{
  "best_metric": 4.1645636558532715,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/passive/lstm/3/checkpoints/checkpoint-457920",
  "epoch": 1.0250006060157382,
  "eval_steps": 10,
  "global_step": 457920,
  "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
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.8921344794207635e+17,
  "trial_name": null,
  "trial_params": null
}