{
  "best_metric": 4.483644008636475,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl/lstm/0/checkpoints/checkpoint-152638",
  "epoch": 1.0250002784396635,
  "eval_steps": 10,
  "global_step": 152638,
  "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": 10.8211,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5721,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0588,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9811,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9532,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.8829,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7246,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6226,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5326,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4313,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.3752,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 6.3029,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 6.2341,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 6.1668,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1204,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.0849,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0468,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.9966,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9615,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9262,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8884,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.861,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 5.8291,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 5.7974,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7687,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7494,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 5.7183,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 5.6958,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6632,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756873037409844e-05,
      "loss": 5.6592,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748487089899324e-05,
      "loss": 5.6252,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 5.6108,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 5.6012,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 5.5843,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5541,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.541,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969818778363994e-05,
      "loss": 5.5346,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968980183612942e-05,
      "loss": 5.5124,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96814158886189e-05,
      "loss": 5.4989,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 5.4793,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4548,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.4442,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4302,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.4231,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963110020355577e-05,
      "loss": 5.412,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962273063484898e-05,
      "loss": 5.406,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 5.3823,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 5.3816,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3863,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.3585,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3484,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95724313285896e-05,
      "loss": 5.335,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.3438,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955565943356856e-05,
      "loss": 5.3057,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954727348605804e-05,
      "loss": 5.3243,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2949,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2919,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2844,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2744,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95053765061129e-05,
      "loss": 5.2638,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949699055860238e-05,
      "loss": 5.2507,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948860461109186e-05,
      "loss": 5.2374,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948021866358134e-05,
      "loss": 5.2435,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947183271607082e-05,
      "loss": 5.2297,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94634467685603e-05,
      "loss": 5.221,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.2142,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.2025,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.1857,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942993573612568e-05,
      "loss": 5.2035,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942154978861516e-05,
      "loss": 5.1805,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941316384110464e-05,
      "loss": 5.1735,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940477789359412e-05,
      "loss": 5.1825,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93963919460836e-05,
      "loss": 5.1582,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.938800599857309e-05,
      "loss": 5.1705,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937962005106256e-05,
      "loss": 5.1605,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937123410355204e-05,
      "loss": 5.1496,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.936284815604152e-05,
      "loss": 5.1484,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935447858733473e-05,
      "loss": 5.123,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346109018627945e-05,
      "loss": 5.1201,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.1183,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.1184,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.1071,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931256522858586e-05,
      "loss": 5.094,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930417928107534e-05,
      "loss": 5.0805,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295793333564825e-05,
      "loss": 5.0863,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0866,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0826,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0781,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0749,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9253879974815954e-05,
      "loss": 5.0595,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9245494027305433e-05,
      "loss": 5.0427,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9237108079794913e-05,
      "loss": 5.0392,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922873851108812e-05,
      "loss": 5.0461,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92203525635776e-05,
      "loss": 5.0394,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921196661606708e-05,
      "loss": 5.018,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920358066855656e-05,
      "loss": 5.0216,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919519472104604e-05,
      "loss": 5.0203,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918680877353552e-05,
      "loss": 5.0194,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917842282602501e-05,
      "loss": 5.011,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917003687851449e-05,
      "loss": 5.0109,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91616673098077e-05,
      "loss": 5.0031,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915328136229718e-05,
      "loss": 4.9935,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914489541478666e-05,
      "loss": 4.9931,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913650946727614e-05,
      "loss": 4.9924,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912813989856935e-05,
      "loss": 4.9781,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911975395105883e-05,
      "loss": 4.9811,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911136800354831e-05,
      "loss": 4.959,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910298205603779e-05,
      "loss": 4.9535,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909462886613473e-05,
      "loss": 4.9521,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908624291862421e-05,
      "loss": 4.9561,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907785697111369e-05,
      "loss": 4.9489,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906947102360317e-05,
      "loss": 4.9277,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906108507609265e-05,
      "loss": 4.9307,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905269912858213e-05,
      "loss": 4.9355,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904431318107161e-05,
      "loss": 4.9412,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903592723356109e-05,
      "loss": 4.9279,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90275576648543e-05,
      "loss": 4.9284,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901917171734378e-05,
      "loss": 4.9217,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901078576983326e-05,
      "loss": 4.9232,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900239982232274e-05,
      "loss": 4.9081,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899403025361595e-05,
      "loss": 4.909,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.898564430610543e-05,
      "loss": 4.8983,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897725835859491e-05,
      "loss": 4.9041,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968872411084396e-05,
      "loss": 4.8847,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960502842377605e-05,
      "loss": 4.8979,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952116894867085e-05,
      "loss": 4.8874,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943730947356565e-05,
      "loss": 4.8943,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8665,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.8771,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918573104825e-05,
      "loss": 4.8866,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891018715731448e-05,
      "loss": 4.86,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890180120980396e-05,
      "loss": 4.8784,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889343164109717e-05,
      "loss": 4.8629,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888504569358665e-05,
      "loss": 4.8651,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8876659746076134e-05,
      "loss": 4.8549,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8868273798565614e-05,
      "loss": 4.8515,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885990422985882e-05,
      "loss": 4.8522,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515182823483e-05,
      "loss": 4.851,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884313233483778e-05,
      "loss": 4.8512,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883474638732726e-05,
      "loss": 4.8428,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882637681862047e-05,
      "loss": 4.847,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881799087110995e-05,
      "loss": 4.8448,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880960492359943e-05,
      "loss": 4.8248,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880121897608891e-05,
      "loss": 4.8251,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879284940738212e-05,
      "loss": 4.8155,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87844634598716e-05,
      "loss": 4.8191,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877607751236109e-05,
      "loss": 4.8171,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876769156485057e-05,
      "loss": 4.8103,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875930561734005e-05,
      "loss": 4.8072,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875091966982953e-05,
      "loss": 4.8082,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.768613815307617,
      "eval_runtime": 545.152,
      "eval_samples_per_second": 699.972,
      "eval_steps_per_second": 21.875,
      "step": 76319
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874253372231901e-05,
      "loss": 4.8151,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873414777480849e-05,
      "loss": 4.8087,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872576182729797e-05,
      "loss": 4.7953,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871739225859118e-05,
      "loss": 4.7973,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709006311080657e-05,
      "loss": 4.7921,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700620363570137e-05,
      "loss": 4.77,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692234416059616e-05,
      "loss": 4.784,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683848468549096e-05,
      "loss": 4.7862,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675462521038576e-05,
      "loss": 4.7692,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667092952331785e-05,
      "loss": 4.7833,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865870700482127e-05,
      "loss": 4.786,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865032105731075e-05,
      "loss": 4.7706,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8641935109800225e-05,
      "loss": 4.764,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633549162289705e-05,
      "loss": 4.7647,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8625163214779185e-05,
      "loss": 4.7659,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8616777267268665e-05,
      "loss": 4.7599,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608391319758145e-05,
      "loss": 4.7554,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8600005372247625e-05,
      "loss": 4.7521,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591619424737105e-05,
      "loss": 4.7497,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8583249856030314e-05,
      "loss": 4.7528,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574863908519794e-05,
      "loss": 4.7525,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566477961009274e-05,
      "loss": 4.7511,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558092013498754e-05,
      "loss": 4.7465,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854970606598824e-05,
      "loss": 4.7437,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854135287608518e-05,
      "loss": 4.7388,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853296692857466e-05,
      "loss": 4.7355,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852458098106414e-05,
      "loss": 4.7392,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851619503355362e-05,
      "loss": 4.7205,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85078090860431e-05,
      "loss": 4.7336,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849943951733631e-05,
      "loss": 4.7159,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8491069948629524e-05,
      "loss": 4.7203,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8482684001119e-05,
      "loss": 4.7292,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847429805360848e-05,
      "loss": 4.7264,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8465912106097964e-05,
      "loss": 4.7123,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8457526158587444e-05,
      "loss": 4.7046,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8449140211076924e-05,
      "loss": 4.7213,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440754263566404e-05,
      "loss": 4.7098,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432368316055884e-05,
      "loss": 4.7075,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8423982368545363e-05,
      "loss": 4.7042,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841561279983857e-05,
      "loss": 4.6911,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840722685232805e-05,
      "loss": 4.6875,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839884090481753e-05,
      "loss": 4.6894,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839045495730701e-05,
      "loss": 4.6925,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838206900979649e-05,
      "loss": 4.6875,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837368306228597e-05,
      "loss": 4.6894,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836529711477545e-05,
      "loss": 4.6821,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835691116726493e-05,
      "loss": 4.6846,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834855797736188e-05,
      "loss": 4.6961,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834017202985136e-05,
      "loss": 4.6759,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833178608234084e-05,
      "loss": 4.6774,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832340013483032e-05,
      "loss": 4.6677,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83150141873198e-05,
      "loss": 4.6886,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830662823980928e-05,
      "loss": 4.6581,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829824229229876e-05,
      "loss": 4.6814,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828985634478824e-05,
      "loss": 4.6577,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828147039727771e-05,
      "loss": 4.6665,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827308444976719e-05,
      "loss": 4.6657,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826469850225667e-05,
      "loss": 4.6604,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256328933549886e-05,
      "loss": 4.6535,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8247942986039366e-05,
      "loss": 4.6523,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8239557038528846e-05,
      "loss": 4.6401,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8231171091018326e-05,
      "loss": 4.6568,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222785143507806e-05,
      "loss": 4.6485,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214399195997286e-05,
      "loss": 4.6422,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8206013248486766e-05,
      "loss": 4.6409,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197643679779975e-05,
      "loss": 4.6397,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189257732269455e-05,
      "loss": 4.6254,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180871784758935e-05,
      "loss": 4.6483,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172485837248415e-05,
      "loss": 4.6315,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164099889737895e-05,
      "loss": 4.6327,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155713942227375e-05,
      "loss": 4.6414,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147344373520584e-05,
      "loss": 4.6264,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138958426010064e-05,
      "loss": 4.641,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813057247849955e-05,
      "loss": 4.6411,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812220290979276e-05,
      "loss": 4.6355,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811381696228224e-05,
      "loss": 4.6346,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810543101477172e-05,
      "loss": 4.617,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80970450672612e-05,
      "loss": 4.6131,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808865911975068e-05,
      "loss": 4.6229,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808028955104389e-05,
      "loss": 4.6253,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807190360353337e-05,
      "loss": 4.6167,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806351765602285e-05,
      "loss": 4.6127,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805513170851233e-05,
      "loss": 4.5971,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804674576100181e-05,
      "loss": 4.6059,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803835981349129e-05,
      "loss": 4.6098,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802997386598077e-05,
      "loss": 4.6104,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802158791847025e-05,
      "loss": 4.6163,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801320197095973e-05,
      "loss": 4.6157,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800481602344921e-05,
      "loss": 4.603,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799643007593869e-05,
      "loss": 4.5961,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798804412842817e-05,
      "loss": 4.5916,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797967455972138e-05,
      "loss": 4.5955,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797128861221086e-05,
      "loss": 4.5992,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796291904350407e-05,
      "loss": 4.5848,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7954533095993546e-05,
      "loss": 4.5865,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7946147148483026e-05,
      "loss": 4.5947,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7937761200972506e-05,
      "loss": 4.5983,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792939163226572e-05,
      "loss": 4.5878,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79210056847552e-05,
      "loss": 4.5928,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791261973724468e-05,
      "loss": 4.5946,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790423378973416e-05,
      "loss": 4.587,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789584784222364e-05,
      "loss": 4.5823,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788746189471312e-05,
      "loss": 4.5914,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787909232600633e-05,
      "loss": 4.5782,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787070637849581e-05,
      "loss": 4.5898,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786232043098529e-05,
      "loss": 4.5687,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785393448347477e-05,
      "loss": 4.5599,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784554853596425e-05,
      "loss": 4.5671,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783716258845373e-05,
      "loss": 4.5747,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782877664094321e-05,
      "loss": 4.5671,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782040707223642e-05,
      "loss": 4.5546,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7812021124725906e-05,
      "loss": 4.5549,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803635177215386e-05,
      "loss": 4.5635,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795249229704866e-05,
      "loss": 4.5743,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786863282194346e-05,
      "loss": 4.5625,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7778477334683826e-05,
      "loss": 4.5715,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7770091387173306e-05,
      "loss": 4.5648,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7761705439662786e-05,
      "loss": 4.5614,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7753335870955995e-05,
      "loss": 4.5577,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7744966302249204e-05,
      "loss": 4.5551,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7736580354738684e-05,
      "loss": 4.5508,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7728194407228164e-05,
      "loss": 4.5601,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7719808459717644e-05,
      "loss": 4.5457,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771143889101086e-05,
      "loss": 4.5486,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770305294350034e-05,
      "loss": 4.5513,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769466699598982e-05,
      "loss": 4.5568,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76862810484793e-05,
      "loss": 4.5359,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767791147977251e-05,
      "loss": 4.5407,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766954191106572e-05,
      "loss": 4.5631,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76611559635552e-05,
      "loss": 4.5367,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765277001604468e-05,
      "loss": 4.5556,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764438406853416e-05,
      "loss": 4.5372,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763599812102364e-05,
      "loss": 4.5501,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762761217351312e-05,
      "loss": 4.5362,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76192262260026e-05,
      "loss": 4.5338,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761084027849208e-05,
      "loss": 4.5397,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760245433098156e-05,
      "loss": 4.5381,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759406838347104e-05,
      "loss": 4.5458,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758568243596052e-05,
      "loss": 4.5405,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757729648845e-05,
      "loss": 4.5414,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756891054093948e-05,
      "loss": 4.5371,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756052459342896e-05,
      "loss": 4.5216,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755213864591844e-05,
      "loss": 4.5315,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7543769077211646e-05,
      "loss": 4.5179,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7535399508504855e-05,
      "loss": 4.5243,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7527013560994335e-05,
      "loss": 4.5279,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7518627613483815e-05,
      "loss": 4.5141,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510241665973295e-05,
      "loss": 4.5202,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750185571846278e-05,
      "loss": 4.5186,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.483644008636475,
      "eval_runtime": 542.8367,
      "eval_samples_per_second": 702.957,
      "eval_steps_per_second": 21.968,
      "step": 152638
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 6.18106757409527e+16,
  "trial_name": null,
  "trial_params": null
}