{
  "best_metric": 4.470288276672363,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-only/lstm/3/checkpoints/checkpoint-152640",
  "epoch": 1.0250006060157382,
  "eval_steps": 10,
  "global_step": 152640,
  "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": 10.8208,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5525,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0564,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9917,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9456,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.8947,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7148,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.608,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5116,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4526,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.4085,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 6.3554,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 6.2919,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989098268236324e-05,
      "loss": 6.2369,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988259673485272e-05,
      "loss": 6.1912,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.98742107873422e-05,
      "loss": 6.1306,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986582483983168e-05,
      "loss": 6.0968,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985743889232116e-05,
      "loss": 6.0456,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9932,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9602,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.9203,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9823927859886547e-05,
      "loss": 5.8789,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 5.8514,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 5.8072,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.786,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7572,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.7303,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6995,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.97652262273129e-05,
      "loss": 5.6648,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.975684027980238e-05,
      "loss": 5.652,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974845433229186e-05,
      "loss": 5.6281,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.6148,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.5876,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.972331286856403e-05,
      "loss": 5.5684,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714926921053515e-05,
      "loss": 5.5576,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706540973542995e-05,
      "loss": 5.5397,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.5201,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.4885,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9681399509815164e-05,
      "loss": 5.4832,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9673013562304644e-05,
      "loss": 5.4514,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4533,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.427,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4288,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.3997,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963110020355577e-05,
      "loss": 5.3924,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962273063484898e-05,
      "loss": 5.3859,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.3776,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 5.3604,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3383,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.3357,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3381,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95724313285896e-05,
      "loss": 5.3259,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.3046,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955565943356856e-05,
      "loss": 5.2941,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954727348605804e-05,
      "loss": 5.2901,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2767,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2868,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2426,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2558,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9505360127309166e-05,
      "loss": 5.2416,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9496974179798646e-05,
      "loss": 5.2263,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948860461109186e-05,
      "loss": 5.234,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948021866358134e-05,
      "loss": 5.212,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947183271607082e-05,
      "loss": 5.2018,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94634467685603e-05,
      "loss": 5.1982,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.209,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.1833,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94383216836362e-05,
      "loss": 5.1782,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942993573612568e-05,
      "loss": 5.1637,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942154978861516e-05,
      "loss": 5.1557,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941316384110464e-05,
      "loss": 5.161,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940477789359412e-05,
      "loss": 5.165,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93963919460836e-05,
      "loss": 5.1376,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.938800599857309e-05,
      "loss": 5.148,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937962005106256e-05,
      "loss": 5.1488,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937123410355204e-05,
      "loss": 5.1286,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362864534845256e-05,
      "loss": 5.1109,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935447858733473e-05,
      "loss": 5.1119,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.934609263982421e-05,
      "loss": 5.1045,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933770669231369e-05,
      "loss": 5.0941,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932932074480317e-05,
      "loss": 5.0918,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.0943,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931256522858586e-05,
      "loss": 5.0933,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930417928107534e-05,
      "loss": 5.0756,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295809712368554e-05,
      "loss": 5.0548,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0617,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0651,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0534,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0504,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9253879974815954e-05,
      "loss": 5.0318,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9245494027305433e-05,
      "loss": 5.0316,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9237108079794913e-05,
      "loss": 5.0262,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922873851108812e-05,
      "loss": 5.0245,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92203525635776e-05,
      "loss": 5.0253,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921196661606708e-05,
      "loss": 5.0075,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920358066855656e-05,
      "loss": 5.0243,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919521109984978e-05,
      "loss": 5.0082,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918682515233926e-05,
      "loss": 4.989,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917843920482874e-05,
      "loss": 4.9839,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917005325731822e-05,
      "loss": 4.9849,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91616673098077e-05,
      "loss": 4.983,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915328136229718e-05,
      "loss": 4.9812,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914489541478666e-05,
      "loss": 4.9736,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913650946727614e-05,
      "loss": 4.9657,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912813989856935e-05,
      "loss": 4.9618,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911975395105883e-05,
      "loss": 4.9411,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911136800354831e-05,
      "loss": 4.945,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9102998434841516e-05,
      "loss": 4.9434,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9094612487330996e-05,
      "loss": 4.9392,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9086226539820476e-05,
      "loss": 4.9336,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907785697111369e-05,
      "loss": 4.9397,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906947102360317e-05,
      "loss": 4.9143,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906108507609265e-05,
      "loss": 4.9181,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905269912858213e-05,
      "loss": 4.9116,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904431318107161e-05,
      "loss": 4.9176,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903592723356109e-05,
      "loss": 4.922,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027541286050565e-05,
      "loss": 4.9052,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019155338540045e-05,
      "loss": 4.9018,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010769391029525e-05,
      "loss": 4.8923,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900239982232274e-05,
      "loss": 4.9004,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994013874812214e-05,
      "loss": 4.894,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89856279273017e-05,
      "loss": 4.8885,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897724197979118e-05,
      "loss": 4.8824,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896885603228066e-05,
      "loss": 4.8746,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960502842377605e-05,
      "loss": 4.8702,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952133273670815e-05,
      "loss": 4.859,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943747326160294e-05,
      "loss": 4.8638,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935361378649774e-05,
      "loss": 4.8584,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926975431139254e-05,
      "loss": 4.8544,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918589483628734e-05,
      "loss": 4.8448,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910219914921943e-05,
      "loss": 4.8607,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8901833967411423e-05,
      "loss": 4.8604,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88934480199009e-05,
      "loss": 4.8495,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.8376,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887667612487986e-05,
      "loss": 4.8368,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886829017736934e-05,
      "loss": 4.8398,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885990422985882e-05,
      "loss": 4.8364,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515182823483e-05,
      "loss": 4.8313,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884313233483778e-05,
      "loss": 4.8279,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883474638732726e-05,
      "loss": 4.8247,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882636043981674e-05,
      "loss": 4.8173,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881797449230622e-05,
      "loss": 4.8127,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88095885447957e-05,
      "loss": 4.8167,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880121897608891e-05,
      "loss": 4.8047,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879284940738212e-05,
      "loss": 4.7993,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87844634598716e-05,
      "loss": 4.8073,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877607751236109e-05,
      "loss": 4.7931,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876769156485057e-05,
      "loss": 4.8016,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875930561734005e-05,
      "loss": 4.7919,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875091966982953e-05,
      "loss": 4.7826,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.750964641571045,
      "eval_runtime": 293.5866,
      "eval_samples_per_second": 1299.756,
      "eval_steps_per_second": 40.618,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874253372231901e-05,
      "loss": 4.7763,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873414777480849e-05,
      "loss": 4.7778,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872576182729797e-05,
      "loss": 4.7878,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871737587978745e-05,
      "loss": 4.7761,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870898993227693e-05,
      "loss": 4.7784,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87006039847664e-05,
      "loss": 4.7652,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692234416059616e-05,
      "loss": 4.7664,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683848468549096e-05,
      "loss": 4.7531,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675478899842306e-05,
      "loss": 4.7645,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866710933113552e-05,
      "loss": 4.7619,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658723383625e-05,
      "loss": 4.7686,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865033743611448e-05,
      "loss": 4.7477,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864195148860396e-05,
      "loss": 4.7477,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863356554109344e-05,
      "loss": 4.7431,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862517959358292e-05,
      "loss": 4.7345,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86167936460724e-05,
      "loss": 4.7372,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608407698561874e-05,
      "loss": 4.7457,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860003812985509e-05,
      "loss": 4.7284,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859165218234457e-05,
      "loss": 4.7292,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858326623483405e-05,
      "loss": 4.7434,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.857488028732352e-05,
      "loss": 4.7308,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.856649433981301e-05,
      "loss": 4.7303,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.855810839230249e-05,
      "loss": 4.7171,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85497388235957e-05,
      "loss": 4.7321,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854135287608518e-05,
      "loss": 4.7177,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853296692857466e-05,
      "loss": 4.7157,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852458098106414e-05,
      "loss": 4.7097,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851619503355362e-05,
      "loss": 4.705,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850782546484683e-05,
      "loss": 4.699,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849943951733631e-05,
      "loss": 4.6999,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849105356982579e-05,
      "loss": 4.7092,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848266762231527e-05,
      "loss": 4.7047,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847428167480475e-05,
      "loss": 4.697,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846589572729423e-05,
      "loss": 4.7064,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8457526158587444e-05,
      "loss": 4.6982,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844915658988065e-05,
      "loss": 4.6946,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844077064237013e-05,
      "loss": 4.6795,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843238469485961e-05,
      "loss": 4.6843,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842399874734909e-05,
      "loss": 4.6725,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841561279983857e-05,
      "loss": 4.6825,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840722685232805e-05,
      "loss": 4.6668,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839884090481753e-05,
      "loss": 4.6819,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839045495730701e-05,
      "loss": 4.6633,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838206900979649e-05,
      "loss": 4.6715,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837368306228597e-05,
      "loss": 4.6691,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836529711477545e-05,
      "loss": 4.6671,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835691116726493e-05,
      "loss": 4.6621,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834854159855814e-05,
      "loss": 4.6513,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834017202985136e-05,
      "loss": 4.649,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833178608234084e-05,
      "loss": 4.6657,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8323416513634046e-05,
      "loss": 4.6578,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8315030566123526e-05,
      "loss": 4.6506,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8306644618613006e-05,
      "loss": 4.6399,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8298258671102486e-05,
      "loss": 4.6497,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289872723591966e-05,
      "loss": 4.6381,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281486776081446e-05,
      "loss": 4.6633,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273100828570926e-05,
      "loss": 4.6257,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264714881060406e-05,
      "loss": 4.6468,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256328933549886e-05,
      "loss": 4.6374,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8247942986039366e-05,
      "loss": 4.6262,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8239557038528846e-05,
      "loss": 4.6387,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823118746982206e-05,
      "loss": 4.6323,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222801522311535e-05,
      "loss": 4.6164,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214415574801015e-05,
      "loss": 4.624,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8206029627290495e-05,
      "loss": 4.6368,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197643679779975e-05,
      "loss": 4.6212,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189274111073184e-05,
      "loss": 4.6185,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180888163562664e-05,
      "loss": 4.6144,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172502216052144e-05,
      "loss": 4.6104,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164116268541624e-05,
      "loss": 4.6206,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155730321031104e-05,
      "loss": 4.6318,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147344373520584e-05,
      "loss": 4.6074,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81389748048138e-05,
      "loss": 4.6249,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813058885730328e-05,
      "loss": 4.625,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812220290979276e-05,
      "loss": 4.6135,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811381696228224e-05,
      "loss": 4.6,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810543101477172e-05,
      "loss": 4.6117,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80970450672612e-05,
      "loss": 4.6056,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808865911975068e-05,
      "loss": 4.5988,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808027317224016e-05,
      "loss": 4.5995,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807190360353337e-05,
      "loss": 4.6023,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806353403482658e-05,
      "loss": 4.6113,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805514808731606e-05,
      "loss": 4.5987,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804676213980554e-05,
      "loss": 4.583,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803837619229502e-05,
      "loss": 4.5932,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8029990244784504e-05,
      "loss": 4.5961,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802162067607771e-05,
      "loss": 4.5923,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801323472856719e-05,
      "loss": 4.5971,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800484878105667e-05,
      "loss": 4.5837,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799646283354615e-05,
      "loss": 4.5795,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798807688603563e-05,
      "loss": 4.58,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797970731732884e-05,
      "loss": 4.5849,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797133774862205e-05,
      "loss": 4.5831,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796295180111153e-05,
      "loss": 4.5763,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795456585360101e-05,
      "loss": 4.5948,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794617990609049e-05,
      "loss": 4.5839,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793779395857997e-05,
      "loss": 4.5648,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792940801106946e-05,
      "loss": 4.5683,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792102206355894e-05,
      "loss": 4.5717,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791263611604842e-05,
      "loss": 4.5727,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790426654734163e-05,
      "loss": 4.572,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7895880599831107e-05,
      "loss": 4.57,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7887494652320587e-05,
      "loss": 4.564,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7879108704810067e-05,
      "loss": 4.5669,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870722757299546e-05,
      "loss": 4.5441,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786233680978902e-05,
      "loss": 4.559,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7853967241082236e-05,
      "loss": 4.5533,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845581293571715e-05,
      "loss": 4.5567,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837195346061195e-05,
      "loss": 4.5505,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7828809398550675e-05,
      "loss": 4.5605,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7820423451040155e-05,
      "loss": 4.5422,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.781205388233337e-05,
      "loss": 4.5448,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803667934822844e-05,
      "loss": 4.5464,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795281987312324e-05,
      "loss": 4.5464,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786896039801804e-05,
      "loss": 4.5611,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777852647109502e-05,
      "loss": 4.5486,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777015690238823e-05,
      "loss": 4.5466,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776177095487771e-05,
      "loss": 4.5397,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775338500736719e-05,
      "loss": 4.546,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774499905985666e-05,
      "loss": 4.544,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773661311234615e-05,
      "loss": 4.538,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772824354363936e-05,
      "loss": 4.5403,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7719857596128845e-05,
      "loss": 4.534,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771147164861832e-05,
      "loss": 4.5328,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77030857011078e-05,
      "loss": 4.5224,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769469975359728e-05,
      "loss": 4.5278,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768631380608676e-05,
      "loss": 4.524,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767792785857624e-05,
      "loss": 4.5293,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766954191106572e-05,
      "loss": 4.516,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76611559635552e-05,
      "loss": 4.5374,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765278639484841e-05,
      "loss": 4.5426,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764440044733789e-05,
      "loss": 4.5288,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763601449982737e-05,
      "loss": 4.5201,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762762855231685e-05,
      "loss": 4.5232,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761924260480633e-05,
      "loss": 4.5259,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7610856657295813e-05,
      "loss": 4.5263,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7602470709785293e-05,
      "loss": 4.5224,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75941011410785e-05,
      "loss": 4.5195,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758571519356798e-05,
      "loss": 4.5177,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757732924605746e-05,
      "loss": 4.5209,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756894329854694e-05,
      "loss": 4.5103,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756055735103642e-05,
      "loss": 4.5213,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75521714035259e-05,
      "loss": 4.5115,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754380183481911e-05,
      "loss": 4.5051,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753541588730859e-05,
      "loss": 4.5171,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752702993979807e-05,
      "loss": 4.5045,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751864399228755e-05,
      "loss": 4.512,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751025804477703e-05,
      "loss": 4.5008,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750187209726651e-05,
      "loss": 4.5004,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.470288276672363,
      "eval_runtime": 282.4462,
      "eval_samples_per_second": 1351.022,
      "eval_steps_per_second": 42.22,
      "step": 152640
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 6.318492144496669e+16,
  "trial_name": null,
  "trial_params": null
}