255,
+      "step": 1444864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.633483974650853e-05,
+      "loss": 3.9278,
+      "step": 1445376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6326453798998015e-05,
+      "loss": 3.9176,
+      "step": 1445888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6318067851487495e-05,
+      "loss": 3.9318,
+      "step": 1446400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6309681903976975e-05,
+      "loss": 3.9231,
+      "step": 1446912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6301295956466448e-05,
+      "loss": 3.9146,
+      "step": 1447424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6292910008955928e-05,
+      "loss": 3.924,
+      "step": 1447936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6284524061445408e-05,
+      "loss": 3.9244,
+      "step": 1448448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.627613811393489e-05,
+      "loss": 3.9232,
+      "step": 1448960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.62677685452281e-05,
+      "loss": 3.9093,
+      "step": 1449472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.625938259771758e-05,
+      "loss": 3.9123,
+      "step": 1449984
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.005962371826172,
+      "eval_runtime": 296.402,
+      "eval_samples_per_second": 1287.41,
+      "eval_steps_per_second": 40.233,
+      "step": 1450080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.625099665020706e-05,
+      "loss": 3.9118,
+      "step": 1450496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.624261070269654e-05,
+      "loss": 3.9113,
+      "step": 1451008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.623422475518602e-05,
+      "loss": 3.933,
+      "step": 1451520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.62258388076755e-05,
+      "loss": 3.9286,
+      "step": 1452032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6217452860164983e-05,
+      "loss": 3.927,
+      "step": 1452544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6209066912654463e-05,
+      "loss": 3.9141,
+      "step": 1453056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6200680965143943e-05,
+      "loss": 3.9238,
+      "step": 1453568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6192295017633423e-05,
+      "loss": 3.9056,
+      "step": 1454080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6183925448926632e-05,
+      "loss": 3.9284,
+      "step": 1454592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6175539501416112e-05,
+      "loss": 3.9238,
+      "step": 1455104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6167153553905592e-05,
+      "loss": 3.9208,
+      "step": 1455616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6158767606395075e-05,
+      "loss": 3.9304,
+      "step": 1456128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6150398037688285e-05,
+      "loss": 3.9165,
+      "step": 1456640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6142012090177764e-05,
+      "loss": 3.9154,
+      "step": 1457152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6133626142667244e-05,
+      "loss": 3.9057,
+      "step": 1457664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6125240195156724e-05,
+      "loss": 3.9183,
+      "step": 1458176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6116854247646204e-05,
+      "loss": 3.9141,
+      "step": 1458688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6108468300135684e-05,
+      "loss": 3.914,
+      "step": 1459200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6100082352625164e-05,
+      "loss": 3.9169,
+      "step": 1459712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.609169640511464e-05,
+      "loss": 3.9405,
+      "step": 1460224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.608331045760412e-05,
+      "loss": 3.9194,
+      "step": 1460736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.60749245100936e-05,
+      "loss": 3.9227,
+      "step": 1461248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.606653856258308e-05,
+      "loss": 3.9218,
+      "step": 1461760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.605815261507256e-05,
+      "loss": 3.9282,
+      "step": 1462272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.604976666756204e-05,
+      "loss": 3.9153,
+      "step": 1462784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6041380720051524e-05,
+      "loss": 3.9209,
+      "step": 1463296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6032994772541004e-05,
+      "loss": 3.9106,
+      "step": 1463808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6024608825030484e-05,
+      "loss": 3.914,
+      "step": 1464320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6016222877519964e-05,
+      "loss": 3.9114,
+      "step": 1464832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6007836930009444e-05,
+      "loss": 3.9188,
+      "step": 1465344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5999450982498924e-05,
+      "loss": 3.9155,
+      "step": 1465856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5991081413792133e-05,
+      "loss": 3.9264,
+      "step": 1466368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5982695466281616e-05,
+      "loss": 3.9202,
+      "step": 1466880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5974309518771096e-05,
+      "loss": 3.9245,
+      "step": 1467392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5965939950064305e-05,
+      "loss": 3.9165,
+      "step": 1467904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5957554002553785e-05,
+      "loss": 3.9262,
+      "step": 1468416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5949168055043265e-05,
+      "loss": 3.9096,
+      "step": 1468928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5940782107532745e-05,
+      "loss": 3.9198,
+      "step": 1469440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5932396160022225e-05,
+      "loss": 3.9018,
+      "step": 1469952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.59240102125117e-05,
+      "loss": 3.9183,
+      "step": 1470464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.591562426500118e-05,
+      "loss": 3.9131,
+      "step": 1470976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.590723831749066e-05,
+      "loss": 3.9303,
+      "step": 1471488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.589885236998014e-05,
+      "loss": 3.9141,
+      "step": 1472000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.589046642246962e-05,
+      "loss": 3.9191,
+      "step": 1472512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5882096853762834e-05,
+      "loss": 3.9221,
+      "step": 1473024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5873710906252314e-05,
+      "loss": 3.9275,
+      "step": 1473536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5865324958741794e-05,
+      "loss": 3.9208,
+      "step": 1474048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5856939011231274e-05,
+      "loss": 3.9091,
+      "step": 1474560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5848569442524483e-05,
+      "loss": 3.9024,
+      "step": 1475072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5840183495013963e-05,
+      "loss": 3.9262,
+      "step": 1475584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5831797547503446e-05,
+      "loss": 3.92,
+      "step": 1476096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5823411599992926e-05,
+      "loss": 3.9172,
+      "step": 1476608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5815025652482406e-05,
+      "loss": 3.9099,
+      "step": 1477120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5806639704971886e-05,
+      "loss": 3.9148,
+      "step": 1477632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5798253757461366e-05,
+      "loss": 3.8962,
+      "step": 1478144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5789884188754575e-05,
+      "loss": 3.9194,
+      "step": 1478656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5781498241244055e-05,
+      "loss": 3.9008,
+      "step": 1479168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.577311229373354e-05,
+      "loss": 3.9164,
+      "step": 1479680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.576472634622302e-05,
+      "loss": 3.9215,
+      "step": 1480192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5756340398712498e-05,
+      "loss": 3.9105,
+      "step": 1480704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5747954451201978e-05,
+      "loss": 3.9046,
+      "step": 1481216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5739568503691458e-05,
+      "loss": 3.9139,
+      "step": 1481728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5731182556180938e-05,
+      "loss": 3.9005,
+      "step": 1482240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5722796608670415e-05,
+      "loss": 3.9059,
+      "step": 1482752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5714427039963627e-05,
+      "loss": 3.9193,
+      "step": 1483264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.570604109245311e-05,
+      "loss": 3.915,
+      "step": 1483776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.569765514494259e-05,
+      "loss": 3.9062,
+      "step": 1484288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5689269197432064e-05,
+      "loss": 3.9016,
+      "step": 1484800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.568089962872528e-05,
+      "loss": 3.9118,
+      "step": 1485312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.567251368121476e-05,
+      "loss": 3.9189,
+      "step": 1485824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.566412773370424e-05,
+      "loss": 3.9184,
+      "step": 1486336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5655741786193716e-05,
+      "loss": 3.9125,
+      "step": 1486848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5647355838683196e-05,
+      "loss": 3.9192,
+      "step": 1487360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.563900264878014e-05,
+      "loss": 3.9211,
+      "step": 1487872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.563061670126962e-05,
+      "loss": 3.9111,
+      "step": 1488384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.56222307537591e-05,
+      "loss": 3.9114,
+      "step": 1488896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.561384480624858e-05,
+      "loss": 3.9185,
+      "step": 1489408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5605458858738064e-05,
+      "loss": 3.9163,
+      "step": 1489920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5597089290031273e-05,
+      "loss": 3.9118,
+      "step": 1490432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5588703342520753e-05,
+      "loss": 3.91,
+      "step": 1490944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5580317395010233e-05,
+      "loss": 3.9177,
+      "step": 1491456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5571931447499713e-05,
+      "loss": 3.9228,
+      "step": 1491968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.556354549998919e-05,
+      "loss": 3.915,
+      "step": 1492480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.555515955247867e-05,
+      "loss": 3.8975,
+      "step": 1492992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.554677360496815e-05,
+      "loss": 3.9018,
+      "step": 1493504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.553840403626136e-05,
+      "loss": 3.912,
+      "step": 1494016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.553001808875084e-05,
+      "loss": 3.917,
+      "step": 1494528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.552163214124032e-05,
+      "loss": 3.9147,
+      "step": 1495040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5513262572533535e-05,
+      "loss": 3.9066,
+      "step": 1495552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.550487662502301e-05,
+      "loss": 3.9196,
+      "step": 1496064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.549649067751249e-05,
+      "loss": 3.9118,
+      "step": 1496576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.548810473000197e-05,
+      "loss": 3.9068,
+      "step": 1497088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.547971878249145e-05,
+      "loss": 3.9115,
+      "step": 1497600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.547133283498093e-05,
+      "loss": 3.9154,
+      "step": 1498112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.546294688747041e-05,
+      "loss": 3.9237,
+      "step": 1498624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5454560939959894e-05,
+      "loss": 3.9104,
+      "step": 1499136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5446174992449374e-05,
+      "loss": 3.8986,
+      "step": 1499648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5437789044938854e-05,
+      "loss": 3.9167,
+      "step": 1500160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5429403097428334e-05,
+      "loss": 3.9092,
+      "step": 1500672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5421017149917814e-05,
+      "loss": 3.9253,
+      "step": 1501184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5412647581211023e-05,
+      "loss": 3.9166,
+      "step": 1501696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5404278012504236e-05,
+      "loss": 3.9063,
+      "step": 1502208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5395892064993716e-05,
+      "loss": 3.9018,
+      "step": 1502720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5387506117483196e-05,
+      "loss": 3.9177,
+      "step": 1503232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5379120169972675e-05,
+      "loss": 3.8974,
+      "step": 1503744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5370734222462155e-05,
+      "loss": 3.9125,
+      "step": 1504256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5362348274951635e-05,
+      "loss": 3.9073,
+      "step": 1504768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5353962327441115e-05,
+      "loss": 3.9141,
+      "step": 1505280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5345592758734328e-05,
+      "loss": 3.902,
+      "step": 1505792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5337206811223808e-05,
+      "loss": 3.9197,
+      "step": 1506304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5328820863713288e-05,
+      "loss": 3.9055,
+      "step": 1506816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5320434916202768e-05,
+      "loss": 3.9076,
+      "step": 1507328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5312048968692248e-05,
+      "loss": 3.9151,
+      "step": 1507840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5303679399985457e-05,
+      "loss": 3.906,
+      "step": 1508352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.529530983127867e-05,
+      "loss": 3.9282,
+      "step": 1508864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.528692388376815e-05,
+      "loss": 3.9198,
+      "step": 1509376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.527853793625763e-05,
+      "loss": 3.9163,
+      "step": 1509888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.527015198874711e-05,
+      "loss": 3.9051,
+      "step": 1510400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.526176604123659e-05,
+      "loss": 3.9112,
+      "step": 1510912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.525338009372607e-05,
+      "loss": 3.9106,
+      "step": 1511424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5244994146215546e-05,
+      "loss": 3.9068,
+      "step": 1511936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5236608198705026e-05,
+      "loss": 3.9185,
+      "step": 1512448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5228222251194506e-05,
+      "loss": 3.9117,
+      "step": 1512960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5219836303683985e-05,
+      "loss": 3.9117,
+      "step": 1513472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5211450356173465e-05,
+      "loss": 3.9091,
+      "step": 1513984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5203080787466678e-05,
+      "loss": 3.9115,
+      "step": 1514496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5194694839956158e-05,
+      "loss": 3.9061,
+      "step": 1515008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5186325271249374e-05,
+      "loss": 3.9185,
+      "step": 1515520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5177939323738847e-05,
+      "loss": 3.9009,
+      "step": 1516032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5169553376228327e-05,
+      "loss": 3.9213,
+      "step": 1516544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5161167428717807e-05,
+      "loss": 3.9194,
+      "step": 1517056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5152781481207287e-05,
+      "loss": 3.9172,
+      "step": 1517568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.514439553369677e-05,
+      "loss": 3.9088,
+      "step": 1518080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.513600958618625e-05,
+      "loss": 3.9101,
+      "step": 1518592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.512762363867573e-05,
+      "loss": 3.9182,
+      "step": 1519104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.511923769116521e-05,
+      "loss": 3.9171,
+      "step": 1519616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.511085174365469e-05,
+      "loss": 3.9083,
+      "step": 1520128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.51024821749479e-05,
+      "loss": 3.9208,
+      "step": 1520640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.509409622743738e-05,
+      "loss": 3.9151,
+      "step": 1521152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.508571027992686e-05,
+      "loss": 3.9214,
+      "step": 1521664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5077324332416342e-05,
+      "loss": 3.9114,
+      "step": 1522176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5068938384905822e-05,
+      "loss": 3.9237,
+      "step": 1522688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5060552437395302e-05,
+      "loss": 3.9172,
+      "step": 1523200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5052166489884782e-05,
+      "loss": 3.9086,
+      "step": 1523712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5043780542374262e-05,
+      "loss": 3.9128,
+      "step": 1524224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5035394594863735e-05,
+      "loss": 3.9206,
+      "step": 1524736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.502702502615695e-05,
+      "loss": 3.9123,
+      "step": 1525248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5018639078646435e-05,
+      "loss": 3.8993,
+      "step": 1525760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5010269509939644e-05,
+      "loss": 3.9137,
+      "step": 1526272
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.003040313720703,
+      "eval_runtime": 293.8362,
+      "eval_samples_per_second": 1298.652,
+      "eval_steps_per_second": 40.584,
+      "step": 1526400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5001883562429124e-05,
+      "loss": 3.9074,
+      "step": 1526784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.49934976149186e-05,
+      "loss": 3.9042,
+      "step": 1527296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.498511166740808e-05,
+      "loss": 3.9228,
+      "step": 1527808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4976725719897563e-05,
+      "loss": 3.9214,
+      "step": 1528320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4968339772387043e-05,
+      "loss": 3.9194,
+      "step": 1528832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4959986582483985e-05,
+      "loss": 3.9116,
+      "step": 1529344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4951600634973465e-05,
+      "loss": 3.9148,
+      "step": 1529856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.494321468746294e-05,
+      "loss": 3.8961,
+      "step": 1530368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4934828739952425e-05,
+      "loss": 3.9207,
+      "step": 1530880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4926442792441905e-05,
+      "loss": 3.9183,
+      "step": 1531392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4918056844931385e-05,
+      "loss": 3.9114,
+      "step": 1531904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4909670897420865e-05,
+      "loss": 3.9202,
+      "step": 1532416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4901301328714074e-05,
+      "loss": 3.9083,
+      "step": 1532928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4892915381203554e-05,
+      "loss": 3.9087,
+      "step": 1533440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4884529433693034e-05,
+      "loss": 3.9019,
+      "step": 1533952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4876143486182517e-05,
+      "loss": 3.904,
+      "step": 1534464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4867757538671997e-05,
+      "loss": 3.9069,
+      "step": 1534976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4859371591161474e-05,
+      "loss": 3.9056,
+      "step": 1535488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4850985643650954e-05,
+      "loss": 3.9114,
+      "step": 1536000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4842599696140434e-05,
+      "loss": 3.9308,
+      "step": 1536512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4834213748629913e-05,
+      "loss": 3.9138,
+      "step": 1537024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4825827801119393e-05,
+      "loss": 3.9181,
+      "step": 1537536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4817441853608873e-05,
+      "loss": 3.9129,
+      "step": 1538048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4809055906098357e-05,
+      "loss": 3.9186,
+      "step": 1538560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4800669958587833e-05,
+      "loss": 3.9076,
+      "step": 1539072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4792300389881046e-05,
+      "loss": 3.9132,
+      "step": 1539584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4783930821174255e-05,
+      "loss": 3.9071,
+      "step": 1540096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4775544873663735e-05,
+      "loss": 3.9035,
+      "step": 1540608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4767158926153218e-05,
+      "loss": 3.9035,
+      "step": 1541120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4758772978642698e-05,
+      "loss": 3.9101,
+      "step": 1541632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4750387031132178e-05,
+      "loss": 3.9096,
+      "step": 1542144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4742017462425387e-05,
+      "loss": 3.9201,
+      "step": 1542656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4733631514914867e-05,
+      "loss": 3.914,
+      "step": 1543168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4725245567404347e-05,
+      "loss": 3.9169,
+      "step": 1543680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4716892377501292e-05,
+      "loss": 3.9091,
+      "step": 1544192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4708506429990772e-05,
+      "loss": 3.9195,
+      "step": 1544704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.470012048248025e-05,
+      "loss": 3.9034,
+      "step": 1545216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.469173453496973e-05,
+      "loss": 3.9104,
+      "step": 1545728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.468334858745921e-05,
+      "loss": 3.8962,
+      "step": 1546240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.467496263994869e-05,
+      "loss": 3.9103,
+      "step": 1546752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4666576692438172e-05,
+      "loss": 3.908,
+      "step": 1547264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4658190744927652e-05,
+      "loss": 3.9201,
+      "step": 1547776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.464980479741713e-05,
+      "loss": 3.9065,
+      "step": 1548288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.464141884990661e-05,
+      "loss": 3.9125,
+      "step": 1548800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4633032902396088e-05,
+      "loss": 3.9161,
+      "step": 1549312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4624646954885568e-05,
+      "loss": 3.9173,
+      "step": 1549824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.461627738617878e-05,
+      "loss": 3.9132,
+      "step": 1550336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.460789143866826e-05,
+      "loss": 3.9022,
+      "step": 1550848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.459950549115774e-05,
+      "loss": 3.8938,
+      "step": 1551360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.459111954364722e-05,
+      "loss": 3.9208,
+      "step": 1551872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.45827335961367e-05,
+      "loss": 3.9132,
+      "step": 1552384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.457434764862618e-05,
+      "loss": 3.9081,
+      "step": 1552896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4565961701115657e-05,
+      "loss": 3.9073,
+      "step": 1553408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.455757575360514e-05,
+      "loss": 3.9062,
+      "step": 1553920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.454918980609462e-05,
+      "loss": 3.8897,
+      "step": 1554432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4540820237387833e-05,
+      "loss": 3.9063,
+      "step": 1554944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.453243428987731e-05,
+      "loss": 3.8961,
+      "step": 1555456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.452404834236679e-05,
+      "loss": 3.9066,
+      "step": 1555968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.451566239485627e-05,
+      "loss": 3.9152,
+      "step": 1556480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4507292826149482e-05,
+      "loss": 3.903,
+      "step": 1556992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4498923257442694e-05,
+      "loss": 3.8952,
+      "step": 1557504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4490537309932174e-05,
+      "loss": 3.9118,
+      "step": 1558016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4482151362421654e-05,
+      "loss": 3.8911,
+      "step": 1558528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4473781793714863e-05,
+      "loss": 3.9003,
+      "step": 1559040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4465395846204343e-05,
+      "loss": 3.9116,
+      "step": 1559552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4457009898693827e-05,
+      "loss": 3.9074,
+      "step": 1560064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4448623951183307e-05,
+      "loss": 3.8977,
+      "step": 1560576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4440238003672783e-05,
+      "loss": 3.8985,
+      "step": 1561088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4431852056162263e-05,
+      "loss": 3.8996,
+      "step": 1561600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4423466108651743e-05,
+      "loss": 3.909,
+      "step": 1562112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4415096539944956e-05,
+      "loss": 3.91,
+      "step": 1562624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4406710592434435e-05,
+      "loss": 3.9049,
+      "step": 1563136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4398324644923915e-05,
+      "loss": 3.9124,
+      "step": 1563648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4389938697413395e-05,
+      "loss": 3.9162,
+      "step": 1564160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4381552749902875e-05,
+      "loss": 3.9042,
+      "step": 1564672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4373166802392355e-05,
+      "loss": 3.8971,
+      "step": 1565184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4364780854881835e-05,
+      "loss": 3.9169,
+      "step": 1565696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4356394907371312e-05,
+      "loss": 3.91,
+      "step": 1566208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4348008959860795e-05,
+      "loss": 3.9023,
+      "step": 1566720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4339639391154008e-05,
+      "loss": 3.904,
+      "step": 1567232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4331269822447217e-05,
+      "loss": 3.9103,
+      "step": 1567744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4322883874936697e-05,
+      "loss": 3.919,
+      "step": 1568256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4314497927426177e-05,
+      "loss": 3.9028,
+      "step": 1568768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4306111979915657e-05,
+      "loss": 3.8911,
+      "step": 1569280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4297726032405137e-05,
+      "loss": 3.9008,
+      "step": 1569792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4289340084894617e-05,
+      "loss": 3.8971,
+      "step": 1570304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4280954137384096e-05,
+      "loss": 3.9092,
+      "step": 1570816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4272568189873576e-05,
+      "loss": 3.9058,
+      "step": 1571328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4264198621166786e-05,
+      "loss": 3.8983,
+      "step": 1571840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4255812673656266e-05,
+      "loss": 3.9124,
+      "step": 1572352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.424742672614575e-05,
+      "loss": 3.9019,
+      "step": 1572864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.423904077863523e-05,
+      "loss": 3.9052,
+      "step": 1573376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.423065483112471e-05,
+      "loss": 3.8995,
+      "step": 1573888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4222285262417918e-05,
+      "loss": 3.9066,
+      "step": 1574400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4213899314907398e-05,
+      "loss": 3.9184,
+      "step": 1574912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4205513367396878e-05,
+      "loss": 3.9077,
+      "step": 1575424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4197127419886358e-05,
+      "loss": 3.8871,
+      "step": 1575936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.418875785117957e-05,
+      "loss": 3.9075,
+      "step": 1576448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.418037190366905e-05,
+      "loss": 3.9056,
+      "step": 1576960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.417198595615853e-05,
+      "loss": 3.912,
+      "step": 1577472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.416360000864801e-05,
+      "loss": 3.9123,
+      "step": 1577984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.415523043994122e-05,
+      "loss": 3.9004,
+      "step": 1578496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.41468444924307e-05,
+      "loss": 3.8955,
+      "step": 1579008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4138458544920182e-05,
+      "loss": 3.9092,
+      "step": 1579520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4130072597409662e-05,
+      "loss": 3.8878,
+      "step": 1580032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.412170302870287e-05,
+      "loss": 3.9047,
+      "step": 1580544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.411331708119235e-05,
+      "loss": 3.8992,
+      "step": 1581056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.410493113368183e-05,
+      "loss": 3.9091,
+      "step": 1581568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.409654518617131e-05,
+      "loss": 3.8939,
+      "step": 1582080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.408815923866079e-05,
+      "loss": 3.9105,
+      "step": 1582592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.407977329115027e-05,
+      "loss": 3.8969,
+      "step": 1583104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.407138734363975e-05,
+      "loss": 3.8993,
+      "step": 1583616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4063017774932964e-05,
+      "loss": 3.9088,
+      "step": 1584128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.405463182742244e-05,
+      "loss": 3.9034,
+      "step": 1584640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.404624587991192e-05,
+      "loss": 3.9203,
+      "step": 1585152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4037859932401404e-05,
+      "loss": 3.9114,
+      "step": 1585664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4029490363694616e-05,
+      "loss": 3.9039,
+      "step": 1586176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4021104416184093e-05,
+      "loss": 3.9072,
+      "step": 1586688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4012718468673573e-05,
+      "loss": 3.8979,
+      "step": 1587200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4004332521163053e-05,
+      "loss": 3.9069,
+      "step": 1587712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3995946573652533e-05,
+      "loss": 3.8975,
+      "step": 1588224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3987560626142012e-05,
+      "loss": 3.9143,
+      "step": 1588736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3979191057435225e-05,
+      "loss": 3.8998,
+      "step": 1589248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3970805109924705e-05,
+      "loss": 3.9023,
+      "step": 1589760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3962419162414185e-05,
+      "loss": 3.9007,
+      "step": 1590272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3954033214903665e-05,
+      "loss": 3.9071,
+      "step": 1590784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3945647267393145e-05,
+      "loss": 3.8985,
+      "step": 1591296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3937277698686357e-05,
+      "loss": 3.91,
+      "step": 1591808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3928891751175837e-05,
+      "loss": 3.8937,
+      "step": 1592320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3920505803665317e-05,
+      "loss": 3.9145,
+      "step": 1592832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3912119856154797e-05,
+      "loss": 3.9095,
+      "step": 1593344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3903733908644274e-05,
+      "loss": 3.9086,
+      "step": 1593856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3895347961133754e-05,
+      "loss": 3.9033,
+      "step": 1594368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3886978392426966e-05,
+      "loss": 3.9012,
+      "step": 1594880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3878592444916446e-05,
+      "loss": 3.9165,
+      "step": 1595392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3870206497405926e-05,
+      "loss": 3.9087,
+      "step": 1595904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.386183692869914e-05,
+      "loss": 3.8997,
+      "step": 1596416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.385345098118862e-05,
+      "loss": 3.9126,
+      "step": 1596928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3845065033678095e-05,
+      "loss": 3.909,
+      "step": 1597440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3836679086167575e-05,
+      "loss": 3.9155,
+      "step": 1597952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.382829313865706e-05,
+      "loss": 3.9053,
+      "step": 1598464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.381990719114654e-05,
+      "loss": 3.9126,
+      "step": 1598976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3811521243636018e-05,
+      "loss": 3.9089,
+      "step": 1599488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3803135296125498e-05,
+      "loss": 3.9015,
+      "step": 1600000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3794749348614978e-05,
+      "loss": 3.9076,
+      "step": 1600512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3786363401104455e-05,
+      "loss": 3.9098,
+      "step": 1601024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3777977453593935e-05,
+      "loss": 3.9086,
+      "step": 1601536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3769591506083415e-05,
+      "loss": 3.8923,
+      "step": 1602048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3761221937376627e-05,
+      "loss": 3.9056,
+      "step": 1602560
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9993226528167725,
+      "eval_runtime": 292.489,
+      "eval_samples_per_second": 1304.634,
+      "eval_steps_per_second": 40.771,
+      "step": 1602720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3752835989866107e-05,
+      "loss": 3.8969,
+      "step": 1603072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3744450042355587e-05,
+      "loss": 3.8953,
+      "step": 1603584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3736064094845067e-05,
+      "loss": 3.9146,
+      "step": 1604096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3727678147334547e-05,
+      "loss": 3.9139,
+      "step": 1604608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3719292199824027e-05,
+      "loss": 3.9088,
+      "step": 1605120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3710906252313507e-05,
+      "loss": 3.905,
+      "step": 1605632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3702520304802987e-05,
+      "loss": 3.9055,
+      "step": 1606144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3694134357292467e-05,
+      "loss": 3.8922,
+      "step": 1606656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3685748409781947e-05,
+      "loss": 3.9079,
+      "step": 1607168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3677362462271427e-05,
+      "loss": 3.9128,
+      "step": 1607680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3668976514760907e-05,
+      "loss": 3.9021,
+      "step": 1608192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3660590567250387e-05,
+      "loss": 3.9111,
+      "step": 1608704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.36522209985436e-05,
+      "loss": 3.9019,
+      "step": 1609216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.364383505103308e-05,
+      "loss": 3.9038,
+      "step": 1609728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.363544910352256e-05,
+      "loss": 3.8944,
+      "step": 1610240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.362706315601204e-05,
+      "loss": 3.897,
+      "step": 1610752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3618693587305248e-05,
+      "loss": 3.899,
+      "step": 1611264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3610307639794728e-05,
+      "loss": 3.8985,
+      "step": 1611776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3601921692284208e-05,
+      "loss": 3.9023,
+      "step": 1612288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.359353574477369e-05,
+      "loss": 3.9184,
+      "step": 1612800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3585149797263168e-05,
+      "loss": 3.9122,
+      "step": 1613312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3576763849752648e-05,
+      "loss": 3.9111,
+      "step": 1613824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3568377902242128e-05,
+      "loss": 3.9043,
+      "step": 1614336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3559991954731608e-05,
+      "loss": 3.914,
+      "step": 1614848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3551606007221088e-05,
+      "loss": 3.8969,
+      "step": 1615360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3543220059710568e-05,
+      "loss": 3.9109,
+      "step": 1615872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3534834112200048e-05,
+      "loss": 3.8977,
+      "step": 1616384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3526448164689528e-05,
+      "loss": 3.8935,
+      "step": 1616896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.351807859598274e-05,
+      "loss": 3.9013,
+      "step": 1617408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.350970902727595e-05,
+      "loss": 3.9003,
+      "step": 1617920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.350132307976543e-05,
+      "loss": 3.9042,
+      "step": 1618432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3492937132254912e-05,
+      "loss": 3.9088,
+      "step": 1618944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3484551184744392e-05,
+      "loss": 3.9086,
+      "step": 1619456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3476165237233872e-05,
+      "loss": 3.911,
+      "step": 1619968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.346781204733081e-05,
+      "loss": 3.902,
+      "step": 1620480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.345942609982029e-05,
+      "loss": 3.9107,
+      "step": 1620992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3451040152309774e-05,
+      "loss": 3.9,
+      "step": 1621504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3442654204799254e-05,
+      "loss": 3.8971,
+      "step": 1622016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3434268257288734e-05,
+      "loss": 3.8949,
+      "step": 1622528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3425882309778214e-05,
+      "loss": 3.9003,
+      "step": 1623040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3417496362267694e-05,
+      "loss": 3.905,
+      "step": 1623552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3409126793560903e-05,
+      "loss": 3.9106,
+      "step": 1624064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400740846050383e-05,
+      "loss": 3.8998,
+      "step": 1624576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3392354898539863e-05,
+      "loss": 3.9021,
+      "step": 1625088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3383968951029346e-05,
+      "loss": 3.9108,
+      "step": 1625600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3375583003518823e-05,
+      "loss": 3.9077,
+      "step": 1626112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3367213434812035e-05,
+      "loss": 3.9088,
+      "step": 1626624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3358827487301515e-05,
+      "loss": 3.8961,
+      "step": 1627136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.335044153979099e-05,
+      "loss": 3.8823,
+      "step": 1627648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3342055592280475e-05,
+      "loss": 3.9134,
+      "step": 1628160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3333686023573687e-05,
+      "loss": 3.9072,
+      "step": 1628672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3325300076063167e-05,
+      "loss": 3.9046,
+      "step": 1629184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3316914128552644e-05,
+      "loss": 3.9024,
+      "step": 1629696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3308528181042124e-05,
+      "loss": 3.8952,
+      "step": 1630208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3300142233531604e-05,
+      "loss": 3.8885,
+      "step": 1630720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3291756286021084e-05,
+      "loss": 3.8968,
+      "step": 1631232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3283370338510567e-05,
+      "loss": 3.8887,
+      "step": 1631744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3274984391000047e-05,
+      "loss": 3.8997,
+      "step": 1632256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3266614822293256e-05,
+      "loss": 3.9093,
+      "step": 1632768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3258228874782736e-05,
+      "loss": 3.8957,
+      "step": 1633280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3249842927272216e-05,
+      "loss": 3.8891,
+      "step": 1633792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3241456979761696e-05,
+      "loss": 3.9074,
+      "step": 1634304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3233071032251176e-05,
+      "loss": 3.8844,
+      "step": 1634816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3224685084740656e-05,
+      "loss": 3.8914,
+      "step": 1635328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.321631551603387e-05,
+      "loss": 3.9036,
+      "step": 1635840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.320792956852335e-05,
+      "loss": 3.9032,
+      "step": 1636352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3199559999816558e-05,
+      "loss": 3.8916,
+      "step": 1636864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3191174052306038e-05,
+      "loss": 3.8914,
+      "step": 1637376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.318278810479552e-05,
+      "loss": 3.8923,
+      "step": 1637888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3174402157285e-05,
+      "loss": 3.8997,
+      "step": 1638400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3166016209774477e-05,
+      "loss": 3.906,
+      "step": 1638912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3157630262263957e-05,
+      "loss": 3.8954,
+      "step": 1639424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3149244314753437e-05,
+      "loss": 3.9037,
+      "step": 1639936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3140858367242917e-05,
+      "loss": 3.9084,
+      "step": 1640448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.313248879853613e-05,
+      "loss": 3.9056,
+      "step": 1640960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.312410285102561e-05,
+      "loss": 3.8894,
+      "step": 1641472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.311571690351509e-05,
+      "loss": 3.905,
+      "step": 1641984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.310733095600457e-05,
+      "loss": 3.9071,
+      "step": 1642496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.309894500849405e-05,
+      "loss": 3.8902,
+      "step": 1643008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.309055906098353e-05,
+      "loss": 3.9011,
+      "step": 1643520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3082173113473006e-05,
+      "loss": 3.9008,
+      "step": 1644032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3073787165962486e-05,
+      "loss": 3.9118,
+      "step": 1644544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.306540121845197e-05,
+      "loss": 3.8984,
+      "step": 1645056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.305701527094145e-05,
+      "loss": 3.8794,
+      "step": 1645568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.304862932343093e-05,
+      "loss": 3.8975,
+      "step": 1646080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.304025975472414e-05,
+      "loss": 3.8921,
+      "step": 1646592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.303187380721362e-05,
+      "loss": 3.9002,
+      "step": 1647104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.30234878597031e-05,
+      "loss": 3.8977,
+      "step": 1647616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3015101912192578e-05,
+      "loss": 3.8947,
+      "step": 1648128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.300671596468206e-05,
+      "loss": 3.9014,
+      "step": 1648640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.299834639597527e-05,
+      "loss": 3.8989,
+      "step": 1649152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.298996044846475e-05,
+      "loss": 3.8919,
+      "step": 1649664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.298157450095423e-05,
+      "loss": 3.8957,
+      "step": 1650176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.297318855344371e-05,
+      "loss": 3.9041,
+      "step": 1650688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.296480260593319e-05,
+      "loss": 3.9112,
+      "step": 1651200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.295641665842267e-05,
+      "loss": 3.8996,
+      "step": 1651712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.294803071091215e-05,
+      "loss": 3.8846,
+      "step": 1652224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.293966114220536e-05,
+      "loss": 3.898,
+      "step": 1652736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.293127519469484e-05,
+      "loss": 3.8985,
+      "step": 1653248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.292288924718432e-05,
+      "loss": 3.9071,
+      "step": 1653760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.29145032996738e-05,
+      "loss": 3.9059,
+      "step": 1654272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.290611735216328e-05,
+      "loss": 3.8878,
+      "step": 1654784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2897731404652763e-05,
+      "loss": 3.8915,
+      "step": 1655296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2889345457142243e-05,
+      "loss": 3.9031,
+      "step": 1655808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.288095950963172e-05,
+      "loss": 3.8805,
+      "step": 1656320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2872589940924932e-05,
+      "loss": 3.9006,
+      "step": 1656832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.286420399341441e-05,
+      "loss": 3.8875,
+      "step": 1657344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2855834424707624e-05,
+      "loss": 3.9009,
+      "step": 1657856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2847448477197104e-05,
+      "loss": 3.8877,
+      "step": 1658368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2839062529686584e-05,
+      "loss": 3.904,
+      "step": 1658880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2830676582176064e-05,
+      "loss": 3.8934,
+      "step": 1659392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.282229063466554e-05,
+      "loss": 3.8879,
+      "step": 1659904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2813921065958753e-05,
+      "loss": 3.9024,
+      "step": 1660416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2805535118448233e-05,
+      "loss": 3.8927,
+      "step": 1660928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2797149170937716e-05,
+      "loss": 3.9132,
+      "step": 1661440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2788763223427193e-05,
+      "loss": 3.9097,
+      "step": 1661952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2780393654720405e-05,
+      "loss": 3.8944,
+      "step": 1662464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2772007707209885e-05,
+      "loss": 3.9041,
+      "step": 1662976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2763621759699365e-05,
+      "loss": 3.8893,
+      "step": 1663488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2755235812188845e-05,
+      "loss": 3.8986,
+      "step": 1664000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2746849864678325e-05,
+      "loss": 3.8945,
+      "step": 1664512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2738463917167805e-05,
+      "loss": 3.9035,
+      "step": 1665024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2730094348461014e-05,
+      "loss": 3.8958,
+      "step": 1665536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2721708400950494e-05,
+      "loss": 3.8941,
+      "step": 1666048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2713322453439974e-05,
+      "loss": 3.8961,
+      "step": 1666560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2704936505929454e-05,
+      "loss": 3.8985,
+      "step": 1667072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2696566937222667e-05,
+      "loss": 3.8899,
+      "step": 1667584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.268819736851588e-05,
+      "loss": 3.9045,
+      "step": 1668096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.267981142100536e-05,
+      "loss": 3.8874,
+      "step": 1668608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.267142547349484e-05,
+      "loss": 3.9121,
+      "step": 1669120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2663039525984316e-05,
+      "loss": 3.8965,
+      "step": 1669632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.26546535784738e-05,
+      "loss": 3.9061,
+      "step": 1670144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.264626763096328e-05,
+      "loss": 3.895,
+      "step": 1670656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.263788168345276e-05,
+      "loss": 3.8989,
+      "step": 1671168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.262949573594224e-05,
+      "loss": 3.9065,
+      "step": 1671680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.262110978843172e-05,
+      "loss": 3.9,
+      "step": 1672192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2612723840921195e-05,
+      "loss": 3.8958,
+      "step": 1672704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2604337893410675e-05,
+      "loss": 3.9073,
+      "step": 1673216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.259598470350762e-05,
+      "loss": 3.9031,
+      "step": 1673728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.25875987559971e-05,
+      "loss": 3.9073,
+      "step": 1674240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.257921280848658e-05,
+      "loss": 3.895,
+      "step": 1674752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.257082686097606e-05,
+      "loss": 3.9063,
+      "step": 1675264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.256244091346554e-05,
+      "loss": 3.9004,
+      "step": 1675776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.255405496595502e-05,
+      "loss": 3.897,
+      "step": 1676288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.25456690184445e-05,
+      "loss": 3.9006,
+      "step": 1676800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.253728307093398e-05,
+      "loss": 3.9016,
+      "step": 1677312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.252889712342346e-05,
+      "loss": 3.9061,
+      "step": 1677824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.252051117591294e-05,
+      "loss": 3.8763,
+      "step": 1678336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.251212522840242e-05,
+      "loss": 3.9041,
+      "step": 1678848
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9966681003570557,
+      "eval_runtime": 291.4303,
+      "eval_samples_per_second": 1309.373,
+      "eval_steps_per_second": 40.919,
+      "step": 1679040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.25037392808919e-05,
+      "loss": 3.8968,
+      "step": 1679360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.249536971218511e-05,
+      "loss": 3.8887,
+      "step": 1679872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2486983764674592e-05,
+      "loss": 3.9042,
+      "step": 1680384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.24786141959678e-05,
+      "loss": 3.9098,
+      "step": 1680896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.247022824845728e-05,
+      "loss": 3.9013,
+      "step": 1681408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2461858679750494e-05,
+      "loss": 3.8956,
+      "step": 1681920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.245347273223997e-05,
+      "loss": 3.8992,
+      "step": 1682432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2445086784729454e-05,
+      "loss": 3.8848,
+      "step": 1682944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2436700837218934e-05,
+      "loss": 3.9013,
+      "step": 1683456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2428314889708414e-05,
+      "loss": 3.9049,
+      "step": 1683968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2419928942197894e-05,
+      "loss": 3.9034,
+      "step": 1684480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2411542994687374e-05,
+      "loss": 3.904,
+      "step": 1684992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2403173425980583e-05,
+      "loss": 3.8991,
+      "step": 1685504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2394787478470063e-05,
+      "loss": 3.8947,
+      "step": 1686016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2386401530959546e-05,
+      "loss": 3.8891,
+      "step": 1686528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2378015583449026e-05,
+      "loss": 3.8851,
+      "step": 1687040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2369629635938502e-05,
+      "loss": 3.8962,
+      "step": 1687552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2361243688427982e-05,
+      "loss": 3.893,
+      "step": 1688064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2352857740917462e-05,
+      "loss": 3.8975,
+      "step": 1688576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2344471793406942e-05,
+      "loss": 3.9109,
+      "step": 1689088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2336102224700155e-05,
+      "loss": 3.9104,
+      "step": 1689600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2327732655993367e-05,
+      "loss": 3.9019,
+      "step": 1690112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2319346708482847e-05,
+      "loss": 3.8985,
+      "step": 1690624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2310960760972324e-05,
+      "loss": 3.9063,
+      "step": 1691136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2302574813461804e-05,
+      "loss": 3.891,
+      "step": 1691648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2294188865951284e-05,
+      "loss": 3.9068,
+      "step": 1692160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2285802918440764e-05,
+      "loss": 3.8943,
+      "step": 1692672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2277416970930247e-05,
+      "loss": 3.8843,
+      "step": 1693184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2269031023419727e-05,
+      "loss": 3.8969,
+      "step": 1693696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2260645075909207e-05,
+      "loss": 3.8922,
+      "step": 1694208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2252259128398684e-05,
+      "loss": 3.8963,
+      "step": 1694720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2243889559691896e-05,
+      "loss": 3.9042,
+      "step": 1695232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2235503612181376e-05,
+      "loss": 3.9014,
+      "step": 1695744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2227117664670856e-05,
+      "loss": 3.9011,
+      "step": 1696256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.221874809596407e-05,
+      "loss": 3.8937,
+      "step": 1696768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.221036214845355e-05,
+      "loss": 3.9112,
+      "step": 1697280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2201976200943028e-05,
+      "loss": 3.892,
+      "step": 1697792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2193590253432505e-05,
+      "loss": 3.8882,
+      "step": 1698304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2185204305921985e-05,
+      "loss": 3.894,
+      "step": 1698816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2176818358411465e-05,
+      "loss": 3.8887,
+      "step": 1699328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2168432410900948e-05,
+      "loss": 3.8982,
+      "step": 1699840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2160046463390428e-05,
+      "loss": 3.9066,
+      "step": 1700352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2151660515879908e-05,
+      "loss": 3.8918,
+      "step": 1700864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2143274568369388e-05,
+      "loss": 3.8974,
+      "step": 1701376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2134904999662597e-05,
+      "loss": 3.9063,
+      "step": 1701888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2126519052152077e-05,
+      "loss": 3.9041,
+      "step": 1702400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2118133104641557e-05,
+      "loss": 3.9033,
+      "step": 1702912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2109747157131037e-05,
+      "loss": 3.8924,
+      "step": 1703424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.210137758842425e-05,
+      "loss": 3.8672,
+      "step": 1703936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.209299164091373e-05,
+      "loss": 3.9111,
+      "step": 1704448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.208460569340321e-05,
+      "loss": 3.9004,
+      "step": 1704960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2076219745892686e-05,
+      "loss": 3.8991,
+      "step": 1705472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2067850177185902e-05,
+      "loss": 3.8907,
+      "step": 1705984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2059464229675382e-05,
+      "loss": 3.8886,
+      "step": 1706496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.205109466096859e-05,
+      "loss": 3.883,
+      "step": 1707008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.204270871345807e-05,
+      "loss": 3.8876,
+      "step": 1707520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.203432276594755e-05,
+      "loss": 3.8865,
+      "step": 1708032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.202593681843703e-05,
+      "loss": 3.8938,
+      "step": 1708544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.201755087092651e-05,
+      "loss": 3.9041,
+      "step": 1709056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.200916492341599e-05,
+      "loss": 3.8864,
+      "step": 1709568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.200077897590547e-05,
+      "loss": 3.8827,
+      "step": 1710080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.199239302839495e-05,
+      "loss": 3.8998,
+      "step": 1710592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.198400708088443e-05,
+      "loss": 3.8829,
+      "step": 1711104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.197562113337391e-05,
+      "loss": 3.8848,
+      "step": 1711616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.196723518586339e-05,
+      "loss": 3.8949,
+      "step": 1712128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.195884923835287e-05,
+      "loss": 3.8959,
+      "step": 1712640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1950479669646083e-05,
+      "loss": 3.8844,
+      "step": 1713152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1942093722135563e-05,
+      "loss": 3.8838,
+      "step": 1713664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1933707774625043e-05,
+      "loss": 3.8838,
+      "step": 1714176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.192532182711452e-05,
+      "loss": 3.8971,
+      "step": 1714688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1916935879604e-05,
+      "loss": 3.8958,
+      "step": 1715200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.190854993209348e-05,
+      "loss": 3.8931,
+      "step": 1715712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1900163984582963e-05,
+      "loss": 3.8908,
+      "step": 1716224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1891778037072443e-05,
+      "loss": 3.9034,
+      "step": 1716736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.188340846836565e-05,
+      "loss": 3.9008,
+      "step": 1717248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.187502252085513e-05,
+      "loss": 3.8825,
+      "step": 1717760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.186663657334461e-05,
+      "loss": 3.8979,
+      "step": 1718272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.185825062583409e-05,
+      "loss": 3.9024,
+      "step": 1718784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1849881057127304e-05,
+      "loss": 3.884,
+      "step": 1719296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1841511488420516e-05,
+      "loss": 3.8935,
+      "step": 1719808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1833125540909993e-05,
+      "loss": 3.895,
+      "step": 1720320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1824739593399473e-05,
+      "loss": 3.9002,
+      "step": 1720832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1816353645888953e-05,
+      "loss": 3.8942,
+      "step": 1721344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1807967698378433e-05,
+      "loss": 3.8733,
+      "step": 1721856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1799581750867913e-05,
+      "loss": 3.8903,
+      "step": 1722368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1791195803357396e-05,
+      "loss": 3.8853,
+      "step": 1722880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1782809855846873e-05,
+      "loss": 3.8959,
+      "step": 1723392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1774423908336353e-05,
+      "loss": 3.8895,
+      "step": 1723904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1766037960825833e-05,
+      "loss": 3.8895,
+      "step": 1724416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1757652013315313e-05,
+      "loss": 3.8928,
+      "step": 1724928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1749298823412258e-05,
+      "loss": 3.8957,
+      "step": 1725440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1740912875901738e-05,
+      "loss": 3.883,
+      "step": 1725952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1732526928391218e-05,
+      "loss": 3.8888,
+      "step": 1726464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1724140980880694e-05,
+      "loss": 3.8985,
+      "step": 1726976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1715755033370174e-05,
+      "loss": 3.9014,
+      "step": 1727488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1707369085859654e-05,
+      "loss": 3.8935,
+      "step": 1728000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1698983138349134e-05,
+      "loss": 3.8786,
+      "step": 1728512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1690597190838617e-05,
+      "loss": 3.891,
+      "step": 1729024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1682211243328097e-05,
+      "loss": 3.888,
+      "step": 1729536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1673841674621306e-05,
+      "loss": 3.9008,
+      "step": 1730048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1665455727110786e-05,
+      "loss": 3.9016,
+      "step": 1730560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1657069779600266e-05,
+      "loss": 3.8859,
+      "step": 1731072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1648683832089746e-05,
+      "loss": 3.8872,
+      "step": 1731584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1640297884579226e-05,
+      "loss": 3.8922,
+      "step": 1732096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1631911937068706e-05,
+      "loss": 3.8759,
+      "step": 1732608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1623525989558186e-05,
+      "loss": 3.8949,
+      "step": 1733120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1615140042047666e-05,
+      "loss": 3.8766,
+      "step": 1733632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1606770473340875e-05,
+      "loss": 3.8953,
+      "step": 1734144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1598384525830355e-05,
+      "loss": 3.884,
+      "step": 1734656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1589998578319835e-05,
+      "loss": 3.8918,
+      "step": 1735168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.158161263080932e-05,
+      "loss": 3.8881,
+      "step": 1735680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.15732266832988e-05,
+      "loss": 3.8875,
+      "step": 1736192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1564857114592007e-05,
+      "loss": 3.8937,
+      "step": 1736704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1556471167081487e-05,
+      "loss": 3.8862,
+      "step": 1737216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1548085219570967e-05,
+      "loss": 3.9063,
+      "step": 1737728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1539699272060447e-05,
+      "loss": 3.898,
+      "step": 1738240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1531313324549927e-05,
+      "loss": 3.8908,
+      "step": 1738752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.152292737703941e-05,
+      "loss": 3.8963,
+      "step": 1739264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1514541429528887e-05,
+      "loss": 3.8828,
+      "step": 1739776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1506155482018367e-05,
+      "loss": 3.8897,
+      "step": 1740288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.149778591331158e-05,
+      "loss": 3.8907,
+      "step": 1740800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.148941634460479e-05,
+      "loss": 3.8934,
+      "step": 1741312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1481030397094272e-05,
+      "loss": 3.891,
+      "step": 1741824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1472644449583752e-05,
+      "loss": 3.8886,
+      "step": 1742336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1464258502073232e-05,
+      "loss": 3.891,
+      "step": 1742848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.145588893336644e-05,
+      "loss": 3.8914,
+      "step": 1743360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.144751936465965e-05,
+      "loss": 3.8866,
+      "step": 1743872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1439133417149134e-05,
+      "loss": 3.8958,
+      "step": 1744384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1430747469638613e-05,
+      "loss": 3.8799,
+      "step": 1744896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1422377900931823e-05,
+      "loss": 3.9042,
+      "step": 1745408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1413991953421303e-05,
+      "loss": 3.8944,
+      "step": 1745920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1405606005910783e-05,
+      "loss": 3.8983,
+      "step": 1746432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1397220058400262e-05,
+      "loss": 3.8824,
+      "step": 1746944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1388834110889742e-05,
+      "loss": 3.8965,
+      "step": 1747456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1380448163379226e-05,
+      "loss": 3.8967,
+      "step": 1747968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1372062215868706e-05,
+      "loss": 3.897,
+      "step": 1748480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1363676268358182e-05,
+      "loss": 3.8861,
+      "step": 1748992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1355290320847662e-05,
+      "loss": 3.9028,
+      "step": 1749504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1346904373337142e-05,
+      "loss": 3.8949,
+      "step": 1750016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1338518425826622e-05,
+      "loss": 3.9045,
+      "step": 1750528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1330132478316102e-05,
+      "loss": 3.8849,
+      "step": 1751040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1321746530805582e-05,
+      "loss": 3.9002,
+      "step": 1751552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1313360583295065e-05,
+      "loss": 3.8954,
+      "step": 1752064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1304974635784542e-05,
+      "loss": 3.892,
+      "step": 1752576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1296605067077754e-05,
+      "loss": 3.8926,
+      "step": 1753088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1288219119567234e-05,
+      "loss": 3.8957,
+      "step": 1753600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.127983317205671e-05,
+      "loss": 3.8978,
+      "step": 1754112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1271447224546194e-05,
+      "loss": 3.8769,
+      "step": 1754624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1263061277035674e-05,
+      "loss": 3.899,
+      "step": 1755136
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9945592880249023,
+      "eval_runtime": 292.1251,
+      "eval_samples_per_second": 1306.259,
+      "eval_steps_per_second": 40.822,
+      "step": 1755360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1254675329525154e-05,
+      "loss": 3.8896,
+      "step": 1755648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1246289382014634e-05,
+      "loss": 3.8822,
+      "step": 1756160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1237903434504114e-05,
+      "loss": 3.8963,
+      "step": 1756672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1229517486993594e-05,
+      "loss": 3.9016,
+      "step": 1757184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.122113153948307e-05,
+      "loss": 3.8949,
+      "step": 1757696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.121274559197255e-05,
+      "loss": 3.8919,
+      "step": 1758208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1204359644462034e-05,
+      "loss": 3.8977,
+      "step": 1758720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1195973696951514e-05,
+      "loss": 3.8805,
+      "step": 1759232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1187604128244723e-05,
+      "loss": 3.8961,
+      "step": 1759744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1179218180734203e-05,
+      "loss": 3.897,
+      "step": 1760256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1170832233223683e-05,
+      "loss": 3.893,
+      "step": 1760768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1162446285713163e-05,
+      "loss": 3.8982,
+      "step": 1761280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1154076717006375e-05,
+      "loss": 3.8898,
+      "step": 1761792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1145690769495855e-05,
+      "loss": 3.887,
+      "step": 1762304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1137304821985335e-05,
+      "loss": 3.8846,
+      "step": 1762816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1128918874474815e-05,
+      "loss": 3.8782,
+      "step": 1763328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1120532926964295e-05,
+      "loss": 3.8899,
+      "step": 1763840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1112146979453775e-05,
+      "loss": 3.8861,
+      "step": 1764352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1103761031943252e-05,
+      "loss": 3.8895,
+      "step": 1764864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1095375084432735e-05,
+      "loss": 3.9001,
+      "step": 1765376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1086989136922215e-05,
+      "loss": 3.9049,
+      "step": 1765888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1078603189411695e-05,
+      "loss": 3.9002,
+      "step": 1766400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1070217241901175e-05,
+      "loss": 3.8866,
+      "step": 1766912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1061831294390655e-05,
+      "loss": 3.8971,
+      "step": 1767424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.105344534688013e-05,
+      "loss": 3.8882,
+      "step": 1767936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1045075778173344e-05,
+      "loss": 3.8976,
+      "step": 1768448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1036689830662827e-05,
+      "loss": 3.8886,
+      "step": 1768960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1028303883152307e-05,
+      "loss": 3.8795,
+      "step": 1769472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1019917935641784e-05,
+      "loss": 3.8922,
+      "step": 1769984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1011531988131264e-05,
+      "loss": 3.881,
+      "step": 1770496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1003146040620744e-05,
+      "loss": 3.8875,
+      "step": 1771008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0994760093110224e-05,
+      "loss": 3.8993,
+      "step": 1771520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0986374145599704e-05,
+      "loss": 3.8939,
+      "step": 1772032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0977988198089184e-05,
+      "loss": 3.8937,
+      "step": 1772544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0969618629382396e-05,
+      "loss": 3.8864,
+      "step": 1773056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0961249060675605e-05,
+      "loss": 3.9052,
+      "step": 1773568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0952863113165085e-05,
+      "loss": 3.8897,
+      "step": 1774080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0944477165654565e-05,
+      "loss": 3.88,
+      "step": 1774592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0936091218144045e-05,
+      "loss": 3.8906,
+      "step": 1775104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.092770527063353e-05,
+      "loss": 3.8807,
+      "step": 1775616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.091931932312301e-05,
+      "loss": 3.8896,
+      "step": 1776128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0910933375612488e-05,
+      "loss": 3.8949,
+      "step": 1776640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0902547428101965e-05,
+      "loss": 3.8911,
+      "step": 1777152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0894177859395177e-05,
+      "loss": 3.8862,
+      "step": 1777664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0885791911884657e-05,
+      "loss": 3.8999,
+      "step": 1778176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0877405964374137e-05,
+      "loss": 3.8964,
+      "step": 1778688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0869020016863617e-05,
+      "loss": 3.8988,
+      "step": 1779200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0860634069353097e-05,
+      "loss": 3.8836,
+      "step": 1779712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0852248121842577e-05,
+      "loss": 3.8589,
+      "step": 1780224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0843878553135786e-05,
+      "loss": 3.908,
+      "step": 1780736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0835492605625266e-05,
+      "loss": 3.8895,
+      "step": 1781248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.082710665811475e-05,
+      "loss": 3.8956,
+      "step": 1781760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.081872071060423e-05,
+      "loss": 3.8891,
+      "step": 1782272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.081033476309371e-05,
+      "loss": 3.8778,
+      "step": 1782784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.080194881558319e-05,
+      "loss": 3.8801,
+      "step": 1783296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.079356286807267e-05,
+      "loss": 3.8807,
+      "step": 1783808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0785176920562146e-05,
+      "loss": 3.8769,
+      "step": 1784320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0776790973051626e-05,
+      "loss": 3.8838,
+      "step": 1784832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.076842140434484e-05,
+      "loss": 3.902,
+      "step": 1785344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.076003545683432e-05,
+      "loss": 3.8774,
+      "step": 1785856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0751649509323798e-05,
+      "loss": 3.8747,
+      "step": 1786368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0743263561813278e-05,
+      "loss": 3.8965,
+      "step": 1786880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0734877614302758e-05,
+      "loss": 3.8744,
+      "step": 1787392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0726491666792238e-05,
+      "loss": 3.8789,
+      "step": 1787904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0718138476889183e-05,
+      "loss": 3.8874,
+      "step": 1788416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0709752529378663e-05,
+      "loss": 3.8888,
+      "step": 1788928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0701382960671872e-05,
+      "loss": 3.8774,
+      "step": 1789440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0692997013161352e-05,
+      "loss": 3.8838,
+      "step": 1789952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0684611065650832e-05,
+      "loss": 3.8732,
+      "step": 1790464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0676225118140312e-05,
+      "loss": 3.891,
+      "step": 1790976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0667839170629792e-05,
+      "loss": 3.89,
+      "step": 1791488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0659453223119272e-05,
+      "loss": 3.8845,
+      "step": 1792000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0651067275608752e-05,
+      "loss": 3.8877,
+      "step": 1792512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0642681328098232e-05,
+      "loss": 3.8915,
+      "step": 1793024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0634295380587712e-05,
+      "loss": 3.8999,
+      "step": 1793536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0625909433077192e-05,
+      "loss": 3.8766,
+      "step": 1794048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0617523485566672e-05,
+      "loss": 3.8901,
+      "step": 1794560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.060913753805615e-05,
+      "loss": 3.8953,
+      "step": 1795072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.060075159054563e-05,
+      "loss": 3.8725,
+      "step": 1795584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0592382021838844e-05,
+      "loss": 3.8928,
+      "step": 1796096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0583996074328324e-05,
+      "loss": 3.8862,
+      "step": 1796608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0575626505621533e-05,
+      "loss": 3.8957,
+      "step": 1797120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0567240558111013e-05,
+      "loss": 3.8916,
+      "step": 1797632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0558854610600493e-05,
+      "loss": 3.8725,
+      "step": 1798144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0550468663089973e-05,
+      "loss": 3.8759,
+      "step": 1798656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0542082715579453e-05,
+      "loss": 3.8825,
+      "step": 1799168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0533696768068933e-05,
+      "loss": 3.892,
+      "step": 1799680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0525327199362145e-05,
+      "loss": 3.8813,
+      "step": 1800192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0516941251851622e-05,
+      "loss": 3.8847,
+      "step": 1800704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0508555304341105e-05,
+      "loss": 3.8859,
+      "step": 1801216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0500169356830585e-05,
+      "loss": 3.8895,
+      "step": 1801728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0491783409320065e-05,
+      "loss": 3.8812,
+      "step": 1802240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0483397461809545e-05,
+      "loss": 3.881,
+      "step": 1802752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0475027893102754e-05,
+      "loss": 3.8936,
+      "step": 1803264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0466641945592234e-05,
+      "loss": 3.8953,
+      "step": 1803776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0458255998081714e-05,
+      "loss": 3.8885,
+      "step": 1804288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0449870050571198e-05,
+      "loss": 3.8747,
+      "step": 1804800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0441484103060678e-05,
+      "loss": 3.8855,
+      "step": 1805312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0433098155550154e-05,
+      "loss": 3.8794,
+      "step": 1805824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0424712208039634e-05,
+      "loss": 3.8951,
+      "step": 1806336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0416326260529114e-05,
+      "loss": 3.8885,
+      "step": 1806848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0407956691822326e-05,
+      "loss": 3.8836,
+      "step": 1807360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0399570744311806e-05,
+      "loss": 3.8819,
+      "step": 1807872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.039120117560502e-05,
+      "loss": 3.8804,
+      "step": 1808384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.03828152280945e-05,
+      "loss": 3.874,
+      "step": 1808896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.037442928058398e-05,
+      "loss": 3.8854,
+      "step": 1809408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0366043333073455e-05,
+      "loss": 3.8739,
+      "step": 1809920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0357657385562935e-05,
+      "loss": 3.8914,
+      "step": 1810432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0349271438052415e-05,
+      "loss": 3.8796,
+      "step": 1810944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.03408854905419e-05,
+      "loss": 3.8845,
+      "step": 1811456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.033249954303138e-05,
+      "loss": 3.8789,
+      "step": 1811968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0324129974324588e-05,
+      "loss": 3.8786,
+      "step": 1812480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.03157604056178e-05,
+      "loss": 3.8895,
+      "step": 1812992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0307374458107277e-05,
+      "loss": 3.8797,
+      "step": 1813504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.029898851059676e-05,
+      "loss": 3.9031,
+      "step": 1814016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.029060256308624e-05,
+      "loss": 3.8935,
+      "step": 1814528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.028221661557572e-05,
+      "loss": 3.8844,
+      "step": 1815040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.027384704686893e-05,
+      "loss": 3.8863,
+      "step": 1815552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.026546109935841e-05,
+      "loss": 3.879,
+      "step": 1816064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.025707515184789e-05,
+      "loss": 3.8816,
+      "step": 1816576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.024868920433737e-05,
+      "loss": 3.8896,
+      "step": 1817088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0240303256826852e-05,
+      "loss": 3.8852,
+      "step": 1817600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0231917309316332e-05,
+      "loss": 3.8844,
+      "step": 1818112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.022354774060954e-05,
+      "loss": 3.8862,
+      "step": 1818624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.021516179309902e-05,
+      "loss": 3.8856,
+      "step": 1819136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.02067758455885e-05,
+      "loss": 3.8849,
+      "step": 1819648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0198406276881714e-05,
+      "loss": 3.8768,
+      "step": 1820160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0190020329371194e-05,
+      "loss": 3.8885,
+      "step": 1820672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0181634381860674e-05,
+      "loss": 3.8768,
+      "step": 1821184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0173248434350154e-05,
+      "loss": 3.8996,
+      "step": 1821696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0164862486839634e-05,
+      "loss": 3.8877,
+      "step": 1822208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.015647653932911e-05,
+      "loss": 3.8907,
+      "step": 1822720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.014809059181859e-05,
+      "loss": 3.8788,
+      "step": 1823232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.013970464430807e-05,
+      "loss": 3.8896,
+      "step": 1823744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0131318696797553e-05,
+      "loss": 3.8886,
+      "step": 1824256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0122932749287033e-05,
+      "loss": 3.8891,
+      "step": 1824768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0114546801776513e-05,
+      "loss": 3.883,
+      "step": 1825280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.010616085426599e-05,
+      "loss": 3.8931,
+      "step": 1825792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0097791285559202e-05,
+      "loss": 3.8858,
+      "step": 1826304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0089405338048682e-05,
+      "loss": 3.9022,
+      "step": 1826816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0081019390538162e-05,
+      "loss": 3.8779,
+      "step": 1827328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0072633443027642e-05,
+      "loss": 3.8955,
+      "step": 1827840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0064263874320855e-05,
+      "loss": 3.8881,
+      "step": 1828352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0055877926810335e-05,
+      "loss": 3.8872,
+      "step": 1828864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.004749197929981e-05,
+      "loss": 3.8878,
+      "step": 1829376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.003910603178929e-05,
+      "loss": 3.8893,
+      "step": 1829888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0030736463082507e-05,
+      "loss": 3.891,
+      "step": 1830400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0022350515571987e-05,
+      "loss": 3.8717,
+      "step": 1830912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0013964568061464e-05,
+      "loss": 3.8906,
+      "step": 1831424
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.992262601852417,
+      "eval_runtime": 290.1775,
+      "eval_samples_per_second": 1315.026,
+      "eval_steps_per_second": 41.096,
+      "step": 1831680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005594999354676e-05,
+      "loss": 3.8779,
+      "step": 1831936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9997209051844156e-05,
+      "loss": 3.8777,
+      "step": 1832448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9988823104333636e-05,
+      "loss": 3.8889,
+      "step": 1832960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9980437156823116e-05,
+      "loss": 3.8978,
+      "step": 1833472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9972051209312596e-05,
+      "loss": 3.891,
+      "step": 1833984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9963665261802076e-05,
+      "loss": 3.8829,
+      "step": 1834496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9955295693095285e-05,
+      "loss": 3.8959,
+      "step": 1835008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9946909745584765e-05,
+      "loss": 3.8746,
+      "step": 1835520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9938523798074245e-05,
+      "loss": 3.8879,
+      "step": 1836032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9930137850563725e-05,
+      "loss": 3.8939,
+      "step": 1836544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9921751903053208e-05,
+      "loss": 3.8863,
+      "step": 1837056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9913365955542688e-05,
+      "loss": 3.8893,
+      "step": 1837568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9904996386835897e-05,
+      "loss": 3.8928,
+      "step": 1838080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9896610439325377e-05,
+      "loss": 3.8778,
+      "step": 1838592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9888224491814857e-05,
+      "loss": 3.8766,
+      "step": 1839104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9879838544304337e-05,
+      "loss": 3.8761,
+      "step": 1839616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9871452596793817e-05,
+      "loss": 3.8804,
+      "step": 1840128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9863066649283297e-05,
+      "loss": 3.884,
+      "step": 1840640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9854680701772777e-05,
+      "loss": 3.8801,
+      "step": 1841152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.984631113306599e-05,
+      "loss": 3.8935,
+      "step": 1841664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9837925185555466e-05,
+      "loss": 3.8996,
+      "step": 1842176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9829539238044946e-05,
+      "loss": 3.892,
+      "step": 1842688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.982115329053443e-05,
+      "loss": 3.8844,
+      "step": 1843200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.981276734302391e-05,
+      "loss": 3.8856,
+      "step": 1843712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.980439777431712e-05,
+      "loss": 3.8904,
+      "step": 1844224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.97960118268066e-05,
+      "loss": 3.8881,
+      "step": 1844736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9787625879296078e-05,
+      "loss": 3.8848,
+      "step": 1845248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9779239931785558e-05,
+      "loss": 3.877,
+      "step": 1845760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9770853984275038e-05,
+      "loss": 3.8837,
+      "step": 1846272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.976246803676452e-05,
+      "loss": 3.8785,
+      "step": 1846784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9754082089254e-05,
+      "loss": 3.88,
+      "step": 1847296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9745696141743478e-05,
+      "loss": 3.8926,
+      "step": 1847808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9737310194232958e-05,
+      "loss": 3.888,
+      "step": 1848320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.972894062552617e-05,
+      "loss": 3.8892,
+      "step": 1848832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9720554678015647e-05,
+      "loss": 3.8828,
+      "step": 1849344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9712185109308863e-05,
+      "loss": 3.8935,
+      "step": 1849856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9703799161798343e-05,
+      "loss": 3.8854,
+      "step": 1850368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9695413214287823e-05,
+      "loss": 3.8769,
+      "step": 1850880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.96870272667773e-05,
+      "loss": 3.8842,
+      "step": 1851392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.967864131926678e-05,
+      "loss": 3.8748,
+      "step": 1851904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9670271750559992e-05,
+      "loss": 3.8825,
+      "step": 1852416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9661885803049472e-05,
+      "loss": 3.8916,
+      "step": 1852928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9653499855538952e-05,
+      "loss": 3.8825,
+      "step": 1853440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9645113908028432e-05,
+      "loss": 3.8827,
+      "step": 1853952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.963672796051791e-05,
+      "loss": 3.8953,
+      "step": 1854464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.962834201300739e-05,
+      "loss": 3.8884,
+      "step": 1854976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.961995606549687e-05,
+      "loss": 3.8926,
+      "step": 1855488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.961157011798635e-05,
+      "loss": 3.8821,
+      "step": 1856000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9603216928083297e-05,
+      "loss": 3.8569,
+      "step": 1856512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9594830980572773e-05,
+      "loss": 3.8988,
+      "step": 1857024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9586445033062253e-05,
+      "loss": 3.8844,
+      "step": 1857536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9578059085551733e-05,
+      "loss": 3.8888,
+      "step": 1858048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9569673138041213e-05,
+      "loss": 3.8868,
+      "step": 1858560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9561287190530693e-05,
+      "loss": 3.8744,
+      "step": 1859072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9552917621823905e-05,
+      "loss": 3.8723,
+      "step": 1859584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9544531674313385e-05,
+      "loss": 3.8776,
+      "step": 1860096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9536145726802865e-05,
+      "loss": 3.8733,
+      "step": 1860608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9527759779292345e-05,
+      "loss": 3.8743,
+      "step": 1861120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9519373831781825e-05,
+      "loss": 3.895,
+      "step": 1861632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9510987884271302e-05,
+      "loss": 3.8737,
+      "step": 1862144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9502601936760785e-05,
+      "loss": 3.868,
+      "step": 1862656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9494215989250265e-05,
+      "loss": 3.8932,
+      "step": 1863168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9485830041739745e-05,
+      "loss": 3.8646,
+      "step": 1863680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9477460473032954e-05,
+      "loss": 3.871,
+      "step": 1864192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9469090904326167e-05,
+      "loss": 3.8862,
+      "step": 1864704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9460704956815647e-05,
+      "loss": 3.8844,
+      "step": 1865216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.945231900930513e-05,
+      "loss": 3.8724,
+      "step": 1865728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9443933061794607e-05,
+      "loss": 3.876,
+      "step": 1866240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9435547114284086e-05,
+      "loss": 3.87,
+      "step": 1866752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9427161166773566e-05,
+      "loss": 3.8859,
+      "step": 1867264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9418775219263046e-05,
+      "loss": 3.8843,
+      "step": 1867776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9410389271752526e-05,
+      "loss": 3.8804,
+      "step": 1868288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9402003324242006e-05,
+      "loss": 3.882,
+      "step": 1868800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.939365013433895e-05,
+      "loss": 3.8863,
+      "step": 1869312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9385264186828428e-05,
+      "loss": 3.8947,
+      "step": 1869824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9376878239317908e-05,
+      "loss": 3.8711,
+      "step": 1870336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9368492291807388e-05,
+      "loss": 3.8863,
+      "step": 1870848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9360106344296868e-05,
+      "loss": 3.8898,
+      "step": 1871360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9351720396786348e-05,
+      "loss": 3.8653,
+      "step": 1871872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.934335082807956e-05,
+      "loss": 3.8838,
+      "step": 1872384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9334981259372773e-05,
+      "loss": 3.8845,
+      "step": 1872896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.932659531186225e-05,
+      "loss": 3.8911,
+      "step": 1873408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.931820936435173e-05,
+      "loss": 3.8847,
+      "step": 1873920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.930982341684121e-05,
+      "loss": 3.87,
+      "step": 1874432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9301437469330693e-05,
+      "loss": 3.8651,
+      "step": 1874944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9293051521820172e-05,
+      "loss": 3.8777,
+      "step": 1875456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9284665574309652e-05,
+      "loss": 3.8867,
+      "step": 1875968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9276279626799132e-05,
+      "loss": 3.8757,
+      "step": 1876480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.926791005809234e-05,
+      "loss": 3.8768,
+      "step": 1876992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.925952411058182e-05,
+      "loss": 3.8787,
+      "step": 1877504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.92511381630713e-05,
+      "loss": 3.887,
+      "step": 1878016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.924275221556078e-05,
+      "loss": 3.8759,
+      "step": 1878528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.923436626805026e-05,
+      "loss": 3.8772,
+      "step": 1879040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.922598032053974e-05,
+      "loss": 3.8837,
+      "step": 1879552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9217610751832954e-05,
+      "loss": 3.8891,
+      "step": 1880064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.920922480432243e-05,
+      "loss": 3.8822,
+      "step": 1880576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.920083885681191e-05,
+      "loss": 3.8728,
+      "step": 1881088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9192452909301394e-05,
+      "loss": 3.8739,
+      "step": 1881600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9184066961790874e-05,
+      "loss": 3.8774,
+      "step": 1882112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9175681014280354e-05,
+      "loss": 3.8894,
+      "step": 1882624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9167295066769833e-05,
+      "loss": 3.8864,
+      "step": 1883136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9158909119259313e-05,
+      "loss": 3.879,
+      "step": 1883648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9150555929356255e-05,
+      "loss": 3.8741,
+      "step": 1884160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9142169981845735e-05,
+      "loss": 3.8716,
+      "step": 1884672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9133784034335215e-05,
+      "loss": 3.8744,
+      "step": 1885184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9125398086824695e-05,
+      "loss": 3.8792,
+      "step": 1885696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9117012139314175e-05,
+      "loss": 3.8657,
+      "step": 1886208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9108626191803655e-05,
+      "loss": 3.8877,
+      "step": 1886720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9100256623096864e-05,
+      "loss": 3.8748,
+      "step": 1887232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9091870675586347e-05,
+      "loss": 3.8762,
+      "step": 1887744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9083484728075827e-05,
+      "loss": 3.8733,
+      "step": 1888256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9075098780565307e-05,
+      "loss": 3.8725,
+      "step": 1888768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9066729211858516e-05,
+      "loss": 3.8862,
+      "step": 1889280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9058343264347996e-05,
+      "loss": 3.8717,
+      "step": 1889792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9049957316837476e-05,
+      "loss": 3.8988,
+      "step": 1890304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9041571369326956e-05,
+      "loss": 3.8877,
+      "step": 1890816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9033185421816436e-05,
+      "loss": 3.8819,
+      "step": 1891328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.902481585310965e-05,
+      "loss": 3.8793,
+      "step": 1891840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.901642990559913e-05,
+      "loss": 3.8783,
+      "step": 1892352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.900804395808861e-05,
+      "loss": 3.8757,
+      "step": 1892864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8999658010578085e-05,
+      "loss": 3.8855,
+      "step": 1893376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8991272063067565e-05,
+      "loss": 3.8767,
+      "step": 1893888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.898288611555705e-05,
+      "loss": 3.8821,
+      "step": 1894400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.897450016804653e-05,
+      "loss": 3.8773,
+      "step": 1894912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8966114220536008e-05,
+      "loss": 3.8797,
+      "step": 1895424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8957744651829217e-05,
+      "loss": 3.8798,
+      "step": 1895936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.894937508312243e-05,
+      "loss": 3.869,
+      "step": 1896448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.894098913561191e-05,
+      "loss": 3.8852,
+      "step": 1896960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.893260318810139e-05,
+      "loss": 3.8719,
+      "step": 1897472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.892421724059087e-05,
+      "loss": 3.8922,
+      "step": 1897984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.891583129308035e-05,
+      "loss": 3.8829,
+      "step": 1898496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.890744534556983e-05,
+      "loss": 3.8854,
+      "step": 1899008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.889905939805931e-05,
+      "loss": 3.8727,
+      "step": 1899520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.889067345054879e-05,
+      "loss": 3.8888,
+      "step": 1900032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8882303881842002e-05,
+      "loss": 3.8802,
+      "step": 1900544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8873917934331482e-05,
+      "loss": 3.8854,
+      "step": 1901056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8865531986820962e-05,
+      "loss": 3.8781,
+      "step": 1901568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8857146039310442e-05,
+      "loss": 3.8873,
+      "step": 1902080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.884877647060365e-05,
+      "loss": 3.8832,
+      "step": 1902592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.884039052309313e-05,
+      "loss": 3.8987,
+      "step": 1903104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.883200457558261e-05,
+      "loss": 3.8729,
+      "step": 1903616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.882361862807209e-05,
+      "loss": 3.8895,
+      "step": 1904128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.881523268056157e-05,
+      "loss": 3.8852,
+      "step": 1904640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8806863111854783e-05,
+      "loss": 3.8865,
+      "step": 1905152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8798477164344263e-05,
+      "loss": 3.8751,
+      "step": 1905664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.879009121683374e-05,
+      "loss": 3.8841,
+      "step": 1906176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8781705269323223e-05,
+      "loss": 3.8832,
+      "step": 1906688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8773335700616436e-05,
+      "loss": 3.876,
+      "step": 1907200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8764966131909645e-05,
+      "loss": 3.8842,
+      "step": 1907712
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9901106357574463,
+      "eval_runtime": 292.1233,
+      "eval_samples_per_second": 1306.267,
+      "eval_steps_per_second": 40.822,
+      "step": 1908000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8756580184399125e-05,
+      "loss": 3.8753,
+      "step": 1908224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8748194236888605e-05,
+      "loss": 3.8715,
+      "step": 1908736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8739808289378085e-05,
+      "loss": 3.8831,
+      "step": 1909248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8731422341867565e-05,
+      "loss": 3.8944,
+      "step": 1909760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8723036394357045e-05,
+      "loss": 3.8891,
+      "step": 1910272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8714650446846524e-05,
+      "loss": 3.874,
+      "step": 1910784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8706264499336004e-05,
+      "loss": 3.8908,
+      "step": 1911296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8697878551825484e-05,
+      "loss": 3.8698,
+      "step": 1911808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8689492604314964e-05,
+      "loss": 3.8795,
+      "step": 1912320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8681106656804444e-05,
+      "loss": 3.8909,
+      "step": 1912832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8672720709293924e-05,
+      "loss": 3.8801,
+      "step": 1913344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8664334761783404e-05,
+      "loss": 3.8841,
+      "step": 1913856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8655965193076617e-05,
+      "loss": 3.8871,
+      "step": 1914368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8647579245566093e-05,
+      "loss": 3.874,
+      "step": 1914880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8639193298055573e-05,
+      "loss": 3.8711,
+      "step": 1915392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8630807350545053e-05,
+      "loss": 3.8687,
+      "step": 1915904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8622421403034533e-05,
+      "loss": 3.8733,
+      "step": 1916416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8614035455524016e-05,
+      "loss": 3.8798,
+      "step": 1916928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8605649508013496e-05,
+      "loss": 3.8754,
+      "step": 1917440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8597263560502976e-05,
+      "loss": 3.888,
+      "step": 1917952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8588893991796185e-05,
+      "loss": 3.8926,
+      "step": 1918464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8580524423089395e-05,
+      "loss": 3.8903,
+      "step": 1918976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8572138475578878e-05,
+      "loss": 3.8845,
+      "step": 1919488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8563752528068358e-05,
+      "loss": 3.8776,
+      "step": 1920000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8555366580557838e-05,
+      "loss": 3.8863,
+      "step": 1920512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8546980633047318e-05,
+      "loss": 3.8791,
+      "step": 1921024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8538594685536798e-05,
+      "loss": 3.8756,
+      "step": 1921536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8530208738026274e-05,
+      "loss": 3.8761,
+      "step": 1922048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8521822790515754e-05,
+      "loss": 3.8766,
+      "step": 1922560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8513436843005234e-05,
+      "loss": 3.8747,
+      "step": 1923072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8505050895494718e-05,
+      "loss": 3.8743,
+      "step": 1923584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8496681326787927e-05,
+      "loss": 3.8841,
+      "step": 1924096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8488295379277407e-05,
+      "loss": 3.8898,
+      "step": 1924608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8479909431766887e-05,
+      "loss": 3.8807,
+      "step": 1925120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8471523484256367e-05,
+      "loss": 3.8818,
+      "step": 1925632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.846315391554958e-05,
+      "loss": 3.8821,
+      "step": 1926144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.845476796803906e-05,
+      "loss": 3.8864,
+      "step": 1926656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.844638202052854e-05,
+      "loss": 3.8719,
+      "step": 1927168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.843799607301802e-05,
+      "loss": 3.8787,
+      "step": 1927680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.84296101255075e-05,
+      "loss": 3.8671,
+      "step": 1928192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.842122417799698e-05,
+      "loss": 3.8819,
+      "step": 1928704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8412838230486455e-05,
+      "loss": 3.8802,
+      "step": 1929216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8404452282975935e-05,
+      "loss": 3.8789,
+      "step": 1929728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.839608271426915e-05,
+      "loss": 3.8807,
+      "step": 1930240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.838769676675863e-05,
+      "loss": 3.8855,
+      "step": 1930752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8379310819248108e-05,
+      "loss": 3.8829,
+      "step": 1931264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8370924871737588e-05,
+      "loss": 3.8871,
+      "step": 1931776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8362538924227068e-05,
+      "loss": 3.8791,
+      "step": 1932288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8354152976716548e-05,
+      "loss": 3.8506,
+      "step": 1932800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8345767029206028e-05,
+      "loss": 3.8896,
+      "step": 1933312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.833738108169551e-05,
+      "loss": 3.8785,
+      "step": 1933824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8329027891792453e-05,
+      "loss": 3.8849,
+      "step": 1934336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.832064194428193e-05,
+      "loss": 3.88,
+      "step": 1934848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.831225599677141e-05,
+      "loss": 3.8698,
+      "step": 1935360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.830387004926089e-05,
+      "loss": 3.8658,
+      "step": 1935872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8295500480554105e-05,
+      "loss": 3.87,
+      "step": 1936384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.828711453304358e-05,
+      "loss": 3.873,
+      "step": 1936896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.827872858553306e-05,
+      "loss": 3.8683,
+      "step": 1937408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.827034263802254e-05,
+      "loss": 3.8892,
+      "step": 1937920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.826195669051202e-05,
+      "loss": 3.8736,
+      "step": 1938432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.82535707430015e-05,
+      "loss": 3.8591,
+      "step": 1938944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.824518479549098e-05,
+      "loss": 3.8903,
+      "step": 1939456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8236815226784194e-05,
+      "loss": 3.8632,
+      "step": 1939968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8228429279273674e-05,
+      "loss": 3.8651,
+      "step": 1940480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8220043331763154e-05,
+      "loss": 3.8826,
+      "step": 1940992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8211657384252634e-05,
+      "loss": 3.8751,
+      "step": 1941504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.820327143674211e-05,
+      "loss": 3.8711,
+      "step": 1942016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.819488548923159e-05,
+      "loss": 3.8702,
+      "step": 1942528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8186515920524806e-05,
+      "loss": 3.8667,
+      "step": 1943040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8178129973014286e-05,
+      "loss": 3.8779,
+      "step": 1943552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8169744025503762e-05,
+      "loss": 3.8781,
+      "step": 1944064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8161358077993242e-05,
+      "loss": 3.8786,
+      "step": 1944576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8152972130482722e-05,
+      "loss": 3.8767,
+      "step": 1945088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8144602561775935e-05,
+      "loss": 3.8812,
+      "step": 1945600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8136216614265415e-05,
+      "loss": 3.8911,
+      "step": 1946112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8127830666754895e-05,
+      "loss": 3.8635,
+      "step": 1946624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8119444719244375e-05,
+      "loss": 3.8796,
+      "step": 1947136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8111058771733855e-05,
+      "loss": 3.8873,
+      "step": 1947648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8102672824223335e-05,
+      "loss": 3.8612,
+      "step": 1948160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8094286876712815e-05,
+      "loss": 3.8762,
+      "step": 1948672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8085917308006027e-05,
+      "loss": 3.8838,
+      "step": 1949184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8077531360495507e-05,
+      "loss": 3.8841,
+      "step": 1949696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8069145412984987e-05,
+      "loss": 3.8848,
+      "step": 1950208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8060759465474467e-05,
+      "loss": 3.8621,
+      "step": 1950720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8052389896767676e-05,
+      "loss": 3.8572,
+      "step": 1951232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8044003949257156e-05,
+      "loss": 3.8767,
+      "step": 1951744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8035618001746636e-05,
+      "loss": 3.8793,
+      "step": 1952256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8027232054236116e-05,
+      "loss": 3.8679,
+      "step": 1952768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.801886248552933e-05,
+      "loss": 3.8772,
+      "step": 1953280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.801047653801881e-05,
+      "loss": 3.8704,
+      "step": 1953792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.800209059050829e-05,
+      "loss": 3.8823,
+      "step": 1954304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7993704642997765e-05,
+      "loss": 3.8737,
+      "step": 1954816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.798533507429098e-05,
+      "loss": 3.871,
+      "step": 1955328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.797694912678046e-05,
+      "loss": 3.8741,
+      "step": 1955840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.796856317926994e-05,
+      "loss": 3.8847,
+      "step": 1956352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7960177231759417e-05,
+      "loss": 3.8789,
+      "step": 1956864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7951791284248897e-05,
+      "loss": 3.8678,
+      "step": 1957376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7943405336738377e-05,
+      "loss": 3.8691,
+      "step": 1957888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7935019389227857e-05,
+      "loss": 3.8717,
+      "step": 1958400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7926633441717337e-05,
+      "loss": 3.8824,
+      "step": 1958912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.791826387301055e-05,
+      "loss": 3.8788,
+      "step": 1959424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.790987792550003e-05,
+      "loss": 3.8745,
+      "step": 1959936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.790149197798951e-05,
+      "loss": 3.8725,
+      "step": 1960448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.789310603047899e-05,
+      "loss": 3.8661,
+      "step": 1960960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.78847364617722e-05,
+      "loss": 3.8737,
+      "step": 1961472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7876366893065414e-05,
+      "loss": 3.8688,
+      "step": 1961984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.786798094555489e-05,
+      "loss": 3.8641,
+      "step": 1962496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.785959499804437e-05,
+      "loss": 3.8832,
+      "step": 1963008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.785120905053385e-05,
+      "loss": 3.8696,
+      "step": 1963520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7842839481827063e-05,
+      "loss": 3.8744,
+      "step": 1964032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7834453534316543e-05,
+      "loss": 3.8659,
+      "step": 1964544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7826067586806023e-05,
+      "loss": 3.8648,
+      "step": 1965056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7817681639295503e-05,
+      "loss": 3.8797,
+      "step": 1965568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7809295691784983e-05,
+      "loss": 3.8691,
+      "step": 1966080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7800909744274463e-05,
+      "loss": 3.8902,
+      "step": 1966592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7792523796763943e-05,
+      "loss": 3.8822,
+      "step": 1967104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.778413784925342e-05,
+      "loss": 3.8776,
+      "step": 1967616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7775768280546636e-05,
+      "loss": 3.8761,
+      "step": 1968128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7767382333036115e-05,
+      "loss": 3.8743,
+      "step": 1968640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7758996385525595e-05,
+      "loss": 3.8696,
+      "step": 1969152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7750610438015072e-05,
+      "loss": 3.8822,
+      "step": 1969664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7742224490504552e-05,
+      "loss": 3.877,
+      "step": 1970176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7733838542994032e-05,
+      "loss": 3.8741,
+      "step": 1970688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7725452595483512e-05,
+      "loss": 3.8714,
+      "step": 1971200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7717066647972995e-05,
+      "loss": 3.875,
+      "step": 1971712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7708680700462475e-05,
+      "loss": 3.8748,
+      "step": 1972224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7700294752951952e-05,
+      "loss": 3.8668,
+      "step": 1972736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.769190880544143e-05,
+      "loss": 3.8814,
+      "step": 1973248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7683539236734644e-05,
+      "loss": 3.8602,
+      "step": 1973760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7675153289224124e-05,
+      "loss": 3.8958,
+      "step": 1974272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7666767341713604e-05,
+      "loss": 3.8742,
+      "step": 1974784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7658381394203084e-05,
+      "loss": 3.8788,
+      "step": 1975296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7649995446692564e-05,
+      "loss": 3.8672,
+      "step": 1975808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7641625877985773e-05,
+      "loss": 3.8862,
+      "step": 1976320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7633239930475253e-05,
+      "loss": 3.8751,
+      "step": 1976832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7624853982964733e-05,
+      "loss": 3.8803,
+      "step": 1977344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7616468035454213e-05,
+      "loss": 3.8764,
+      "step": 1977856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7608114845551158e-05,
+      "loss": 3.8775,
+      "step": 1978368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7599728898040638e-05,
+      "loss": 3.8781,
+      "step": 1978880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7591342950530118e-05,
+      "loss": 3.8947,
+      "step": 1979392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7582957003019598e-05,
+      "loss": 3.8675,
+      "step": 1979904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7574571055509074e-05,
+      "loss": 3.8837,
+      "step": 1980416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7566185107998558e-05,
+      "loss": 3.8842,
+      "step": 1980928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7557799160488038e-05,
+      "loss": 3.8792,
+      "step": 1981440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7549429591781247e-05,
+      "loss": 3.867,
+      "step": 1981952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7541043644270727e-05,
+      "loss": 3.8821,
+      "step": 1982464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7532657696760207e-05,
+      "loss": 3.8777,
+      "step": 1982976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7524271749249687e-05,
+      "loss": 3.8709,
+      "step": 1983488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.75159021805429e-05,
+      "loss": 3.877,
+      "step": 1984000
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9887855052948,
+      "eval_runtime": 295.0239,
+      "eval_samples_per_second": 1293.424,
+      "eval_steps_per_second": 40.42,
+      "step": 1984320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.750751623303238e-05,
+      "loss": 3.8675,
+      "step": 1984512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749913028552186e-05,
+      "loss": 3.8636,
+      "step": 1985024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749074433801134e-05,
+      "loss": 3.8793,
+      "step": 1985536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7482374769304548e-05,
+      "loss": 3.8872,
+      "step": 1986048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.747400520059776e-05,
+      "loss": 3.8852,
+      "step": 1986560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7465619253087244e-05,
+      "loss": 3.8718,
+      "step": 1987072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7457249684380453e-05,
+      "loss": 3.8818,
+      "step": 1987584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7448863736869933e-05,
+      "loss": 3.8707,
+      "step": 1988096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7440477789359413e-05,
+      "loss": 3.8716,
+      "step": 1988608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7432091841848893e-05,
+      "loss": 3.8844,
+      "step": 1989120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7423705894338373e-05,
+      "loss": 3.8754,
+      "step": 1989632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7415319946827853e-05,
+      "loss": 3.8825,
+      "step": 1990144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7406950378121065e-05,
+      "loss": 3.8846,
+      "step": 1990656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7398564430610545e-05,
+      "loss": 3.8681,
+      "step": 1991168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7390178483100022e-05,
+      "loss": 3.8651,
+      "step": 1991680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7381792535589502e-05,
+      "loss": 3.8615,
+      "step": 1992192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7373406588078982e-05,
+      "loss": 3.8727,
+      "step": 1992704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7365020640568465e-05,
+      "loss": 3.8766,
+      "step": 1993216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7356634693057945e-05,
+      "loss": 3.8671,
+      "step": 1993728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7348248745547425e-05,
+      "loss": 3.8831,
+      "step": 1994240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.73398627980369e-05,
+      "loss": 3.888,
+      "step": 1994752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.733147685052638e-05,
+      "loss": 3.8822,
+      "step": 1995264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.732309090301586e-05,
+      "loss": 3.8807,
+      "step": 1995776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.731470495550534e-05,
+      "loss": 3.8738,
+      "step": 1996288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.730631900799482e-05,
+      "loss": 3.8799,
+      "step": 1996800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7297933060484305e-05,
+      "loss": 3.8779,
+      "step": 1997312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7289563491777514e-05,
+      "loss": 3.8691,
+      "step": 1997824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7281177544266994e-05,
+      "loss": 3.8674,
+      "step": 1998336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7272791596756474e-05,
+      "loss": 3.8763,
+      "step": 1998848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7264422028049683e-05,
+      "loss": 3.8694,
+      "step": 1999360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7256036080539166e-05,
+      "loss": 3.8725,
+      "step": 1999872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7247650133028646e-05,
+      "loss": 3.88,
+      "step": 2000384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7239264185518126e-05,
+      "loss": 3.8796,
+      "step": 2000896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7230878238007606e-05,
+      "loss": 3.8834,
+      "step": 2001408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7222508669300815e-05,
+      "loss": 3.8706,
+      "step": 2001920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7214139100594028e-05,
+      "loss": 3.882,
+      "step": 2002432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7205753153083508e-05,
+      "loss": 3.8825,
+      "step": 2002944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7197367205572988e-05,
+      "loss": 3.8607,
+      "step": 2003456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7188981258062467e-05,
+      "loss": 3.8745,
+      "step": 2003968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7180595310551947e-05,
+      "loss": 3.8647,
+      "step": 2004480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7172209363041427e-05,
+      "loss": 3.8758,
+      "step": 2004992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7163823415530907e-05,
+      "loss": 3.8729,
+      "step": 2005504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7155437468020384e-05,
+      "loss": 3.8723,
+      "step": 2006016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7147051520509867e-05,
+      "loss": 3.8785,
+      "step": 2006528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.713868195180308e-05,
+      "loss": 3.8819,
+      "step": 2007040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7130296004292556e-05,
+      "loss": 3.8797,
+      "step": 2007552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7121910056782036e-05,
+      "loss": 3.8843,
+      "step": 2008064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7113524109271516e-05,
+      "loss": 3.8721,
+      "step": 2008576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7105138161760996e-05,
+      "loss": 3.8463,
+      "step": 2009088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7096752214250476e-05,
+      "loss": 3.8843,
+      "step": 2009600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.708836626673996e-05,
+      "loss": 3.8739,
+      "step": 2010112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.707998031922944e-05,
+      "loss": 3.8802,
+      "step": 2010624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7071594371718916e-05,
+      "loss": 3.8737,
+      "step": 2011136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7063208424208396e-05,
+      "loss": 3.868,
+      "step": 2011648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7054822476697876e-05,
+      "loss": 3.8587,
+      "step": 2012160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.704645290799109e-05,
+      "loss": 3.8641,
+      "step": 2012672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.703806696048057e-05,
+      "loss": 3.8702,
+      "step": 2013184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.702968101297005e-05,
+      "loss": 3.8573,
+      "step": 2013696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7021295065459528e-05,
+      "loss": 3.8863,
+      "step": 2014208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7012925496752737e-05,
+      "loss": 3.8689,
+      "step": 2014720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7004539549242217e-05,
+      "loss": 3.855,
+      "step": 2015232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6996153601731697e-05,
+      "loss": 3.8815,
+      "step": 2015744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6987767654221177e-05,
+      "loss": 3.8589,
+      "step": 2016256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.697938170671066e-05,
+      "loss": 3.8608,
+      "step": 2016768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.697099575920014e-05,
+      "loss": 3.8777,
+      "step": 2017280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.696260981168962e-05,
+      "loss": 3.8681,
+      "step": 2017792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.695424024298283e-05,
+      "loss": 3.8657,
+      "step": 2018304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.694587067427604e-05,
+      "loss": 3.8637,
+      "step": 2018816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6937484726765522e-05,
+      "loss": 3.8622,
+      "step": 2019328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6929098779255002e-05,
+      "loss": 3.8705,
+      "step": 2019840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6920712831744482e-05,
+      "loss": 3.8788,
+      "step": 2020352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6912326884233962e-05,
+      "loss": 3.8695,
+      "step": 2020864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6903940936723442e-05,
+      "loss": 3.8715,
+      "step": 2021376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.689555498921292e-05,
+      "loss": 3.8705,
+      "step": 2021888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.688718542050613e-05,
+      "loss": 3.8882,
+      "step": 2022400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6878799472995614e-05,
+      "loss": 3.8635,
+      "step": 2022912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6870413525485094e-05,
+      "loss": 3.8747,
+      "step": 2023424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.686202757797457e-05,
+      "loss": 3.8756,
+      "step": 2023936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.685364163046405e-05,
+      "loss": 3.8629,
+      "step": 2024448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.684525568295353e-05,
+      "loss": 3.8694,
+      "step": 2024960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.683686973544301e-05,
+      "loss": 3.8757,
+      "step": 2025472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.682848378793249e-05,
+      "loss": 3.8772,
+      "step": 2025984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6820114219225703e-05,
+      "loss": 3.8832,
+      "step": 2026496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6811728271715183e-05,
+      "loss": 3.8523,
+      "step": 2027008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6803358703008392e-05,
+      "loss": 3.8583,
+      "step": 2027520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6794972755497872e-05,
+      "loss": 3.8715,
+      "step": 2028032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6786586807987352e-05,
+      "loss": 3.8739,
+      "step": 2028544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6778200860476835e-05,
+      "loss": 3.8619,
+      "step": 2029056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6769831291770044e-05,
+      "loss": 3.8728,
+      "step": 2029568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6761445344259524e-05,
+      "loss": 3.8678,
+      "step": 2030080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6753059396749004e-05,
+      "loss": 3.8773,
+      "step": 2030592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6744673449238484e-05,
+      "loss": 3.8699,
+      "step": 2031104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6736287501727964e-05,
+      "loss": 3.8641,
+      "step": 2031616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6727901554217444e-05,
+      "loss": 3.8719,
+      "step": 2032128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6719515606706924e-05,
+      "loss": 3.8785,
+      "step": 2032640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6711129659196404e-05,
+      "loss": 3.8764,
+      "step": 2033152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6702743711685884e-05,
+      "loss": 3.8602,
+      "step": 2033664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6694357764175364e-05,
+      "loss": 3.8635,
+      "step": 2034176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6685988195468573e-05,
+      "loss": 3.8719,
+      "step": 2034688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6677602247958053e-05,
+      "loss": 3.8774,
+      "step": 2035200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6669216300447536e-05,
+      "loss": 3.8739,
+      "step": 2035712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6660830352937016e-05,
+      "loss": 3.8677,
+      "step": 2036224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6652444405426496e-05,
+      "loss": 3.8684,
+      "step": 2036736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6644058457915976e-05,
+      "loss": 3.8615,
+      "step": 2037248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6635688889209185e-05,
+      "loss": 3.8682,
+      "step": 2037760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6627302941698665e-05,
+      "loss": 3.8633,
+      "step": 2038272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6618916994188145e-05,
+      "loss": 3.8628,
+      "step": 2038784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.661053104667763e-05,
+      "loss": 3.877,
+      "step": 2039296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6602145099167105e-05,
+      "loss": 3.8653,
+      "step": 2039808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6593759151656585e-05,
+      "loss": 3.8677,
+      "step": 2040320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6585373204146065e-05,
+      "loss": 3.8631,
+      "step": 2040832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6577003635439278e-05,
+      "loss": 3.8618,
+      "step": 2041344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6568617687928754e-05,
+      "loss": 3.8738,
+      "step": 2041856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6560231740418238e-05,
+      "loss": 3.8703,
+      "step": 2042368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6551845792907718e-05,
+      "loss": 3.8797,
+      "step": 2042880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6543459845397197e-05,
+      "loss": 3.8765,
+      "step": 2043392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6535090276690407e-05,
+      "loss": 3.8743,
+      "step": 2043904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.652672070798362e-05,
+      "loss": 3.8721,
+      "step": 2044416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.65183347604731e-05,
+      "loss": 3.8703,
+      "step": 2044928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.650994881296258e-05,
+      "loss": 3.8653,
+      "step": 2045440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.650156286545206e-05,
+      "loss": 3.8766,
+      "step": 2045952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.649317691794154e-05,
+      "loss": 3.871,
+      "step": 2046464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.648479097043102e-05,
+      "loss": 3.8732,
+      "step": 2046976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.64764050229205e-05,
+      "loss": 3.8637,
+      "step": 2047488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.646801907540998e-05,
+      "loss": 3.8739,
+      "step": 2048000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.645963312789946e-05,
+      "loss": 3.8663,
+      "step": 2048512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.645124718038894e-05,
+      "loss": 3.8641,
+      "step": 2049024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.644287761168215e-05,
+      "loss": 3.8731,
+      "step": 2049536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.643449166417163e-05,
+      "loss": 3.8558,
+      "step": 2050048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6426105716661108e-05,
+      "loss": 3.8914,
+      "step": 2050560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6417719769150588e-05,
+      "loss": 3.8685,
+      "step": 2051072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6409333821640068e-05,
+      "loss": 3.879,
+      "step": 2051584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6400964252933283e-05,
+      "loss": 3.8639,
+      "step": 2052096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.639257830542276e-05,
+      "loss": 3.8775,
+      "step": 2052608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.638419235791224e-05,
+      "loss": 3.8704,
+      "step": 2053120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.637580641040172e-05,
+      "loss": 3.8766,
+      "step": 2053632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.63674204628912e-05,
+      "loss": 3.8704,
+      "step": 2054144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.635903451538068e-05,
+      "loss": 3.8699,
+      "step": 2054656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.635064856787016e-05,
+      "loss": 3.8747,
+      "step": 2055168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.634226262035964e-05,
+      "loss": 3.8929,
+      "step": 2055680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6333893051652852e-05,
+      "loss": 3.8576,
+      "step": 2056192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6325507104142332e-05,
+      "loss": 3.8835,
+      "step": 2056704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6317121156631812e-05,
+      "loss": 3.8793,
+      "step": 2057216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.630875158792502e-05,
+      "loss": 3.8722,
+      "step": 2057728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.63003656404145e-05,
+      "loss": 3.8647,
+      "step": 2058240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6291979692903985e-05,
+      "loss": 3.873,
+      "step": 2058752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6283593745393464e-05,
+      "loss": 3.8716,
+      "step": 2059264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.627520779788294e-05,
+      "loss": 3.8694,
+      "step": 2059776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.626682185037242e-05,
+      "loss": 3.871,
+      "step": 2060288
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9870848655700684,
+      "eval_runtime": 289.9516,
+      "eval_samples_per_second": 1316.051,
+      "eval_steps_per_second": 41.128,
+      "step": 2060640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.62584359028619e-05,
+      "loss": 3.8623,
+      "step": 2060800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.625004995535138e-05,
+      "loss": 3.8606,
+      "step": 2061312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.624166400784086e-05,
+      "loss": 3.8703,
+      "step": 2061824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.623327806033034e-05,
+      "loss": 3.8809,
+      "step": 2062336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6224892112819824e-05,
+      "loss": 3.8819,
+      "step": 2062848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.62165061653093e-05,
+      "loss": 3.8679,
+      "step": 2063360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.620812021779878e-05,
+      "loss": 3.8797,
+      "step": 2063872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.619973427028826e-05,
+      "loss": 3.8665,
+      "step": 2064384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.619134832277774e-05,
+      "loss": 3.8642,
+      "step": 2064896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.618296237526722e-05,
+      "loss": 3.8757,
+      "step": 2065408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.61745764277567e-05,
+      "loss": 3.8772,
+      "step": 2065920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.616619048024618e-05,
+      "loss": 3.8792,
+      "step": 2066432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6157820911539393e-05,
+      "loss": 3.8756,
+      "step": 2066944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6149434964028873e-05,
+      "loss": 3.8632,
+      "step": 2067456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6141049016518353e-05,
+      "loss": 3.8648,
+      "step": 2067968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.613266306900783e-05,
+      "loss": 3.8546,
+      "step": 2068480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.612427712149731e-05,
+      "loss": 3.8713,
+      "step": 2068992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6115891173986793e-05,
+      "loss": 3.8715,
+      "step": 2069504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6107505226476273e-05,
+      "loss": 3.8604,
+      "step": 2070016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6099119278965753e-05,
+      "loss": 3.8756,
+      "step": 2070528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6090733331455233e-05,
+      "loss": 3.8846,
+      "step": 2071040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6082347383944713e-05,
+      "loss": 3.8761,
+      "step": 2071552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.607396143643419e-05,
+      "loss": 3.8761,
+      "step": 2072064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.606557548892367e-05,
+      "loss": 3.8714,
+      "step": 2072576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6057205920216885e-05,
+      "loss": 3.8747,
+      "step": 2073088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.604881997270636e-05,
+      "loss": 3.8735,
+      "step": 2073600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.604043402519584e-05,
+      "loss": 3.8632,
+      "step": 2074112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6032064456489054e-05,
+      "loss": 3.8607,
+      "step": 2074624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6023678508978534e-05,
+      "loss": 3.8722,
+      "step": 2075136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.601529256146801e-05,
+      "loss": 3.8662,
+      "step": 2075648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6006906613957494e-05,
+      "loss": 3.8719,
+      "step": 2076160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5998520666446974e-05,
+      "loss": 3.8714,
+      "step": 2076672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5990151097740186e-05,
+      "loss": 3.8703,
+      "step": 2077184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5981765150229663e-05,
+      "loss": 3.8807,
+      "step": 2077696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5973379202719143e-05,
+      "loss": 3.8647,
+      "step": 2078208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5965009634012355e-05,
+      "loss": 3.8756,
+      "step": 2078720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5956623686501835e-05,
+      "loss": 3.8789,
+      "step": 2079232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5948237738991315e-05,
+      "loss": 3.8556,
+      "step": 2079744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5939851791480795e-05,
+      "loss": 3.8733,
+      "step": 2080256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5931465843970275e-05,
+      "loss": 3.8633,
+      "step": 2080768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5923079896459755e-05,
+      "loss": 3.8663,
+      "step": 2081280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5914693948949235e-05,
+      "loss": 3.8688,
+      "step": 2081792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5906308001438715e-05,
+      "loss": 3.87,
+      "step": 2082304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5897922053928195e-05,
+      "loss": 3.8731,
+      "step": 2082816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5889536106417675e-05,
+      "loss": 3.8759,
+      "step": 2083328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5881150158907155e-05,
+      "loss": 3.8773,
+      "step": 2083840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5872764211396635e-05,
+      "loss": 3.8776,
+      "step": 2084352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5864394642689844e-05,
+      "loss": 3.8692,
+      "step": 2084864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5856008695179324e-05,
+      "loss": 3.8436,
+      "step": 2085376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5847622747668807e-05,
+      "loss": 3.8793,
+      "step": 2085888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5839253178962016e-05,
+      "loss": 3.8683,
+      "step": 2086400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5830867231451496e-05,
+      "loss": 3.8779,
+      "step": 2086912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5822481283940976e-05,
+      "loss": 3.8696,
+      "step": 2087424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5814095336430456e-05,
+      "loss": 3.8596,
+      "step": 2087936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5805709388919936e-05,
+      "loss": 3.8629,
+      "step": 2088448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.579733982021315e-05,
+      "loss": 3.854,
+      "step": 2088960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.578895387270263e-05,
+      "loss": 3.8677,
+      "step": 2089472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.578056792519211e-05,
+      "loss": 3.8492,
+      "step": 2089984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.577218197768159e-05,
+      "loss": 3.8848,
+      "step": 2090496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.576379603017107e-05,
+      "loss": 3.8672,
+      "step": 2091008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5755410082660545e-05,
+      "loss": 3.8477,
+      "step": 2091520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5747040513953757e-05,
+      "loss": 3.878,
+      "step": 2092032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.573865456644324e-05,
+      "loss": 3.8546,
+      "step": 2092544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.573026861893272e-05,
+      "loss": 3.8566,
+      "step": 2093056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5721882671422197e-05,
+      "loss": 3.8676,
+      "step": 2093568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.571351310271541e-05,
+      "loss": 3.8654,
+      "step": 2094080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.570512715520489e-05,
+      "loss": 3.8604,
+      "step": 2094592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.569674120769437e-05,
+      "loss": 3.8646,
+      "step": 2095104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.568835526018385e-05,
+      "loss": 3.8559,
+      "step": 2095616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5679985691477062e-05,
+      "loss": 3.8683,
+      "step": 2096128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5671599743966542e-05,
+      "loss": 3.8731,
+      "step": 2096640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.566321379645602e-05,
+      "loss": 3.8631,
+      "step": 2097152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56548278489455e-05,
+      "loss": 3.8705,
+      "step": 2097664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.564644190143498e-05,
+      "loss": 3.8621,
+      "step": 2098176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5638055953924462e-05,
+      "loss": 3.8864,
+      "step": 2098688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5629670006413942e-05,
+      "loss": 3.8545,
+      "step": 2099200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5621284058903422e-05,
+      "loss": 3.8726,
+      "step": 2099712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5612898111392902e-05,
+      "loss": 3.864,
+      "step": 2100224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.560452854268611e-05,
+      "loss": 3.8663,
+      "step": 2100736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.559614259517559e-05,
+      "loss": 3.8632,
+      "step": 2101248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.558775664766507e-05,
+      "loss": 3.8714,
+      "step": 2101760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5579387078958283e-05,
+      "loss": 3.8755,
+      "step": 2102272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5571001131447763e-05,
+      "loss": 3.8779,
+      "step": 2102784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5562615183937243e-05,
+      "loss": 3.8483,
+      "step": 2103296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5554229236426723e-05,
+      "loss": 3.8517,
+      "step": 2103808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.55458432889162e-05,
+      "loss": 3.866,
+      "step": 2104320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.553745734140568e-05,
+      "loss": 3.8715,
+      "step": 2104832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5529071393895163e-05,
+      "loss": 3.8566,
+      "step": 2105344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5520701825188376e-05,
+      "loss": 3.8675,
+      "step": 2105856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5512315877677852e-05,
+      "loss": 3.8648,
+      "step": 2106368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5503929930167332e-05,
+      "loss": 3.8686,
+      "step": 2106880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5495543982656812e-05,
+      "loss": 3.8667,
+      "step": 2107392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5487158035146292e-05,
+      "loss": 3.8611,
+      "step": 2107904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5478772087635772e-05,
+      "loss": 3.8689,
+      "step": 2108416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5470386140125255e-05,
+      "loss": 3.8708,
+      "step": 2108928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5462000192614735e-05,
+      "loss": 3.8762,
+      "step": 2109440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5453630623907944e-05,
+      "loss": 3.8563,
+      "step": 2109952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5445244676397424e-05,
+      "loss": 3.8551,
+      "step": 2110464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5436858728886904e-05,
+      "loss": 3.8663,
+      "step": 2110976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5428489160180117e-05,
+      "loss": 3.8691,
+      "step": 2111488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5420103212669597e-05,
+      "loss": 3.872,
+      "step": 2112000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5411717265159077e-05,
+      "loss": 3.863,
+      "step": 2112512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5403331317648557e-05,
+      "loss": 3.8645,
+      "step": 2113024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5394961748941766e-05,
+      "loss": 3.8578,
+      "step": 2113536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5386575801431246e-05,
+      "loss": 3.8635,
+      "step": 2114048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5378189853920726e-05,
+      "loss": 3.8619,
+      "step": 2114560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.536980390641021e-05,
+      "loss": 3.8583,
+      "step": 2115072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5361417958899685e-05,
+      "loss": 3.8671,
+      "step": 2115584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5353032011389165e-05,
+      "loss": 3.8618,
+      "step": 2116096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5344662442682378e-05,
+      "loss": 3.8612,
+      "step": 2116608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5336276495171855e-05,
+      "loss": 3.8578,
+      "step": 2117120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5327890547661334e-05,
+      "loss": 3.8537,
+      "step": 2117632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5319504600150818e-05,
+      "loss": 3.8706,
+      "step": 2118144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.531113503144403e-05,
+      "loss": 3.8657,
+      "step": 2118656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5302749083933507e-05,
+      "loss": 3.8755,
+      "step": 2119168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5294363136422987e-05,
+      "loss": 3.8685,
+      "step": 2119680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5285977188912467e-05,
+      "loss": 3.8735,
+      "step": 2120192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.527760762020568e-05,
+      "loss": 3.8684,
+      "step": 2120704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.526922167269516e-05,
+      "loss": 3.8573,
+      "step": 2121216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.526083572518464e-05,
+      "loss": 3.8683,
+      "step": 2121728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5252449777674119e-05,
+      "loss": 3.8666,
+      "step": 2122240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5244063830163599e-05,
+      "loss": 3.8646,
+      "step": 2122752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5235677882653079e-05,
+      "loss": 3.8721,
+      "step": 2123264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5227291935142559e-05,
+      "loss": 3.8622,
+      "step": 2123776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5218905987632037e-05,
+      "loss": 3.867,
+      "step": 2124288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5210520040121517e-05,
+      "loss": 3.8596,
+      "step": 2124800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5202134092610997e-05,
+      "loss": 3.8626,
+      "step": 2125312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5193748145100479e-05,
+      "loss": 3.868,
+      "step": 2125824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5185378576393688e-05,
+      "loss": 3.8578,
+      "step": 2126336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.517699262888317e-05,
+      "loss": 3.8822,
+      "step": 2126848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.516860668137265e-05,
+      "loss": 3.8656,
+      "step": 2127360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.516022073386213e-05,
+      "loss": 3.8726,
+      "step": 2127872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.515183478635161e-05,
+      "loss": 3.8591,
+      "step": 2128384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.514346521764482e-05,
+      "loss": 3.8715,
+      "step": 2128896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.51350792701343e-05,
+      "loss": 3.8682,
+      "step": 2129408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.512669332262378e-05,
+      "loss": 3.8719,
+      "step": 2129920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.511830737511326e-05,
+      "loss": 3.8652,
+      "step": 2130432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5109937806406471e-05,
+      "loss": 3.8656,
+      "step": 2130944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.510155185889595e-05,
+      "loss": 3.87,
+      "step": 2131456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5093182290189162e-05,
+      "loss": 3.8878,
+      "step": 2131968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5084796342678642e-05,
+      "loss": 3.8591,
+      "step": 2132480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5076410395168122e-05,
+      "loss": 3.8778,
+      "step": 2132992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5068024447657603e-05,
+      "loss": 3.8761,
+      "step": 2133504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5059638500147083e-05,
+      "loss": 3.8642,
+      "step": 2134016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5051252552636563e-05,
+      "loss": 3.8686,
+      "step": 2134528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5042866605126041e-05,
+      "loss": 3.8654,
+      "step": 2135040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5034480657615521e-05,
+      "loss": 3.8669,
+      "step": 2135552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5026094710105001e-05,
+      "loss": 3.8666,
+      "step": 2136064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5017708762594481e-05,
+      "loss": 3.867,
+      "step": 2136576
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9856343269348145,
+      "eval_runtime": 296.3366,
+      "eval_samples_per_second": 1287.694,
+      "eval_steps_per_second": 40.241,
+      "step": 2136960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5009322815083963e-05,
+      "loss": 3.8647,
+      "step": 2137088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5000936867573443e-05,
+      "loss": 3.8515,
+      "step": 2137600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4992567298866652e-05,
+      "loss": 3.8651,
+      "step": 2138112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4984181351356134e-05,
+      "loss": 3.8809,
+      "step": 2138624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4975795403845614e-05,
+      "loss": 3.8723,
+      "step": 2139136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4967409456335093e-05,
+      "loss": 3.8653,
+      "step": 2139648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4959023508824573e-05,
+      "loss": 3.8805,
+      "step": 2140160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4950637561314052e-05,
+      "loss": 3.8592,
+      "step": 2140672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4942267992607264e-05,
+      "loss": 3.8588,
+      "step": 2141184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4933882045096744e-05,
+      "loss": 3.8717,
+      "step": 2141696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4925496097586222e-05,
+      "loss": 3.8727,
+      "step": 2142208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4917110150075702e-05,
+      "loss": 3.8709,
+      "step": 2142720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4908740581368917e-05,
+      "loss": 3.8769,
+      "step": 2143232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4900354633858396e-05,
+      "loss": 3.8571,
+      "step": 2143744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4891968686347873e-05,
+      "loss": 3.8569,
+      "step": 2144256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4883582738837355e-05,
+      "loss": 3.8511,
+      "step": 2144768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4875196791326835e-05,
+      "loss": 3.8682,
+      "step": 2145280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4866810843816315e-05,
+      "loss": 3.8686,
+      "step": 2145792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4858424896305795e-05,
+      "loss": 3.8605,
+      "step": 2146304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4850038948795275e-05,
+      "loss": 3.8653,
+      "step": 2146816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4841653001284756e-05,
+      "loss": 3.8853,
+      "step": 2147328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4833283432577965e-05,
+      "loss": 3.8667,
+      "step": 2147840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4824897485067445e-05,
+      "loss": 3.8732,
+      "step": 2148352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4816511537556927e-05,
+      "loss": 3.8679,
+      "step": 2148864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4808125590046403e-05,
+      "loss": 3.8731,
+      "step": 2149376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4799739642535883e-05,
+      "loss": 3.8677,
+      "step": 2149888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4791370073829098e-05,
+      "loss": 3.8624,
+      "step": 2150400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4782984126318578e-05,
+      "loss": 3.8544,
+      "step": 2150912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4774598178808056e-05,
+      "loss": 3.8691,
+      "step": 2151424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4766212231297536e-05,
+      "loss": 3.8559,
+      "step": 2151936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4757826283787016e-05,
+      "loss": 3.8664,
+      "step": 2152448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4749456715080228e-05,
+      "loss": 3.8661,
+      "step": 2152960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4741070767569706e-05,
+      "loss": 3.8713,
+      "step": 2153472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4732684820059186e-05,
+      "loss": 3.8733,
+      "step": 2153984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4724298872548666e-05,
+      "loss": 3.8647,
+      "step": 2154496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4715929303841877e-05,
+      "loss": 3.8701,
+      "step": 2155008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4707543356331357e-05,
+      "loss": 3.8733,
+      "step": 2155520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4699157408820837e-05,
+      "loss": 3.8506,
+      "step": 2156032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4690771461310319e-05,
+      "loss": 3.867,
+      "step": 2156544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4682385513799799e-05,
+      "loss": 3.8584,
+      "step": 2157056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4673999566289279e-05,
+      "loss": 3.8611,
+      "step": 2157568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4665613618778759e-05,
+      "loss": 3.8692,
+      "step": 2158080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4657227671268237e-05,
+      "loss": 3.8671,
+      "step": 2158592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.464885810256145e-05,
+      "loss": 3.8687,
+      "step": 2159104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.464047215505093e-05,
+      "loss": 3.8696,
+      "step": 2159616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4632086207540408e-05,
+      "loss": 3.8721,
+      "step": 2160128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4623700260029888e-05,
+      "loss": 3.8733,
+      "step": 2160640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4615330691323102e-05,
+      "loss": 3.8669,
+      "step": 2161152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4606944743812582e-05,
+      "loss": 3.8384,
+      "step": 2161664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.459857517510579e-05,
+      "loss": 3.8712,
+      "step": 2162176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4590189227595272e-05,
+      "loss": 3.8652,
+      "step": 2162688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4581803280084752e-05,
+      "loss": 3.872,
+      "step": 2163200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4573417332574232e-05,
+      "loss": 3.8678,
+      "step": 2163712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4565047763867443e-05,
+      "loss": 3.8582,
+      "step": 2164224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4556661816356923e-05,
+      "loss": 3.8508,
+      "step": 2164736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4548275868846403e-05,
+      "loss": 3.853,
+      "step": 2165248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4539906300139614e-05,
+      "loss": 3.8611,
+      "step": 2165760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4531520352629094e-05,
+      "loss": 3.845,
+      "step": 2166272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4523134405118574e-05,
+      "loss": 3.881,
+      "step": 2166784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4514748457608054e-05,
+      "loss": 3.8636,
+      "step": 2167296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4506362510097532e-05,
+      "loss": 3.8453,
+      "step": 2167808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4497976562587012e-05,
+      "loss": 3.8697,
+      "step": 2168320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4489590615076492e-05,
+      "loss": 3.8541,
+      "step": 2168832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4481204667565973e-05,
+      "loss": 3.853,
+      "step": 2169344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4472818720055453e-05,
+      "loss": 3.861,
+      "step": 2169856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4464432772544933e-05,
+      "loss": 3.864,
+      "step": 2170368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4456046825034413e-05,
+      "loss": 3.851,
+      "step": 2170880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4447660877523892e-05,
+      "loss": 3.8663,
+      "step": 2171392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4439291308817104e-05,
+      "loss": 3.8501,
+      "step": 2171904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4430905361306584e-05,
+      "loss": 3.8595,
+      "step": 2172416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4422552171403527e-05,
+      "loss": 3.8686,
+      "step": 2172928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4414166223893006e-05,
+      "loss": 3.8614,
+      "step": 2173440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4405780276382486e-05,
+      "loss": 3.8622,
+      "step": 2173952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4397394328871966e-05,
+      "loss": 3.861,
+      "step": 2174464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4389008381361445e-05,
+      "loss": 3.8833,
+      "step": 2174976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4380622433850927e-05,
+      "loss": 3.8465,
+      "step": 2175488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4372236486340407e-05,
+      "loss": 3.8685,
+      "step": 2176000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4363850538829887e-05,
+      "loss": 3.8581,
+      "step": 2176512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4355464591319365e-05,
+      "loss": 3.8603,
+      "step": 2177024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4347078643808845e-05,
+      "loss": 3.862,
+      "step": 2177536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4338725453905789e-05,
+      "loss": 3.8615,
+      "step": 2178048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4330339506395269e-05,
+      "loss": 3.8759,
+      "step": 2178560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4321953558884749e-05,
+      "loss": 3.8741,
+      "step": 2179072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4313567611374228e-05,
+      "loss": 3.8459,
+      "step": 2179584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.430518166386371e-05,
+      "loss": 3.8489,
+      "step": 2180096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4296795716353187e-05,
+      "loss": 3.8549,
+      "step": 2180608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.42884261476464e-05,
+      "loss": 3.8692,
+      "step": 2181120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.428004020013588e-05,
+      "loss": 3.8551,
+      "step": 2181632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.427165425262536e-05,
+      "loss": 3.8598,
+      "step": 2182144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4263284683918572e-05,
+      "loss": 3.8656,
+      "step": 2182656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4254898736408052e-05,
+      "loss": 3.862,
+      "step": 2183168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4246512788897531e-05,
+      "loss": 3.8654,
+      "step": 2183680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.423812684138701e-05,
+      "loss": 3.855,
+      "step": 2184192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.422974089387649e-05,
+      "loss": 3.8645,
+      "step": 2184704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.422135494636597e-05,
+      "loss": 3.8651,
+      "step": 2185216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.421296899885545e-05,
+      "loss": 3.872,
+      "step": 2185728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.420458305134493e-05,
+      "loss": 3.8527,
+      "step": 2186240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4196197103834411e-05,
+      "loss": 3.8539,
+      "step": 2186752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4187811156323891e-05,
+      "loss": 3.8623,
+      "step": 2187264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4179425208813368e-05,
+      "loss": 3.8635,
+      "step": 2187776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.417103926130285e-05,
+      "loss": 3.8727,
+      "step": 2188288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.416265331379233e-05,
+      "loss": 3.8569,
+      "step": 2188800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4154283745085542e-05,
+      "loss": 3.8592,
+      "step": 2189312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.414589779757502e-05,
+      "loss": 3.8547,
+      "step": 2189824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.41375118500645e-05,
+      "loss": 3.853,
+      "step": 2190336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.412912590255398e-05,
+      "loss": 3.8594,
+      "step": 2190848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.412073995504346e-05,
+      "loss": 3.8582,
+      "step": 2191360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.411237038633667e-05,
+      "loss": 3.8626,
+      "step": 2191872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.410398443882615e-05,
+      "loss": 3.8578,
+      "step": 2192384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.409559849131563e-05,
+      "loss": 3.8564,
+      "step": 2192896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4087212543805112e-05,
+      "loss": 3.8586,
+      "step": 2193408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4078826596294592e-05,
+      "loss": 3.8486,
+      "step": 2193920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4070457027587803e-05,
+      "loss": 3.8652,
+      "step": 2194432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4062071080077283e-05,
+      "loss": 3.8653,
+      "step": 2194944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4053685132566763e-05,
+      "loss": 3.8659,
+      "step": 2195456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4045299185056243e-05,
+      "loss": 3.8687,
+      "step": 2195968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4036913237545721e-05,
+      "loss": 3.8674,
+      "step": 2196480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4028543668838934e-05,
+      "loss": 3.8614,
+      "step": 2196992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4020157721328414e-05,
+      "loss": 3.855,
+      "step": 2197504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4011771773817894e-05,
+      "loss": 3.8669,
+      "step": 2198016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4003385826307372e-05,
+      "loss": 3.8585,
+      "step": 2198528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3995016257600584e-05,
+      "loss": 3.8601,
+      "step": 2199040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3986630310090066e-05,
+      "loss": 3.8656,
+      "step": 2199552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3978244362579546e-05,
+      "loss": 3.857,
+      "step": 2200064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3969858415069022e-05,
+      "loss": 3.8633,
+      "step": 2200576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3961472467558504e-05,
+      "loss": 3.8529,
+      "step": 2201088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3953102898851717e-05,
+      "loss": 3.8577,
+      "step": 2201600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3944716951341195e-05,
+      "loss": 3.864,
+      "step": 2202112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3936331003830675e-05,
+      "loss": 3.855,
+      "step": 2202624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3927945056320155e-05,
+      "loss": 3.8731,
+      "step": 2203136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3919559108809635e-05,
+      "loss": 3.8637,
+      "step": 2203648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3911173161299115e-05,
+      "loss": 3.8666,
+      "step": 2204160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3902803592592325e-05,
+      "loss": 3.8566,
+      "step": 2204672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3894417645081805e-05,
+      "loss": 3.8622,
+      "step": 2205184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3886031697571285e-05,
+      "loss": 3.8672,
+      "step": 2205696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3877645750060767e-05,
+      "loss": 3.8673,
+      "step": 2206208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3869259802550247e-05,
+      "loss": 3.8625,
+      "step": 2206720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3860873855039727e-05,
+      "loss": 3.8577,
+      "step": 2207232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3852504286332938e-05,
+      "loss": 3.8672,
+      "step": 2207744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3844134717626147e-05,
+      "loss": 3.8799,
+      "step": 2208256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3835748770115629e-05,
+      "loss": 3.8596,
+      "step": 2208768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3827362822605108e-05,
+      "loss": 3.8692,
+      "step": 2209280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3818976875094588e-05,
+      "loss": 3.8724,
+      "step": 2209792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3810590927584068e-05,
+      "loss": 3.8628,
+      "step": 2210304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.380220498007355e-05,
+      "loss": 3.8634,
+      "step": 2210816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3793819032563027e-05,
+      "loss": 3.8592,
+      "step": 2211328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3785433085052507e-05,
+      "loss": 3.8625,
+      "step": 2211840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3777047137541988e-05,
+      "loss": 3.8638,
+      "step": 2212352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3768661190031468e-05,
+      "loss": 3.8583,
+      "step": 2212864
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9842782020568848,
+      "eval_runtime": 296.6564,
+      "eval_samples_per_second": 1286.306,
+      "eval_steps_per_second": 40.198,
+      "step": 2213280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3760275242520948e-05,
+      "loss": 3.8733,
+      "step": 2213376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3751889295010428e-05,
+      "loss": 3.8476,
+      "step": 2213888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3743503347499908e-05,
+      "loss": 3.8613,
+      "step": 2214400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3735117399989386e-05,
+      "loss": 3.8798,
+      "step": 2214912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3726731452478866e-05,
+      "loss": 3.8681,
+      "step": 2215424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3718345504968346e-05,
+      "loss": 3.8606,
+      "step": 2215936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3709959557457828e-05,
+      "loss": 3.8696,
+      "step": 2216448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3701573609947308e-05,
+      "loss": 3.8597,
+      "step": 2216960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3693187662436788e-05,
+      "loss": 3.8547,
+      "step": 2217472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3684801714926266e-05,
+      "loss": 3.8714,
+      "step": 2217984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3676415767415746e-05,
+      "loss": 3.864,
+      "step": 2218496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3668029819905226e-05,
+      "loss": 3.87,
+      "step": 2219008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3659660251198438e-05,
+      "loss": 3.8722,
+      "step": 2219520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3651274303687917e-05,
+      "loss": 3.8522,
+      "step": 2220032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3642888356177397e-05,
+      "loss": 3.8539,
+      "step": 2220544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3634502408666877e-05,
+      "loss": 3.8476,
+      "step": 2221056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3626116461156358e-05,
+      "loss": 3.8638,
+      "step": 2221568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3617730513645838e-05,
+      "loss": 3.8632,
+      "step": 2222080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3609344566135318e-05,
+      "loss": 3.8574,
+      "step": 2222592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3600958618624798e-05,
+      "loss": 3.8573,
+      "step": 2223104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3592572671114276e-05,
+      "loss": 3.8829,
+      "step": 2223616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3584186723603756e-05,
+      "loss": 3.8615,
+      "step": 2224128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3575817154896969e-05,
+      "loss": 3.8685,
+      "step": 2224640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3567431207386447e-05,
+      "loss": 3.8642,
+      "step": 2225152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3559045259875927e-05,
+      "loss": 3.8693,
+      "step": 2225664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3550659312365407e-05,
+      "loss": 3.8607,
+      "step": 2226176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3542289743658621e-05,
+      "loss": 3.8589,
+      "step": 2226688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.353392017495183e-05,
+      "loss": 3.8566,
+      "step": 2227200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3525534227441312e-05,
+      "loss": 3.8582,
+      "step": 2227712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3517148279930792e-05,
+      "loss": 3.8547,
+      "step": 2228224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3508762332420268e-05,
+      "loss": 3.8622,
+      "step": 2228736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3500376384909748e-05,
+      "loss": 3.8578,
+      "step": 2229248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.349199043739923e-05,
+      "loss": 3.8708,
+      "step": 2229760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.348360448988871e-05,
+      "loss": 3.8651,
+      "step": 2230272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.347521854237819e-05,
+      "loss": 3.8609,
+      "step": 2230784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34668489736714e-05,
+      "loss": 3.8716,
+      "step": 2231296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.345846302616088e-05,
+      "loss": 3.8685,
+      "step": 2231808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.345007707865036e-05,
+      "loss": 3.8442,
+      "step": 2232320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.344169113113984e-05,
+      "loss": 3.8641,
+      "step": 2232832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3433305183629322e-05,
+      "loss": 3.853,
+      "step": 2233344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3424919236118802e-05,
+      "loss": 3.8562,
+      "step": 2233856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3416549667412013e-05,
+      "loss": 3.8661,
+      "step": 2234368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3408163719901493e-05,
+      "loss": 3.8663,
+      "step": 2234880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3399777772390973e-05,
+      "loss": 3.8604,
+      "step": 2235392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3391391824880451e-05,
+      "loss": 3.8679,
+      "step": 2235904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3383005877369931e-05,
+      "loss": 3.8674,
+      "step": 2236416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3374619929859411e-05,
+      "loss": 3.8671,
+      "step": 2236928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3366233982348891e-05,
+      "loss": 3.8624,
+      "step": 2237440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3357848034838371e-05,
+      "loss": 3.8372,
+      "step": 2237952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3349462087327853e-05,
+      "loss": 3.8607,
+      "step": 2238464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3341076139817333e-05,
+      "loss": 3.8665,
+      "step": 2238976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3332690192306809e-05,
+      "loss": 3.8644,
+      "step": 2239488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.332430424479629e-05,
+      "loss": 3.8614,
+      "step": 2240000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3315934676089503e-05,
+      "loss": 3.8559,
+      "step": 2240512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3307565107382714e-05,
+      "loss": 3.8483,
+      "step": 2241024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3299179159872194e-05,
+      "loss": 3.8456,
+      "step": 2241536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3290793212361674e-05,
+      "loss": 3.8623,
+      "step": 2242048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3282407264851154e-05,
+      "loss": 3.8435,
+      "step": 2242560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3274021317340632e-05,
+      "loss": 3.8693,
+      "step": 2243072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3265635369830112e-05,
+      "loss": 3.862,
+      "step": 2243584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3257249422319592e-05,
+      "loss": 3.846,
+      "step": 2244096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3248863474809072e-05,
+      "loss": 3.8634,
+      "step": 2244608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3240493906102283e-05,
+      "loss": 3.8473,
+      "step": 2245120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3232107958591763e-05,
+      "loss": 3.8527,
+      "step": 2245632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3223722011081244e-05,
+      "loss": 3.8546,
+      "step": 2246144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3215336063570724e-05,
+      "loss": 3.8638,
+      "step": 2246656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3206966494863934e-05,
+      "loss": 3.8498,
+      "step": 2247168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3198580547353415e-05,
+      "loss": 3.8601,
+      "step": 2247680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3190194599842895e-05,
+      "loss": 3.8443,
+      "step": 2248192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3181808652332375e-05,
+      "loss": 3.8572,
+      "step": 2248704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3173422704821855e-05,
+      "loss": 3.8629,
+      "step": 2249216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3165069514918798e-05,
+      "loss": 3.8595,
+      "step": 2249728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3156683567408278e-05,
+      "loss": 3.8548,
+      "step": 2250240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3148297619897757e-05,
+      "loss": 3.8616,
+      "step": 2250752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3139911672387237e-05,
+      "loss": 3.8766,
+      "step": 2251264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3131525724876716e-05,
+      "loss": 3.8443,
+      "step": 2251776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3123139777366198e-05,
+      "loss": 3.8672,
+      "step": 2252288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3114753829855678e-05,
+      "loss": 3.8556,
+      "step": 2252800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3106367882345158e-05,
+      "loss": 3.8534,
+      "step": 2253312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3098014692442101e-05,
+      "loss": 3.8555,
+      "step": 2253824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.308964512373531e-05,
+      "loss": 3.8569,
+      "step": 2254336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3081259176224792e-05,
+      "loss": 3.8668,
+      "step": 2254848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3072873228714272e-05,
+      "loss": 3.8722,
+      "step": 2255360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3064487281203752e-05,
+      "loss": 3.8457,
+      "step": 2255872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.305610133369323e-05,
+      "loss": 3.8447,
+      "step": 2256384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3047731764986443e-05,
+      "loss": 3.8519,
+      "step": 2256896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3039345817475923e-05,
+      "loss": 3.8635,
+      "step": 2257408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3030959869965403e-05,
+      "loss": 3.8501,
+      "step": 2257920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3022573922454881e-05,
+      "loss": 3.8554,
+      "step": 2258432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3014187974944361e-05,
+      "loss": 3.8633,
+      "step": 2258944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3005802027433841e-05,
+      "loss": 3.8537,
+      "step": 2259456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2997416079923323e-05,
+      "loss": 3.8647,
+      "step": 2259968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2989046511216532e-05,
+      "loss": 3.8491,
+      "step": 2260480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2980660563706013e-05,
+      "loss": 3.8613,
+      "step": 2260992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2972274616195493e-05,
+      "loss": 3.8549,
+      "step": 2261504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2963888668684973e-05,
+      "loss": 3.8695,
+      "step": 2262016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2955502721174453e-05,
+      "loss": 3.8468,
+      "step": 2262528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2947116773663933e-05,
+      "loss": 3.8494,
+      "step": 2263040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2938730826153411e-05,
+      "loss": 3.864,
+      "step": 2263552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2930344878642891e-05,
+      "loss": 3.8577,
+      "step": 2264064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2921958931132371e-05,
+      "loss": 3.8661,
+      "step": 2264576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2913589362425582e-05,
+      "loss": 3.8541,
+      "step": 2265088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2905203414915062e-05,
+      "loss": 3.8525,
+      "step": 2265600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2896817467404542e-05,
+      "loss": 3.8521,
+      "step": 2266112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2888431519894024e-05,
+      "loss": 3.8539,
+      "step": 2266624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2880061951187233e-05,
+      "loss": 3.8523,
+      "step": 2267136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2871676003676714e-05,
+      "loss": 3.8539,
+      "step": 2267648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2863290056166194e-05,
+      "loss": 3.859,
+      "step": 2268160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2854904108655674e-05,
+      "loss": 3.8522,
+      "step": 2268672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2846518161145154e-05,
+      "loss": 3.8551,
+      "step": 2269184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2838148592438365e-05,
+      "loss": 3.8556,
+      "step": 2269696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2829762644927845e-05,
+      "loss": 3.8437,
+      "step": 2270208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2821376697417325e-05,
+      "loss": 3.8633,
+      "step": 2270720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2812990749906807e-05,
+      "loss": 3.8584,
+      "step": 2271232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2804604802396287e-05,
+      "loss": 3.8609,
+      "step": 2271744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2796218854885763e-05,
+      "loss": 3.8664,
+      "step": 2272256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2787849286178977e-05,
+      "loss": 3.8644,
+      "step": 2272768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2779463338668457e-05,
+      "loss": 3.8557,
+      "step": 2273280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2771077391157937e-05,
+      "loss": 3.8517,
+      "step": 2273792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2762691443647415e-05,
+      "loss": 3.8614,
+      "step": 2274304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2754305496136895e-05,
+      "loss": 3.8573,
+      "step": 2274816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2745919548626375e-05,
+      "loss": 3.8526,
+      "step": 2275328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2737549979919588e-05,
+      "loss": 3.8625,
+      "step": 2275840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2729164032409066e-05,
+      "loss": 3.8547,
+      "step": 2276352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2720778084898546e-05,
+      "loss": 3.8586,
+      "step": 2276864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2712392137388026e-05,
+      "loss": 3.8515,
+      "step": 2277376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2704006189877508e-05,
+      "loss": 3.8512,
+      "step": 2277888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2695620242366988e-05,
+      "loss": 3.8608,
+      "step": 2278400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2687250673660198e-05,
+      "loss": 3.8508,
+      "step": 2278912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2678864726149678e-05,
+      "loss": 3.8644,
+      "step": 2279424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2670478778639158e-05,
+      "loss": 3.8668,
+      "step": 2279936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2662092831128638e-05,
+      "loss": 3.8594,
+      "step": 2280448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2653706883618118e-05,
+      "loss": 3.8518,
+      "step": 2280960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2645320936107596e-05,
+      "loss": 3.864,
+      "step": 2281472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2636934988597076e-05,
+      "loss": 3.861,
+      "step": 2281984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2628565419890289e-05,
+      "loss": 3.8618,
+      "step": 2282496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.262017947237977e-05,
+      "loss": 3.8555,
+      "step": 2283008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2611793524869247e-05,
+      "loss": 3.8553,
+      "step": 2283520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2603407577358727e-05,
+      "loss": 3.8672,
+      "step": 2284032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2595021629848209e-05,
+      "loss": 3.8752,
+      "step": 2284544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2586635682337689e-05,
+      "loss": 3.8523,
+      "step": 2285056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25782661136309e-05,
+      "loss": 3.8657,
+      "step": 2285568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.256988016612038e-05,
+      "loss": 3.8672,
+      "step": 2286080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.256149421860986e-05,
+      "loss": 3.8575,
+      "step": 2286592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.255312464990307e-05,
+      "loss": 3.8591,
+      "step": 2287104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.254473870239255e-05,
+      "loss": 3.8562,
+      "step": 2287616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.253635275488203e-05,
+      "loss": 3.8577,
+      "step": 2288128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.252796680737151e-05,
+      "loss": 3.8675,
+      "step": 2288640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2519580859860992e-05,
+      "loss": 3.8468,
+      "step": 2289152
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.982823610305786,
+      "eval_runtime": 293.4983,
+      "eval_samples_per_second": 1300.147,
+      "eval_steps_per_second": 40.631,
+      "step": 2289600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2511194912350472e-05,
+      "loss": 3.8588,
+      "step": 2289664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2502808964839952e-05,
+      "loss": 3.8458,
+      "step": 2290176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2494439396133162e-05,
+      "loss": 3.8553,
+      "step": 2290688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2486053448622642e-05,
+      "loss": 3.8714,
+      "step": 2291200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.247766750111212e-05,
+      "loss": 3.8673,
+      "step": 2291712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.24692815536016e-05,
+      "loss": 3.8556,
+      "step": 2292224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2460895606091082e-05,
+      "loss": 3.8637,
+      "step": 2292736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.245250965858056e-05,
+      "loss": 3.8534,
+      "step": 2293248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.244412371107004e-05,
+      "loss": 3.8501,
+      "step": 2293760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.243573776355952e-05,
+      "loss": 3.8662,
+      "step": 2294272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2427351816049e-05,
+      "loss": 3.8614,
+      "step": 2294784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.241896586853848e-05,
+      "loss": 3.8659,
+      "step": 2295296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2410612678635424e-05,
+      "loss": 3.869,
+      "step": 2295808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2402226731124904e-05,
+      "loss": 3.8491,
+      "step": 2296320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2393840783614384e-05,
+      "loss": 3.8474,
+      "step": 2296832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2385454836103864e-05,
+      "loss": 3.8441,
+      "step": 2297344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2377068888593343e-05,
+      "loss": 3.86,
+      "step": 2297856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2368682941082822e-05,
+      "loss": 3.8559,
+      "step": 2298368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2360296993572303e-05,
+      "loss": 3.8545,
+      "step": 2298880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2351911046061783e-05,
+      "loss": 3.8585,
+      "step": 2299392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2343525098551263e-05,
+      "loss": 3.8745,
+      "step": 2299904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2335139151040742e-05,
+      "loss": 3.8629,
+      "step": 2300416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2326769582333954e-05,
+      "loss": 3.8593,
+      "step": 2300928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2318383634823434e-05,
+      "loss": 3.8616,
+      "step": 2301440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2309997687312912e-05,
+      "loss": 3.8661,
+      "step": 2301952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2301611739802394e-05,
+      "loss": 3.8596,
+      "step": 2302464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2293225792291874e-05,
+      "loss": 3.8524,
+      "step": 2302976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2284839844781354e-05,
+      "loss": 3.8548,
+      "step": 2303488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2276453897270832e-05,
+      "loss": 3.8499,
+      "step": 2304000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2268084328564045e-05,
+      "loss": 3.8549,
+      "step": 2304512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2259698381053525e-05,
+      "loss": 3.8564,
+      "step": 2305024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2251312433543004e-05,
+      "loss": 3.8581,
+      "step": 2305536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2242926486032484e-05,
+      "loss": 3.8618,
+      "step": 2306048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2234540538521964e-05,
+      "loss": 3.8617,
+      "step": 2306560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2226154591011444e-05,
+      "loss": 3.8602,
+      "step": 2307072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2217785022304655e-05,
+      "loss": 3.8666,
+      "step": 2307584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2209399074794135e-05,
+      "loss": 3.8647,
+      "step": 2308096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2201013127283615e-05,
+      "loss": 3.8406,
+      "step": 2308608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2192627179773095e-05,
+      "loss": 3.8651,
+      "step": 2309120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2184241232262575e-05,
+      "loss": 3.8433,
+      "step": 2309632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2175855284752055e-05,
+      "loss": 3.8542,
+      "step": 2310144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2167469337241535e-05,
+      "loss": 3.8578,
+      "step": 2310656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2159083389731015e-05,
+      "loss": 3.8689,
+      "step": 2311168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2150697442220495e-05,
+      "loss": 3.8536,
+      "step": 2311680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2142327873513706e-05,
+      "loss": 3.8592,
+      "step": 2312192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2133941926003186e-05,
+      "loss": 3.8653,
+      "step": 2312704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2125555978492665e-05,
+      "loss": 3.8599,
+      "step": 2313216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2117170030982145e-05,
+      "loss": 3.8616,
+      "step": 2313728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2108800462275358e-05,
+      "loss": 3.8318,
+      "step": 2314240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2100414514764836e-05,
+      "loss": 3.8546,
+      "step": 2314752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2092028567254316e-05,
+      "loss": 3.8639,
+      "step": 2315264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2083658998547529e-05,
+      "loss": 3.8616,
+      "step": 2315776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2075273051037009e-05,
+      "loss": 3.8579,
+      "step": 2316288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2066887103526489e-05,
+      "loss": 3.8485,
+      "step": 2316800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2058501156015968e-05,
+      "loss": 3.8504,
+      "step": 2317312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.205013158730918e-05,
+      "loss": 3.8366,
+      "step": 2317824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.204174563979866e-05,
+      "loss": 3.8638,
+      "step": 2318336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.203335969228814e-05,
+      "loss": 3.8379,
+      "step": 2318848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2024973744777619e-05,
+      "loss": 3.8612,
+      "step": 2319360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2016587797267099e-05,
+      "loss": 3.8572,
+      "step": 2319872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.200821822856031e-05,
+      "loss": 3.8454,
+      "step": 2320384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.199983228104979e-05,
+      "loss": 3.8586,
+      "step": 2320896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1991462712343002e-05,
+      "loss": 3.8434,
+      "step": 2321408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1983076764832482e-05,
+      "loss": 3.8486,
+      "step": 2321920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.197469081732196e-05,
+      "loss": 3.8472,
+      "step": 2322432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.196630486981144e-05,
+      "loss": 3.8605,
+      "step": 2322944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1957918922300922e-05,
+      "loss": 3.8425,
+      "step": 2323456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1949549353594131e-05,
+      "loss": 3.8556,
+      "step": 2323968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1941163406083613e-05,
+      "loss": 3.842,
+      "step": 2324480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1932777458573093e-05,
+      "loss": 3.8493,
+      "step": 2324992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1924391511062573e-05,
+      "loss": 3.8629,
+      "step": 2325504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1916005563552051e-05,
+      "loss": 3.8592,
+      "step": 2326016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1907619616041533e-05,
+      "loss": 3.851,
+      "step": 2326528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1899233668531013e-05,
+      "loss": 3.8579,
+      "step": 2327040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1890847721020491e-05,
+      "loss": 3.8696,
+      "step": 2327552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1882461773509971e-05,
+      "loss": 3.8453,
+      "step": 2328064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1874092204803183e-05,
+      "loss": 3.8611,
+      "step": 2328576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1865706257292663e-05,
+      "loss": 3.8501,
+      "step": 2329088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1857320309782143e-05,
+      "loss": 3.854,
+      "step": 2329600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1848934362271623e-05,
+      "loss": 3.8495,
+      "step": 2330112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1840548414761103e-05,
+      "loss": 3.854,
+      "step": 2330624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1832162467250581e-05,
+      "loss": 3.8664,
+      "step": 2331136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1823792898543794e-05,
+      "loss": 3.8658,
+      "step": 2331648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1815406951033274e-05,
+      "loss": 3.8513,
+      "step": 2332160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1807021003522754e-05,
+      "loss": 3.835,
+      "step": 2332672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1798635056012234e-05,
+      "loss": 3.8459,
+      "step": 2333184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1790249108501714e-05,
+      "loss": 3.8581,
+      "step": 2333696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1781863160991194e-05,
+      "loss": 3.8488,
+      "step": 2334208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1773477213480674e-05,
+      "loss": 3.8499,
+      "step": 2334720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1765091265970154e-05,
+      "loss": 3.8583,
+      "step": 2335232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1756721697263364e-05,
+      "loss": 3.8545,
+      "step": 2335744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1748335749752844e-05,
+      "loss": 3.8552,
+      "step": 2336256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1739949802242324e-05,
+      "loss": 3.8478,
+      "step": 2336768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1731563854731804e-05,
+      "loss": 3.8611,
+      "step": 2337280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1723177907221284e-05,
+      "loss": 3.8484,
+      "step": 2337792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1714808338514495e-05,
+      "loss": 3.8679,
+      "step": 2338304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1706422391003975e-05,
+      "loss": 3.8449,
+      "step": 2338816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1698036443493455e-05,
+      "loss": 3.8444,
+      "step": 2339328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1689650495982935e-05,
+      "loss": 3.8586,
+      "step": 2339840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1681280927276146e-05,
+      "loss": 3.8546,
+      "step": 2340352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1672894979765626e-05,
+      "loss": 3.8621,
+      "step": 2340864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1664525411058838e-05,
+      "loss": 3.8515,
+      "step": 2341376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1656139463548316e-05,
+      "loss": 3.8509,
+      "step": 2341888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1647753516037798e-05,
+      "loss": 3.8452,
+      "step": 2342400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1639367568527278e-05,
+      "loss": 3.8514,
+      "step": 2342912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1630981621016758e-05,
+      "loss": 3.8432,
+      "step": 2343424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1622595673506236e-05,
+      "loss": 3.8506,
+      "step": 2343936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1614209725995718e-05,
+      "loss": 3.8543,
+      "step": 2344448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1605823778485198e-05,
+      "loss": 3.8478,
+      "step": 2344960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1597454209778409e-05,
+      "loss": 3.8474,
+      "step": 2345472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1589068262267889e-05,
+      "loss": 3.8559,
+      "step": 2345984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1580682314757369e-05,
+      "loss": 3.8397,
+      "step": 2346496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1572296367246848e-05,
+      "loss": 3.8603,
+      "step": 2347008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1563910419736328e-05,
+      "loss": 3.8535,
+      "step": 2347520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.155554085102954e-05,
+      "loss": 3.8601,
+      "step": 2348032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.154715490351902e-05,
+      "loss": 3.8603,
+      "step": 2348544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1538785334812232e-05,
+      "loss": 3.8624,
+      "step": 2349056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.153039938730171e-05,
+      "loss": 3.8521,
+      "step": 2349568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.152201343979119e-05,
+      "loss": 3.8491,
+      "step": 2350080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1513627492280672e-05,
+      "loss": 3.8541,
+      "step": 2350592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.150524154477015e-05,
+      "loss": 3.8541,
+      "step": 2351104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.149687197606336e-05,
+      "loss": 3.8449,
+      "step": 2351616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1488486028552842e-05,
+      "loss": 3.8644,
+      "step": 2352128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1480100081042322e-05,
+      "loss": 3.8485,
+      "step": 2352640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.14717141335318e-05,
+      "loss": 3.857,
+      "step": 2353152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1463328186021282e-05,
+      "loss": 3.8473,
+      "step": 2353664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1454942238510762e-05,
+      "loss": 3.8465,
+      "step": 2354176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1446572669803971e-05,
+      "loss": 3.8594,
+      "step": 2354688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1438186722293453e-05,
+      "loss": 3.8507,
+      "step": 2355200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1429800774782933e-05,
+      "loss": 3.8531,
+      "step": 2355712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1421414827272413e-05,
+      "loss": 3.8659,
+      "step": 2356224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1413028879761891e-05,
+      "loss": 3.8562,
+      "step": 2356736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1404642932251373e-05,
+      "loss": 3.8514,
+      "step": 2357248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1396273363544583e-05,
+      "loss": 3.8601,
+      "step": 2357760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1387887416034063e-05,
+      "loss": 3.8601,
+      "step": 2358272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1379501468523543e-05,
+      "loss": 3.8585,
+      "step": 2358784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1371115521013023e-05,
+      "loss": 3.8499,
+      "step": 2359296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1362729573502503e-05,
+      "loss": 3.8535,
+      "step": 2359808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1354343625991983e-05,
+      "loss": 3.8696,
+      "step": 2360320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1345957678481463e-05,
+      "loss": 3.8687,
+      "step": 2360832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1337571730970943e-05,
+      "loss": 3.854,
+      "step": 2361344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1329185783460421e-05,
+      "loss": 3.8556,
+      "step": 2361856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1320832593557365e-05,
+      "loss": 3.8651,
+      "step": 2362368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1312446646046845e-05,
+      "loss": 3.8581,
+      "step": 2362880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1304060698536326e-05,
+      "loss": 3.8566,
+      "step": 2363392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1295674751025805e-05,
+      "loss": 3.8546,
+      "step": 2363904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1287288803515285e-05,
+      "loss": 3.8501,
+      "step": 2364416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1278902856004764e-05,
+      "loss": 3.8645,
+      "step": 2364928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1270549666101708e-05,
+      "loss": 3.845,
+      "step": 2365440
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9818062782287598,
+      "eval_runtime": 291.518,
+      "eval_samples_per_second": 1308.979,
+      "eval_steps_per_second": 40.907,
+      "step": 2365920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1262163718591188e-05,
+      "loss": 3.8227,
+      "step": 2365952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1253777771080668e-05,
+      "loss": 3.848,
+      "step": 2366464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1245391823570148e-05,
+      "loss": 3.8484,
+      "step": 2366976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1237005876059626e-05,
+      "loss": 3.8689,
+      "step": 2367488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1228619928549108e-05,
+      "loss": 3.8645,
+      "step": 2368000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1220233981038588e-05,
+      "loss": 3.8513,
+      "step": 2368512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1211848033528067e-05,
+      "loss": 3.8605,
+      "step": 2369024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1203462086017546e-05,
+      "loss": 3.8523,
+      "step": 2369536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1195076138507027e-05,
+      "loss": 3.8501,
+      "step": 2370048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1186690190996507e-05,
+      "loss": 3.8619,
+      "step": 2370560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1178304243485986e-05,
+      "loss": 3.8542,
+      "step": 2371072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1169918295975467e-05,
+      "loss": 3.8605,
+      "step": 2371584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1161548727268678e-05,
+      "loss": 3.8671,
+      "step": 2372096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1153162779758156e-05,
+      "loss": 3.8479,
+      "step": 2372608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1144776832247638e-05,
+      "loss": 3.8417,
+      "step": 2373120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1136390884737118e-05,
+      "loss": 3.841,
+      "step": 2373632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1128004937226598e-05,
+      "loss": 3.8532,
+      "step": 2374144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1119618989716076e-05,
+      "loss": 3.8554,
+      "step": 2374656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1111233042205558e-05,
+      "loss": 3.8496,
+      "step": 2375168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1102847094695038e-05,
+      "loss": 3.8524,
+      "step": 2375680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1094461147184516e-05,
+      "loss": 3.8691,
+      "step": 2376192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1086075199673996e-05,
+      "loss": 3.8557,
+      "step": 2376704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1077689252163478e-05,
+      "loss": 3.8616,
+      "step": 2377216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1069303304652956e-05,
+      "loss": 3.854,
+      "step": 2377728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1060917357142436e-05,
+      "loss": 3.8652,
+      "step": 2378240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1052531409631916e-05,
+      "loss": 3.8531,
+      "step": 2378752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1044145462121397e-05,
+      "loss": 3.8528,
+      "step": 2379264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1035792272218339e-05,
+      "loss": 3.8502,
+      "step": 2379776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1027406324707819e-05,
+      "loss": 3.8493,
+      "step": 2380288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1019020377197299e-05,
+      "loss": 3.8507,
+      "step": 2380800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1010634429686779e-05,
+      "loss": 3.8514,
+      "step": 2381312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.100226486097999e-05,
+      "loss": 3.8555,
+      "step": 2381824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.099387891346947e-05,
+      "loss": 3.8558,
+      "step": 2382336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.098549296595895e-05,
+      "loss": 3.858,
+      "step": 2382848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.097710701844843e-05,
+      "loss": 3.8615,
+      "step": 2383360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.096873744974164e-05,
+      "loss": 3.8598,
+      "step": 2383872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0960351502231122e-05,
+      "loss": 3.8614,
+      "step": 2384384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0951965554720602e-05,
+      "loss": 3.8383,
+      "step": 2384896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.094357960721008e-05,
+      "loss": 3.8628,
+      "step": 2385408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.093519365969956e-05,
+      "loss": 3.8343,
+      "step": 2385920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0926807712189042e-05,
+      "loss": 3.8532,
+      "step": 2386432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.091842176467852e-05,
+      "loss": 3.853,
+      "step": 2386944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0910035817168e-05,
+      "loss": 3.8643,
+      "step": 2387456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.090164986965748e-05,
+      "loss": 3.8489,
+      "step": 2387968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0893280300950692e-05,
+      "loss": 3.8604,
+      "step": 2388480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.088489435344017e-05,
+      "loss": 3.8622,
+      "step": 2388992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.087650840592965e-05,
+      "loss": 3.8585,
+      "step": 2389504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0868138837222863e-05,
+      "loss": 3.8564,
+      "step": 2390016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0859752889712343e-05,
+      "loss": 3.8338,
+      "step": 2390528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0851366942201823e-05,
+      "loss": 3.8514,
+      "step": 2391040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0842980994691303e-05,
+      "loss": 3.8605,
+      "step": 2391552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0834595047180783e-05,
+      "loss": 3.8567,
+      "step": 2392064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0826209099670261e-05,
+      "loss": 3.8525,
+      "step": 2392576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0817823152159743e-05,
+      "loss": 3.8446,
+      "step": 2393088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0809437204649223e-05,
+      "loss": 3.8476,
+      "step": 2393600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0801051257138701e-05,
+      "loss": 3.8334,
+      "step": 2394112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0792665309628181e-05,
+      "loss": 3.8637,
+      "step": 2394624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0784279362117663e-05,
+      "loss": 3.834,
+      "step": 2395136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0775893414607143e-05,
+      "loss": 3.858,
+      "step": 2395648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0767507467096621e-05,
+      "loss": 3.8544,
+      "step": 2396160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0759137898389833e-05,
+      "loss": 3.8453,
+      "step": 2396672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0750751950879313e-05,
+      "loss": 3.8498,
+      "step": 2397184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0742366003368792e-05,
+      "loss": 3.8403,
+      "step": 2397696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0733980055858273e-05,
+      "loss": 3.847,
+      "step": 2398208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0725610487151484e-05,
+      "loss": 3.8387,
+      "step": 2398720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0717240918444697e-05,
+      "loss": 3.8604,
+      "step": 2399232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0708854970934175e-05,
+      "loss": 3.8445,
+      "step": 2399744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0700469023423655e-05,
+      "loss": 3.8448,
+      "step": 2400256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0692083075913135e-05,
+      "loss": 3.8394,
+      "step": 2400768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0683697128402616e-05,
+      "loss": 3.849,
+      "step": 2401280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0675327559695826e-05,
+      "loss": 3.862,
+      "step": 2401792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0666941612185307e-05,
+      "loss": 3.8527,
+      "step": 2402304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0658555664674787e-05,
+      "loss": 3.8427,
+      "step": 2402816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0650169717164265e-05,
+      "loss": 3.8578,
+      "step": 2403328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0641783769653745e-05,
+      "loss": 3.8641,
+      "step": 2403840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0633397822143227e-05,
+      "loss": 3.8394,
+      "step": 2404352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0625011874632705e-05,
+      "loss": 3.859,
+      "step": 2404864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0616625927122185e-05,
+      "loss": 3.8474,
+      "step": 2405376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0608256358415398e-05,
+      "loss": 3.8493,
+      "step": 2405888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0599870410904878e-05,
+      "loss": 3.845,
+      "step": 2406400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0591484463394356e-05,
+      "loss": 3.8522,
+      "step": 2406912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0583114894687568e-05,
+      "loss": 3.8622,
+      "step": 2407424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0574728947177048e-05,
+      "loss": 3.8606,
+      "step": 2407936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0566342999666528e-05,
+      "loss": 3.8475,
+      "step": 2408448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0557957052156008e-05,
+      "loss": 3.8291,
+      "step": 2408960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0549571104645488e-05,
+      "loss": 3.8469,
+      "step": 2409472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0541185157134968e-05,
+      "loss": 3.8508,
+      "step": 2409984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0532799209624446e-05,
+      "loss": 3.8513,
+      "step": 2410496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0524413262113928e-05,
+      "loss": 3.8422,
+      "step": 2411008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0516043693407139e-05,
+      "loss": 3.8558,
+      "step": 2411520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0507657745896619e-05,
+      "loss": 3.8513,
+      "step": 2412032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0499271798386099e-05,
+      "loss": 3.8539,
+      "step": 2412544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0490885850875579e-05,
+      "loss": 3.8432,
+      "step": 2413056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0482499903365059e-05,
+      "loss": 3.8561,
+      "step": 2413568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.047413033465827e-05,
+      "loss": 3.8445,
+      "step": 2414080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.046574438714775e-05,
+      "loss": 3.8641,
+      "step": 2414592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.045735843963723e-05,
+      "loss": 3.8477,
+      "step": 2415104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.044897249212671e-05,
+      "loss": 3.8377,
+      "step": 2415616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.044060292341992e-05,
+      "loss": 3.852,
+      "step": 2416128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.04322169759094e-05,
+      "loss": 3.8524,
+      "step": 2416640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0423831028398882e-05,
+      "loss": 3.8577,
+      "step": 2417152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.041544508088836e-05,
+      "loss": 3.85,
+      "step": 2417664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.040705913337784e-05,
+      "loss": 3.846,
+      "step": 2418176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.039867318586732e-05,
+      "loss": 3.8421,
+      "step": 2418688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0390287238356802e-05,
+      "loss": 3.8506,
+      "step": 2419200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.038190129084628e-05,
+      "loss": 3.8368,
+      "step": 2419712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0373531722139492e-05,
+      "loss": 3.8501,
+      "step": 2420224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0365145774628972e-05,
+      "loss": 3.8498,
+      "step": 2420736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.035675982711845e-05,
+      "loss": 3.8473,
+      "step": 2421248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.034837387960793e-05,
+      "loss": 3.8406,
+      "step": 2421760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0340004310901143e-05,
+      "loss": 3.8519,
+      "step": 2422272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0331618363390623e-05,
+      "loss": 3.8398,
+      "step": 2422784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0323248794683834e-05,
+      "loss": 3.8527,
+      "step": 2423296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0314862847173314e-05,
+      "loss": 3.8488,
+      "step": 2423808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0306476899662794e-05,
+      "loss": 3.8506,
+      "step": 2424320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0298090952152274e-05,
+      "loss": 3.8582,
+      "step": 2424832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0289705004641754e-05,
+      "loss": 3.8571,
+      "step": 2425344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0281319057131233e-05,
+      "loss": 3.853,
+      "step": 2425856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0272933109620713e-05,
+      "loss": 3.8445,
+      "step": 2426368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0264547162110193e-05,
+      "loss": 3.8527,
+      "step": 2426880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0256161214599673e-05,
+      "loss": 3.8519,
+      "step": 2427392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0247808024696617e-05,
+      "loss": 3.8427,
+      "step": 2427904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0239422077186097e-05,
+      "loss": 3.8562,
+      "step": 2428416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0231036129675575e-05,
+      "loss": 3.8487,
+      "step": 2428928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0222650182165055e-05,
+      "loss": 3.8548,
+      "step": 2429440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0214264234654536e-05,
+      "loss": 3.8413,
+      "step": 2429952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0205878287144015e-05,
+      "loss": 3.8452,
+      "step": 2430464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0197508718437227e-05,
+      "loss": 3.8529,
+      "step": 2430976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0189122770926707e-05,
+      "loss": 3.8463,
+      "step": 2431488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0180736823416187e-05,
+      "loss": 3.8462,
+      "step": 2432000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0172350875905665e-05,
+      "loss": 3.8656,
+      "step": 2432512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0163981307198878e-05,
+      "loss": 3.8549,
+      "step": 2433024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0155595359688358e-05,
+      "loss": 3.8473,
+      "step": 2433536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0147209412177838e-05,
+      "loss": 3.8545,
+      "step": 2434048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0138823464667318e-05,
+      "loss": 3.8532,
+      "step": 2434560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0130437517156798e-05,
+      "loss": 3.8551,
+      "step": 2435072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0122051569646278e-05,
+      "loss": 3.8505,
+      "step": 2435584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0113682000939488e-05,
+      "loss": 3.8472,
+      "step": 2436096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0105296053428968e-05,
+      "loss": 3.8628,
+      "step": 2436608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.009692648472218e-05,
+      "loss": 3.8641,
+      "step": 2437120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0088540537211661e-05,
+      "loss": 3.8527,
+      "step": 2437632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0080154589701139e-05,
+      "loss": 3.852,
+      "step": 2438144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0071768642190619e-05,
+      "loss": 3.8625,
+      "step": 2438656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.00633826946801e-05,
+      "loss": 3.8483,
+      "step": 2439168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0054996747169579e-05,
+      "loss": 3.8531,
+      "step": 2439680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0046610799659059e-05,
+      "loss": 3.8513,
+      "step": 2440192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0038224852148539e-05,
+      "loss": 3.8473,
+      "step": 2440704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.002983890463802e-05,
+      "loss": 3.8611,
+      "step": 2441216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0021452957127499e-05,
+      "loss": 3.8433,
+      "step": 2441728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0013067009616979e-05,
+      "loss": 3.8488,
+      "step": 2442240
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9805712699890137,
+      "eval_runtime": 293.3539,
+      "eval_samples_per_second": 1300.787,
+      "eval_steps_per_second": 40.651,
+      "step": 2442240
+    }
+  ],
+  "logging_steps": 512,
+  "max_steps": 3052726,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 10,
+  "total_flos": 1.0091383890244074e+18,
+  "trial_name": null,
+  "trial_params": null
+}