{
  "best_metric": 4.338195323944092,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/existential-there-quantifier/lstm/1/checkpoints/checkpoint-228960",
  "epoch": 0.025000606015738065,
  "eval_steps": 10,
  "global_step": 228960,
  "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": 10.8197,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5496,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0587,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 7.0003,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9467,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9226,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7712,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6663,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5618,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4812,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.4138,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990777095618801e-05,
      "loss": 6.3469,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 6.2679,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 6.2028,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1428,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.0893,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0413,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 6.008,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9596,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9256,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.893,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.844,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8126,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.7802,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979875363855125e-05,
      "loss": 5.7629,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979036769104073e-05,
      "loss": 5.7265,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.978198174353021e-05,
      "loss": 5.6971,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6779,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.97652262273129e-05,
      "loss": 5.6502,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6223,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.5997,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.5903,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.5673,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 5.5576,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5353,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 5.516,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969818778363994e-05,
      "loss": 5.5113,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968980183612942e-05,
      "loss": 5.4703,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96814158886189e-05,
      "loss": 5.4597,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 5.4415,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4355,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.4107,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4177,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.3821,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963110020355577e-05,
      "loss": 5.3815,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962271425604525e-05,
      "loss": 5.3764,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.368,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960595873982795e-05,
      "loss": 5.3586,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3361,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.3157,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.328,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95724313285896e-05,
      "loss": 5.3083,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.2947,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 5.2771,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.2829,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2633,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2768,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2462,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2388,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9505360127309166e-05,
      "loss": 5.2396,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9496974179798646e-05,
      "loss": 5.2243,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948860461109186e-05,
      "loss": 5.2019,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948023504238507e-05,
      "loss": 5.2062,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947184909487455e-05,
      "loss": 5.1845,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946346314736403e-05,
      "loss": 5.1856,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.1951,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.1871,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.1688,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942993573612568e-05,
      "loss": 5.1493,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942154978861516e-05,
      "loss": 5.1433,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941316384110464e-05,
      "loss": 5.1539,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940477789359412e-05,
      "loss": 5.1439,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396408324887336e-05,
      "loss": 5.1386,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388022377376816e-05,
      "loss": 5.1373,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379636429866296e-05,
      "loss": 5.1358,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371250482355776e-05,
      "loss": 5.1271,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362864534845256e-05,
      "loss": 5.0987,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935447858733473e-05,
      "loss": 5.1043,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.934609263982421e-05,
      "loss": 5.0974,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933770669231369e-05,
      "loss": 5.0837,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.1039,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.0823,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931256522858586e-05,
      "loss": 5.0872,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930417928107534e-05,
      "loss": 5.0723,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295809712368554e-05,
      "loss": 5.0478,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0536,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0566,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0512,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92622823011302e-05,
      "loss": 5.0458,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.0322,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924551040610916e-05,
      "loss": 5.0309,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923712445859864e-05,
      "loss": 5.0197,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922873851108812e-05,
      "loss": 5.0126,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922036894238133e-05,
      "loss": 5.0105,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921198299487081e-05,
      "loss": 5.0034,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920359704736029e-05,
      "loss": 5.0118,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919521109984978e-05,
      "loss": 4.9931,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918682515233926e-05,
      "loss": 4.9869,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917843920482874e-05,
      "loss": 4.9804,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917005325731822e-05,
      "loss": 4.9826,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.916168368861143e-05,
      "loss": 4.9811,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.9709,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9144928172394116e-05,
      "loss": 4.9625,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136542224883596e-05,
      "loss": 4.947,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128156277373076e-05,
      "loss": 4.9555,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119770329862556e-05,
      "loss": 4.9424,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111384382352036e-05,
      "loss": 4.9377,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9102998434841516e-05,
      "loss": 4.9391,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909462886613473e-05,
      "loss": 4.9372,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908624291862421e-05,
      "loss": 4.9334,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907785697111369e-05,
      "loss": 4.9291,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906947102360317e-05,
      "loss": 4.9181,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906108507609265e-05,
      "loss": 4.9194,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905269912858213e-05,
      "loss": 4.9087,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904431318107161e-05,
      "loss": 4.9033,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903592723356109e-05,
      "loss": 4.9125,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90275576648543e-05,
      "loss": 4.903,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901918809614751e-05,
      "loss": 4.9011,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901080214863699e-05,
      "loss": 4.8915,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900241620112647e-05,
      "loss": 4.8998,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899403025361595e-05,
      "loss": 4.8955,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.898564430610543e-05,
      "loss": 4.8764,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897725835859491e-05,
      "loss": 4.8924,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968872411084396e-05,
      "loss": 4.8791,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960502842377605e-05,
      "loss": 4.8805,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952116894867085e-05,
      "loss": 4.8589,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943730947356565e-05,
      "loss": 4.8543,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8493,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.8538,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918589483628734e-05,
      "loss": 4.8411,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910203536118214e-05,
      "loss": 4.8599,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890181758860769e-05,
      "loss": 4.8563,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889343164109717e-05,
      "loss": 4.8386,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.8425,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887667612487986e-05,
      "loss": 4.8379,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886829017736934e-05,
      "loss": 4.839,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885990422985882e-05,
      "loss": 4.8388,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515182823483e-05,
      "loss": 4.8401,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884313233483778e-05,
      "loss": 4.8208,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883476276613099e-05,
      "loss": 4.8245,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882637681862047e-05,
      "loss": 4.8175,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881799087110995e-05,
      "loss": 4.8087,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880960492359943e-05,
      "loss": 4.8209,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880123535489264e-05,
      "loss": 4.8079,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879284940738212e-05,
      "loss": 4.7971,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87844634598716e-05,
      "loss": 4.8133,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877607751236109e-05,
      "loss": 4.795,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876769156485057e-05,
      "loss": 4.7923,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875932199614378e-05,
      "loss": 4.7955,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.7869,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.752319812774658,
      "eval_runtime": 371.8701,
      "eval_samples_per_second": 1026.141,
      "eval_steps_per_second": 32.068,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.7731,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873416415361222e-05,
      "loss": 4.7828,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87257782061017e-05,
      "loss": 4.7988,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871739225859118e-05,
      "loss": 4.7743,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709022689884386e-05,
      "loss": 4.7864,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700636742373866e-05,
      "loss": 4.7606,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692250794863346e-05,
      "loss": 4.7765,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683864847352826e-05,
      "loss": 4.7533,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867549527864604e-05,
      "loss": 4.7644,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866710933113552e-05,
      "loss": 4.7637,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658723383625e-05,
      "loss": 4.7656,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865035381491821e-05,
      "loss": 4.7624,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864196786740769e-05,
      "loss": 4.7452,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863358191989717e-05,
      "loss": 4.7406,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862519597238665e-05,
      "loss": 4.7364,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861681002487613e-05,
      "loss": 4.7408,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860844045616934e-05,
      "loss": 4.7538,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860005450865882e-05,
      "loss": 4.7361,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85916685611483e-05,
      "loss": 4.7426,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858328261363778e-05,
      "loss": 4.7542,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.857489666612726e-05,
      "loss": 4.7285,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.856651071861674e-05,
      "loss": 4.7256,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558124771106226e-05,
      "loss": 4.7253,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549755202399435e-05,
      "loss": 4.7377,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541369254888915e-05,
      "loss": 4.7161,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532983307378395e-05,
      "loss": 4.719,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524597359867875e-05,
      "loss": 4.7158,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851621141235735e-05,
      "loss": 4.7143,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850782546484683e-05,
      "loss": 4.698,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849943951733631e-05,
      "loss": 4.7018,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849105356982579e-05,
      "loss": 4.6985,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848266762231527e-05,
      "loss": 4.7072,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847428167480475e-05,
      "loss": 4.712,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846589572729423e-05,
      "loss": 4.6989,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8457526158587444e-05,
      "loss": 4.6939,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8449140211076924e-05,
      "loss": 4.7055,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440754263566404e-05,
      "loss": 4.6778,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432368316055884e-05,
      "loss": 4.6842,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8423982368545363e-05,
      "loss": 4.6729,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8415596421034843e-05,
      "loss": 4.6804,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840721047352432e-05,
      "loss": 4.6722,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839884090481753e-05,
      "loss": 4.6871,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839045495730701e-05,
      "loss": 4.6686,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838206900979649e-05,
      "loss": 4.6755,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837368306228597e-05,
      "loss": 4.6698,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836529711477545e-05,
      "loss": 4.675,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835691116726493e-05,
      "loss": 4.6698,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834852521975441e-05,
      "loss": 4.6601,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834013927224389e-05,
      "loss": 4.6475,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833176970353711e-05,
      "loss": 4.671,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832338375602659e-05,
      "loss": 4.6575,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831499780851606e-05,
      "loss": 4.6509,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830661186100554e-05,
      "loss": 4.6451,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829822591349502e-05,
      "loss": 4.6524,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82898399659845e-05,
      "loss": 4.6422,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828147039727771e-05,
      "loss": 4.6603,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827308444976719e-05,
      "loss": 4.6417,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826469850225667e-05,
      "loss": 4.6397,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825631255474615e-05,
      "loss": 4.6487,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.824792660723563e-05,
      "loss": 4.6292,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823954065972511e-05,
      "loss": 4.623,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8231171091018326e-05,
      "loss": 4.6348,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222785143507806e-05,
      "loss": 4.6162,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214399195997286e-05,
      "loss": 4.6231,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8206013248486766e-05,
      "loss": 4.6331,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197627300976246e-05,
      "loss": 4.6369,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189241353465726e-05,
      "loss": 4.622,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180855405955206e-05,
      "loss": 4.615,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172469458444685e-05,
      "loss": 4.6064,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164099889737895e-05,
      "loss": 4.627,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155713942227375e-05,
      "loss": 4.6201,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147327994716854e-05,
      "loss": 4.6211,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138942047206334e-05,
      "loss": 4.6187,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813057247849955e-05,
      "loss": 4.6278,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812218653098903e-05,
      "loss": 4.6229,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811381696228224e-05,
      "loss": 4.5999,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810543101477172e-05,
      "loss": 4.6113,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80970450672612e-05,
      "loss": 4.6103,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808867549855441e-05,
      "loss": 4.5921,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808028955104389e-05,
      "loss": 4.622,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807190360353337e-05,
      "loss": 4.6067,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806351765602285e-05,
      "loss": 4.6119,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805513170851233e-05,
      "loss": 4.6084,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804674576100181e-05,
      "loss": 4.5826,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803835981349129e-05,
      "loss": 4.598,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802997386598077e-05,
      "loss": 4.5967,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802158791847025e-05,
      "loss": 4.5988,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801320197095973e-05,
      "loss": 4.5967,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800481602344921e-05,
      "loss": 4.592,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799643007593869e-05,
      "loss": 4.5934,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798804412842817e-05,
      "loss": 4.5889,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797967455972138e-05,
      "loss": 4.574,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797128861221086e-05,
      "loss": 4.5811,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796290266470034e-05,
      "loss": 4.5837,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795451671718982e-05,
      "loss": 4.5868,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79461307696793e-05,
      "loss": 4.5754,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793774482216878e-05,
      "loss": 4.5675,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7929375253461986e-05,
      "loss": 4.5695,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792098930595147e-05,
      "loss": 4.5793,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791260335844095e-05,
      "loss": 4.5778,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790421741093043e-05,
      "loss": 4.571,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789584784222364e-05,
      "loss": 4.5621,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788746189471312e-05,
      "loss": 4.5552,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78790759472026e-05,
      "loss": 4.5664,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787068999969208e-05,
      "loss": 4.5527,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786230405218156e-05,
      "loss": 4.5563,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785393448347477e-05,
      "loss": 4.5582,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784554853596425e-05,
      "loss": 4.5593,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783716258845373e-05,
      "loss": 4.5562,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782877664094321e-05,
      "loss": 4.5601,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782039069343269e-05,
      "loss": 4.5469,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.781200474592217e-05,
      "loss": 4.5504,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780361879841165e-05,
      "loss": 4.5472,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795265608508595e-05,
      "loss": 4.5405,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786879660998075e-05,
      "loss": 4.5567,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7778493713487555e-05,
      "loss": 4.5508,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7770107765977035e-05,
      "loss": 4.5493,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7761721818466515e-05,
      "loss": 4.5423,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7753335870955995e-05,
      "loss": 4.549,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7744949923445475e-05,
      "loss": 4.5466,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7736563975934955e-05,
      "loss": 4.5323,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7728178028424435e-05,
      "loss": 4.5536,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771979208091391e-05,
      "loss": 4.5395,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771140613340339e-05,
      "loss": 4.5421,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7703020185892875e-05,
      "loss": 4.5258,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7694650617186084e-05,
      "loss": 4.5267,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686264669675564e-05,
      "loss": 4.5204,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767789510096878e-05,
      "loss": 4.5246,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766950915345826e-05,
      "loss": 4.5187,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766112320594773e-05,
      "loss": 4.5364,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765273725843721e-05,
      "loss": 4.5376,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764435131092669e-05,
      "loss": 4.5252,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763596536341617e-05,
      "loss": 4.5201,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762757941590565e-05,
      "loss": 4.5254,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761919346839513e-05,
      "loss": 4.5269,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761082389968834e-05,
      "loss": 4.5338,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760245433098156e-05,
      "loss": 4.5285,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759406838347104e-05,
      "loss": 4.5198,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758568243596052e-05,
      "loss": 4.5134,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757729648845e-05,
      "loss": 4.5203,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756891054093948e-05,
      "loss": 4.5134,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756052459342896e-05,
      "loss": 4.5177,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755213864591844e-05,
      "loss": 4.5143,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754375269840792e-05,
      "loss": 4.507,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7535383129701126e-05,
      "loss": 4.5215,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7526997182190606e-05,
      "loss": 4.5078,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7518627613483815e-05,
      "loss": 4.5013,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510241665973295e-05,
      "loss": 4.5067,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750185571846278e-05,
      "loss": 4.5021,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.473540782928467,
      "eval_runtime": 372.6137,
      "eval_samples_per_second": 1024.093,
      "eval_steps_per_second": 32.004,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749346977095226e-05,
      "loss": 4.4882,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748508382344174e-05,
      "loss": 4.5017,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747669787593122e-05,
      "loss": 4.518,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.74683119284207e-05,
      "loss": 4.4959,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745992598091018e-05,
      "loss": 4.5079,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745154003339966e-05,
      "loss": 4.4907,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744315408588914e-05,
      "loss": 4.5037,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743476813837862e-05,
      "loss": 4.4786,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7426382190868095e-05,
      "loss": 4.4985,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7417996243357575e-05,
      "loss": 4.4961,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7409610295847055e-05,
      "loss": 4.4961,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401224348336535e-05,
      "loss": 4.5026,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.739285477962975e-05,
      "loss": 4.4785,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.738446883211923e-05,
      "loss": 4.4801,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.737608288460871e-05,
      "loss": 4.4763,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736771331590192e-05,
      "loss": 4.4851,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73593273683914e-05,
      "loss": 4.4926,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735094142088088e-05,
      "loss": 4.4825,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.734255547337036e-05,
      "loss": 4.485,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733416952585984e-05,
      "loss": 4.5019,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732578357834932e-05,
      "loss": 4.479,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73173976308388e-05,
      "loss": 4.4731,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730901168332828e-05,
      "loss": 4.4801,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730062573581776e-05,
      "loss": 4.4901,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.729223978830724e-05,
      "loss": 4.473,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728385384079672e-05,
      "loss": 4.4736,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.72754678932862e-05,
      "loss": 4.4716,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726708194577568e-05,
      "loss": 4.4723,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.725869599826516e-05,
      "loss": 4.4573,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7250326429558375e-05,
      "loss": 4.4644,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7241940482047855e-05,
      "loss": 4.4662,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7233554534537335e-05,
      "loss": 4.4666,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722516858702681e-05,
      "loss": 4.4798,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721678263951629e-05,
      "loss": 4.4599,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7208413070809504e-05,
      "loss": 4.4636,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7200027123298984e-05,
      "loss": 4.4719,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719164117578846e-05,
      "loss": 4.4526,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718325522827794e-05,
      "loss": 4.4516,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717486928076742e-05,
      "loss": 4.4487,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716649971206063e-05,
      "loss": 4.4554,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715811376455011e-05,
      "loss": 4.4449,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714972781703959e-05,
      "loss": 4.4644,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714134186952907e-05,
      "loss": 4.4432,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713295592201855e-05,
      "loss": 4.46,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.712456997450803e-05,
      "loss": 4.4503,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711618402699751e-05,
      "loss": 4.4502,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710781445829072e-05,
      "loss": 4.4547,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70994285107802e-05,
      "loss": 4.4407,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709104256326968e-05,
      "loss": 4.4324,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.708267299456289e-05,
      "loss": 4.4561,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.707428704705237e-05,
      "loss": 4.4449,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706590109954185e-05,
      "loss": 4.4358,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.705751515203134e-05,
      "loss": 4.4378,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704912920452082e-05,
      "loss": 4.4452,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70407432570103e-05,
      "loss": 4.4263,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.703235730949978e-05,
      "loss": 4.4495,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702397136198926e-05,
      "loss": 4.4342,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701558541447874e-05,
      "loss": 4.4316,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7007215845771946e-05,
      "loss": 4.445,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6998829898261426e-05,
      "loss": 4.4247,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6990460329554635e-05,
      "loss": 4.421,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6982074382044115e-05,
      "loss": 4.4362,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6973688434533595e-05,
      "loss": 4.4178,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6965302487023075e-05,
      "loss": 4.4203,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6956916539512555e-05,
      "loss": 4.4329,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6948530592002035e-05,
      "loss": 4.4387,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6940144644491515e-05,
      "loss": 4.421,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6931758696980995e-05,
      "loss": 4.4218,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6923372749470475e-05,
      "loss": 4.4079,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6914986801959955e-05,
      "loss": 4.4322,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6906600854449435e-05,
      "loss": 4.425,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.689824766454638e-05,
      "loss": 4.4304,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688986171703586e-05,
      "loss": 4.4248,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688147576952534e-05,
      "loss": 4.434,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.687308982201481e-05,
      "loss": 4.4371,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.686470387450429e-05,
      "loss": 4.41,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.685631792699377e-05,
      "loss": 4.4202,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.684793197948325e-05,
      "loss": 4.4252,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683954603197274e-05,
      "loss": 4.4052,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683116008446222e-05,
      "loss": 4.4345,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68227741369517e-05,
      "loss": 4.421,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.681438818944118e-05,
      "loss": 4.4267,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680600224193066e-05,
      "loss": 4.4259,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679763267322387e-05,
      "loss": 4.4005,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678924672571335e-05,
      "loss": 4.4169,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678086077820283e-05,
      "loss": 4.4126,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677249120949604e-05,
      "loss": 4.4193,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676410526198552e-05,
      "loss": 4.4117,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6755719314475e-05,
      "loss": 4.415,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674733336696448e-05,
      "loss": 4.4136,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673894741945396e-05,
      "loss": 4.4094,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673057785074717e-05,
      "loss": 4.3976,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672219190323665e-05,
      "loss": 4.403,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671380595572613e-05,
      "loss": 4.4091,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.670542000821561e-05,
      "loss": 4.4127,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.669703406070509e-05,
      "loss": 4.405,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668864811319457e-05,
      "loss": 4.3924,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668026216568405e-05,
      "loss": 4.3974,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.667187621817353e-05,
      "loss": 4.4066,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6663490270663006e-05,
      "loss": 4.4032,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6655104323152486e-05,
      "loss": 4.399,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6646718375641966e-05,
      "loss": 4.3941,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6638332428131446e-05,
      "loss": 4.3844,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662996285942466e-05,
      "loss": 4.4016,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662157691191414e-05,
      "loss": 4.3798,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661320734320736e-05,
      "loss": 4.3924,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660482139569683e-05,
      "loss": 4.3844,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.659643544818631e-05,
      "loss": 4.4007,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658804950067579e-05,
      "loss": 4.3861,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657966355316527e-05,
      "loss": 4.397,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657127760565475e-05,
      "loss": 4.3847,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.656289165814423e-05,
      "loss": 4.3831,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.655450571063371e-05,
      "loss": 4.3897,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654613614192692e-05,
      "loss": 4.376,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65377501944164e-05,
      "loss": 4.395,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652936424690588e-05,
      "loss": 4.3929,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6520994678199095e-05,
      "loss": 4.3905,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6512608730688575e-05,
      "loss": 4.3808,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6504222783178055e-05,
      "loss": 4.3869,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6495836835667535e-05,
      "loss": 4.3904,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6487467266960744e-05,
      "loss": 4.3705,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6479081319450224e-05,
      "loss": 4.395,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6470695371939704e-05,
      "loss": 4.3841,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646232580323291e-05,
      "loss": 4.3839,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645393985572239e-05,
      "loss": 4.3737,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644555390821187e-05,
      "loss": 4.3753,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643716796070135e-05,
      "loss": 4.361,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642878201319083e-05,
      "loss": 4.3724,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642041244448405e-05,
      "loss": 4.366,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641202649697353e-05,
      "loss": 4.3818,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640364054946301e-05,
      "loss": 4.3851,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639525460195249e-05,
      "loss": 4.3763,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638686865444197e-05,
      "loss": 4.3707,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637848270693145e-05,
      "loss": 4.3733,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637009675942093e-05,
      "loss": 4.3774,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636171081191041e-05,
      "loss": 4.3824,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635332486439989e-05,
      "loss": 4.378,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.634493891688936e-05,
      "loss": 4.3664,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633658572698631e-05,
      "loss": 4.3736,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632819977947579e-05,
      "loss": 4.3755,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631981383196527e-05,
      "loss": 4.3623,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631142788445475e-05,
      "loss": 4.3728,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630304193694423e-05,
      "loss": 4.3667,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629465598943371e-05,
      "loss": 4.3624,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628627004192319e-05,
      "loss": 4.3761,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6277884094412666e-05,
      "loss": 4.3648,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6269498146902146e-05,
      "loss": 4.3581,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6261112199391626e-05,
      "loss": 4.3637,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6252726251881106e-05,
      "loss": 4.3565,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.338195323944092,
      "eval_runtime": 303.1267,
      "eval_samples_per_second": 1258.85,
      "eval_steps_per_second": 39.34,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.420222446476171e+16,
  "trial_name": null,
  "trial_params": null
}