{
  "best_metric": 3.9505887031555176,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/full2/transformer/3/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": 10.9241,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 6.8212,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 6.1757,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 5.9703,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 5.808,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 5.6985,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 5.5946,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 5.5279,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 5.4387,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 5.3907,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 5.3396,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 5.3156,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 5.2603,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 5.1973,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 5.1716,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 5.1336,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 5.1049,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.0781,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.0526,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.0119,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.0047,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9823927859886547e-05,
      "loss": 4.9693,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 4.9494,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 4.9211,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 4.9195,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790400448648195e-05,
      "loss": 4.8928,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 4.8653,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 4.8486,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 4.8312,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756873037409844e-05,
      "loss": 4.8205,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748487089899324e-05,
      "loss": 4.7977,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740117521192533e-05,
      "loss": 4.7909,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731731573682013e-05,
      "loss": 4.7772,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.97233456261715e-05,
      "loss": 4.7644,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.971495967866098e-05,
      "loss": 4.7614,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 4.7466,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969820416244367e-05,
      "loss": 4.7305,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 4.7047,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 4.7002,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 4.6809,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 4.6805,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 4.6659,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 4.6594,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 4.6438,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 4.6489,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 4.6318,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 4.6351,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 4.6235,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 4.5969,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95892359812181e-05,
      "loss": 4.5892,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958085003370758e-05,
      "loss": 4.6022,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957246408619706e-05,
      "loss": 4.5871,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956407813868654e-05,
      "loss": 4.5717,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9555692191176016e-05,
      "loss": 4.5584,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9547306243665496e-05,
      "loss": 4.5618,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9538920296154976e-05,
      "loss": 4.5453,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 4.558,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.952216477993767e-05,
      "loss": 4.5245,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951377883242715e-05,
      "loss": 4.5362,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 4.5343,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 4.5112,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 4.5263,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94802514211888e-05,
      "loss": 4.4983,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947186547367828e-05,
      "loss": 4.4984,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946347952616776e-05,
      "loss": 4.4859,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945510995746097e-05,
      "loss": 4.5022,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944672400995045e-05,
      "loss": 4.474,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943833806243993e-05,
      "loss": 4.479,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 4.4624,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942156616741889e-05,
      "loss": 4.4604,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413196598712105e-05,
      "loss": 4.473,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940482703000532e-05,
      "loss": 4.4654,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396441082494794e-05,
      "loss": 4.456,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388055134984274e-05,
      "loss": 4.4562,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379669187473754e-05,
      "loss": 4.4617,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371283239963234e-05,
      "loss": 4.437,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.936291367125644e-05,
      "loss": 4.4508,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935452772374592e-05,
      "loss": 4.4244,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93461417762354e-05,
      "loss": 4.4234,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933775582872488e-05,
      "loss": 4.4223,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932936988121436e-05,
      "loss": 4.4254,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932098393370384e-05,
      "loss": 4.4175,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931259798619332e-05,
      "loss": 4.4251,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930422841748654e-05,
      "loss": 4.4086,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929584246997602e-05,
      "loss": 4.3894,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92874565224655e-05,
      "loss": 4.391,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927907057495498e-05,
      "loss": 4.4016,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927070100624819e-05,
      "loss": 4.3874,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92623314375414e-05,
      "loss": 4.3859,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925394549003088e-05,
      "loss": 4.389,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924555954252036e-05,
      "loss": 4.3712,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923717359500984e-05,
      "loss": 4.3787,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9228804026303046e-05,
      "loss": 4.373,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922041807879253e-05,
      "loss": 4.3774,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921203213128201e-05,
      "loss": 4.3744,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920364618377149e-05,
      "loss": 4.3688,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919526023626097e-05,
      "loss": 4.3495,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918687428875045e-05,
      "loss": 4.3566,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917848834123993e-05,
      "loss": 4.348,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917010239372941e-05,
      "loss": 4.3525,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.916171644621889e-05,
      "loss": 4.3569,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915333049870837e-05,
      "loss": 4.3363,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914494455119785e-05,
      "loss": 4.341,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136558603687325e-05,
      "loss": 4.3412,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912818903498054e-05,
      "loss": 4.3347,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911980308747002e-05,
      "loss": 4.3252,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911143351876323e-05,
      "loss": 4.3272,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910304757125272e-05,
      "loss": 4.3169,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90946616237422e-05,
      "loss": 4.3151,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908627567623168e-05,
      "loss": 4.3145,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907788972872115e-05,
      "loss": 4.3138,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906950378121063e-05,
      "loss": 4.3071,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906111783370011e-05,
      "loss": 4.3121,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905273188618959e-05,
      "loss": 4.3038,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90443623174828e-05,
      "loss": 4.3067,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903597636997228e-05,
      "loss": 4.3131,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027606801265495e-05,
      "loss": 4.3093,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019220853754975e-05,
      "loss": 4.3034,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010834906244455e-05,
      "loss": 4.294,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9002448958733935e-05,
      "loss": 4.2936,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994063011223415e-05,
      "loss": 4.2923,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985677063712895e-05,
      "loss": 4.2791,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977307495006104e-05,
      "loss": 4.2812,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968921547495584e-05,
      "loss": 4.2796,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960535599985064e-05,
      "loss": 4.2837,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952149652474544e-05,
      "loss": 4.2739,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943763704964024e-05,
      "loss": 4.2818,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935377757453504e-05,
      "loss": 4.2747,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926991809942984e-05,
      "loss": 4.2663,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891862224123619e-05,
      "loss": 4.261,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891023629372567e-05,
      "loss": 4.2785,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890185034621515e-05,
      "loss": 4.2694,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889346439870464e-05,
      "loss": 4.261,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888507845119412e-05,
      "loss": 4.2628,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88766925036836e-05,
      "loss": 4.2647,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886830655617308e-05,
      "loss": 4.2615,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885992060866256e-05,
      "loss": 4.2638,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885155103995577e-05,
      "loss": 4.2533,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884316509244525e-05,
      "loss": 4.2528,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883477914493473e-05,
      "loss": 4.2518,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882639319742421e-05,
      "loss": 4.2424,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881800724991369e-05,
      "loss": 4.2386,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880962130240316e-05,
      "loss": 4.2404,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880123535489264e-05,
      "loss": 4.2414,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879286578618586e-05,
      "loss": 4.234,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878449621747907e-05,
      "loss": 4.2425,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877611026996855e-05,
      "loss": 4.2337,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876772432245803e-05,
      "loss": 4.2361,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875933837494751e-05,
      "loss": 4.2409,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8750952427436986e-05,
      "loss": 4.2206,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.191587448120117,
      "eval_runtime": 303.7914,
      "eval_samples_per_second": 1256.095,
      "eval_steps_per_second": 39.254,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8742566479926466e-05,
      "loss": 4.2108,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8734180532415946e-05,
      "loss": 4.2076,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8725794584905426e-05,
      "loss": 4.2321,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871744139500237e-05,
      "loss": 4.2169,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870905544749185e-05,
      "loss": 4.2194,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870066949998133e-05,
      "loss": 4.2116,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.869228355247081e-05,
      "loss": 4.2051,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.868389760496029e-05,
      "loss": 4.196,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675528036253507e-05,
      "loss": 4.2055,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667142088742986e-05,
      "loss": 4.1978,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865875614123246e-05,
      "loss": 4.2208,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865037019372194e-05,
      "loss": 4.2054,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8642000625015155e-05,
      "loss": 4.187,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633614677504635e-05,
      "loss": 4.1882,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862522872999411e-05,
      "loss": 4.1902,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861684278248359e-05,
      "loss": 4.1772,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860845683497307e-05,
      "loss": 4.1913,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860007088746255e-05,
      "loss": 4.1867,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859168493995203e-05,
      "loss": 4.174,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858329899244151e-05,
      "loss": 4.2,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574913044930995e-05,
      "loss": 4.179,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566543476224204e-05,
      "loss": 4.1813,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558157528713684e-05,
      "loss": 4.1719,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549771581203164e-05,
      "loss": 4.1894,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541385633692644e-05,
      "loss": 4.1664,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532999686182124e-05,
      "loss": 4.1687,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524613738671604e-05,
      "loss": 4.1636,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851624416996481e-05,
      "loss": 4.1628,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850785822245429e-05,
      "loss": 4.1611,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849947227494377e-05,
      "loss": 4.1587,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849108632743325e-05,
      "loss": 4.1626,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848271675872646e-05,
      "loss": 4.1629,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847433081121595e-05,
      "loss": 4.16,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846594486370543e-05,
      "loss": 4.1648,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845755891619491e-05,
      "loss": 4.1591,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844918934748812e-05,
      "loss": 4.1589,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84408033999776e-05,
      "loss": 4.1444,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843241745246708e-05,
      "loss": 4.1477,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842403150495656e-05,
      "loss": 4.1376,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841566193624977e-05,
      "loss": 4.1462,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840727598873925e-05,
      "loss": 4.1376,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839889004122873e-05,
      "loss": 4.1471,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839050409371821e-05,
      "loss": 4.1289,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8382134525011416e-05,
      "loss": 4.1472,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83737485775009e-05,
      "loss": 4.1389,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836536262999038e-05,
      "loss": 4.1402,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835697668247986e-05,
      "loss": 4.1379,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834860711377307e-05,
      "loss": 4.1222,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834022116626255e-05,
      "loss": 4.1244,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833183521875203e-05,
      "loss": 4.1387,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832344927124151e-05,
      "loss": 4.1291,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831506332373099e-05,
      "loss": 4.1222,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83066937550242e-05,
      "loss": 4.1158,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829830780751368e-05,
      "loss": 4.1245,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828992186000316e-05,
      "loss": 4.1049,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828153591249264e-05,
      "loss": 4.1315,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273166343785856e-05,
      "loss": 4.1021,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264780396275336e-05,
      "loss": 4.1221,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256394448764816e-05,
      "loss": 4.1188,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8248008501254296e-05,
      "loss": 4.1058,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8239638932547505e-05,
      "loss": 4.1184,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8231252985036985e-05,
      "loss": 4.1064,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222867037526465e-05,
      "loss": 4.104,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214481090015945e-05,
      "loss": 4.0931,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8206111521309154e-05,
      "loss": 4.1177,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197725573798634e-05,
      "loss": 4.092,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189339626288114e-05,
      "loss": 4.1039,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180953678777594e-05,
      "loss": 4.0828,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817258411007081e-05,
      "loss": 4.095,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816419816256029e-05,
      "loss": 4.1056,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.815581221504977e-05,
      "loss": 4.1028,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814742626753924e-05,
      "loss": 4.0987,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813905669883246e-05,
      "loss": 4.1031,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813067075132194e-05,
      "loss": 4.1071,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812228480381141e-05,
      "loss": 4.0928,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811389885630089e-05,
      "loss": 4.1021,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810552928759411e-05,
      "loss": 4.0867,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809714334008359e-05,
      "loss": 4.0894,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808875739257306e-05,
      "loss": 4.0845,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808037144506255e-05,
      "loss": 4.0939,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807200187635576e-05,
      "loss": 4.0931,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8063615928845243e-05,
      "loss": 4.0986,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8055229981334717e-05,
      "loss": 4.0888,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046844033824197e-05,
      "loss": 4.0683,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803847446511741e-05,
      "loss": 4.0729,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8030088517606886e-05,
      "loss": 4.0847,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8021702570096366e-05,
      "loss": 4.0794,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8013316622585846e-05,
      "loss": 4.0761,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800494705387906e-05,
      "loss": 4.0831,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799657748517227e-05,
      "loss": 4.0694,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798819153766175e-05,
      "loss": 4.0714,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797980559015123e-05,
      "loss": 4.0745,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797141964264071e-05,
      "loss": 4.0829,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796303369513019e-05,
      "loss": 4.0808,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795464774761967e-05,
      "loss": 4.0838,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794626180010915e-05,
      "loss": 4.0589,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793787585259863e-05,
      "loss": 4.0692,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792950628389184e-05,
      "loss": 4.0653,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792112033638132e-05,
      "loss": 4.0667,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79127343888708e-05,
      "loss": 4.0789,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790434844136028e-05,
      "loss": 4.0574,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789597887265349e-05,
      "loss": 4.062,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788759292514297e-05,
      "loss": 4.0692,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787920697763245e-05,
      "loss": 4.0666,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870821030121935e-05,
      "loss": 4.0508,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786246784021888e-05,
      "loss": 4.0626,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785408189270836e-05,
      "loss": 4.0543,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784569594519783e-05,
      "loss": 4.0507,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783730999768731e-05,
      "loss": 4.0485,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782892405017679e-05,
      "loss": 4.0504,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782053810266627e-05,
      "loss": 4.0483,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.781215215515575e-05,
      "loss": 4.0593,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780376620764523e-05,
      "loss": 4.0458,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779539663893844e-05,
      "loss": 4.0548,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778701069142792e-05,
      "loss": 4.0652,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77786247439174e-05,
      "loss": 4.0572,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777025517521062e-05,
      "loss": 4.0558,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77618692277001e-05,
      "loss": 4.0504,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775348328018958e-05,
      "loss": 4.0468,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774509733267906e-05,
      "loss": 4.0483,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773671138516854e-05,
      "loss": 4.0392,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7728341816461747e-05,
      "loss": 4.0396,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7719955868951227e-05,
      "loss": 4.0421,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7711569921440707e-05,
      "loss": 4.0472,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7703183973930186e-05,
      "loss": 4.0422,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7694798026419666e-05,
      "loss": 4.0442,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686412078909146e-05,
      "loss": 4.0435,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7678026131398626e-05,
      "loss": 4.0316,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766965656269184e-05,
      "loss": 4.031,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766127061518132e-05,
      "loss": 4.0513,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76528846676708e-05,
      "loss": 4.0454,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764449872016028e-05,
      "loss": 4.0391,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763612915145349e-05,
      "loss": 4.0339,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762774320394297e-05,
      "loss": 4.0406,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761935725643245e-05,
      "loss": 4.037,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761097130892193e-05,
      "loss": 4.0458,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760260174021514e-05,
      "loss": 4.036,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759421579270462e-05,
      "loss": 4.0347,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75858298451941e-05,
      "loss": 4.0337,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757744389768358e-05,
      "loss": 4.0291,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7569074328976796e-05,
      "loss": 4.0231,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7560688381466276e-05,
      "loss": 4.0293,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7552302433955756e-05,
      "loss": 4.034,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7543916486445236e-05,
      "loss": 4.0184,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7535546917738445e-05,
      "loss": 4.0364,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7527160970227925e-05,
      "loss": 4.0276,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7518775022717405e-05,
      "loss": 4.0261,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510389075206885e-05,
      "loss": 4.0301,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7502019506500094e-05,
      "loss": 4.0169,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.023649215698242,
      "eval_runtime": 303.8573,
      "eval_samples_per_second": 1255.823,
      "eval_steps_per_second": 39.245,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7493633558989574e-05,
      "loss": 4.0052,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7485247611479054e-05,
      "loss": 4.0069,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7476861663968534e-05,
      "loss": 4.0317,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7468475716458014e-05,
      "loss": 4.0181,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7460089768947494e-05,
      "loss": 4.0264,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7451703821436974e-05,
      "loss": 4.0103,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7443317873926453e-05,
      "loss": 4.0143,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7434931926415933e-05,
      "loss": 4.0003,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7426545978905413e-05,
      "loss": 4.0137,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.741816003139489e-05,
      "loss": 4.0083,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.74097904626881e-05,
      "loss": 4.0262,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740140451517758e-05,
      "loss": 4.0187,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.739303494647079e-05,
      "loss": 3.9954,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.738464899896027e-05,
      "loss": 4.0004,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.737626305144975e-05,
      "loss": 4.0056,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736787710393923e-05,
      "loss": 3.99,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735949115642871e-05,
      "loss": 4.0028,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73511052089182e-05,
      "loss": 4.0038,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.734271926140768e-05,
      "loss": 3.9945,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733433331389716e-05,
      "loss": 4.0224,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732594736638664e-05,
      "loss": 3.9979,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.731756141887612e-05,
      "loss": 4.0057,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730919185016933e-05,
      "loss": 3.9946,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730080590265881e-05,
      "loss": 4.0123,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.729241995514829e-05,
      "loss": 3.9882,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728403400763777e-05,
      "loss": 3.9949,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.727564806012725e-05,
      "loss": 3.9899,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7267278491420456e-05,
      "loss": 3.9943,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7258892543909936e-05,
      "loss": 3.9874,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7250506596399416e-05,
      "loss": 3.9863,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7242120648888896e-05,
      "loss": 3.9943,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.723373470137838e-05,
      "loss": 3.9957,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7225348753867856e-05,
      "loss": 3.9883,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7216962806357336e-05,
      "loss": 3.9999,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720859323765055e-05,
      "loss": 3.9871,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720020729014003e-05,
      "loss": 3.9926,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7191821342629505e-05,
      "loss": 3.9803,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7183435395118985e-05,
      "loss": 3.9849,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7175049447608465e-05,
      "loss": 3.9718,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716667987890168e-05,
      "loss": 3.9859,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7158293931391154e-05,
      "loss": 3.9738,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7149907983880634e-05,
      "loss": 3.9883,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714152203637012e-05,
      "loss": 3.9713,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71331360888596e-05,
      "loss": 3.9832,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.712475014134908e-05,
      "loss": 3.9873,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711636419383856e-05,
      "loss": 3.9821,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710797824632804e-05,
      "loss": 3.9833,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709960867762125e-05,
      "loss": 3.9654,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709122273011073e-05,
      "loss": 3.9684,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.708283678260021e-05,
      "loss": 3.9838,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.707445083508969e-05,
      "loss": 3.9744,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70660812663829e-05,
      "loss": 3.9714,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.705769531887238e-05,
      "loss": 3.9612,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704930937136186e-05,
      "loss": 3.9736,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704092342385134e-05,
      "loss": 3.9515,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7032553855144554e-05,
      "loss": 3.9843,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7024167907634034e-05,
      "loss": 3.9517,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7015781960123514e-05,
      "loss": 3.9724,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7007396012612994e-05,
      "loss": 3.9738,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699904282270993e-05,
      "loss": 3.9556,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699065687519941e-05,
      "loss": 3.9752,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.698227092768889e-05,
      "loss": 3.9597,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.697388498017837e-05,
      "loss": 3.9588,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.696549903266785e-05,
      "loss": 3.9478,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.695711308515733e-05,
      "loss": 3.9722,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694874351645054e-05,
      "loss": 3.9511,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694035756894003e-05,
      "loss": 3.9581,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.693197162142951e-05,
      "loss": 3.9449,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.692358567391899e-05,
      "loss": 3.9509,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69152161052122e-05,
      "loss": 3.9637,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6906830157701677e-05,
      "loss": 3.9667,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6898444210191157e-05,
      "loss": 3.9525,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6890058262680636e-05,
      "loss": 3.9677,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6881672315170116e-05,
      "loss": 3.9663,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6873286367659596e-05,
      "loss": 3.957,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6864900420149076e-05,
      "loss": 3.9579,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6856514472638556e-05,
      "loss": 3.9571,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6848144903931765e-05,
      "loss": 3.9515,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683977533522498e-05,
      "loss": 3.9489,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683138938771446e-05,
      "loss": 3.9565,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.682300344020394e-05,
      "loss": 3.9604,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.681461749269342e-05,
      "loss": 3.9661,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68062315451829e-05,
      "loss": 3.9524,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679784559767238e-05,
      "loss": 3.9316,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678945965016186e-05,
      "loss": 3.9422,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678109008145507e-05,
      "loss": 3.9524,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677270413394455e-05,
      "loss": 3.9471,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676431818643403e-05,
      "loss": 3.9438,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.675593223892351e-05,
      "loss": 3.9512,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674756267021672e-05,
      "loss": 3.945,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.67391767227062e-05,
      "loss": 3.9376,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673079077519568e-05,
      "loss": 3.9451,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672240482768516e-05,
      "loss": 3.9541,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671401888017464e-05,
      "loss": 3.9493,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6705649311467855e-05,
      "loss": 3.9553,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6697263363957335e-05,
      "loss": 3.936,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6688877416446815e-05,
      "loss": 3.9418,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668049146893629e-05,
      "loss": 3.939,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.667210552142577e-05,
      "loss": 3.9439,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.666371957391525e-05,
      "loss": 3.9508,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.665535000520846e-05,
      "loss": 3.9352,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664696405769794e-05,
      "loss": 3.9357,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663857811018742e-05,
      "loss": 3.944,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66301921626769e-05,
      "loss": 3.9467,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6621806215166383e-05,
      "loss": 3.9267,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661343664645959e-05,
      "loss": 3.9351,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660505069894907e-05,
      "loss": 3.9377,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.659666475143855e-05,
      "loss": 3.9278,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658827880392803e-05,
      "loss": 3.9274,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657989285641751e-05,
      "loss": 3.9298,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657152328771072e-05,
      "loss": 3.9321,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65631373402002e-05,
      "loss": 3.9369,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.655475139268968e-05,
      "loss": 3.9253,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654636544517916e-05,
      "loss": 3.9325,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.653799587647237e-05,
      "loss": 3.9431,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652960992896185e-05,
      "loss": 3.9443,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6521240360255066e-05,
      "loss": 3.9347,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6512854412744546e-05,
      "loss": 3.9352,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6504468465234026e-05,
      "loss": 3.9296,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6496082517723506e-05,
      "loss": 3.9319,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6487696570212986e-05,
      "loss": 3.9198,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6479310622702466e-05,
      "loss": 3.9219,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6470924675191946e-05,
      "loss": 3.9262,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6462538727681426e-05,
      "loss": 3.936,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6454169158974635e-05,
      "loss": 3.9274,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6445783211464115e-05,
      "loss": 3.9296,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6437397263953595e-05,
      "loss": 3.9293,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6429011316443075e-05,
      "loss": 3.9188,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642064174773629e-05,
      "loss": 3.9162,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641225580022577e-05,
      "loss": 3.9381,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640386985271525e-05,
      "loss": 3.9355,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639548390520473e-05,
      "loss": 3.9241,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638709795769421e-05,
      "loss": 3.9217,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637872838898742e-05,
      "loss": 3.925,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.63703424414769e-05,
      "loss": 3.9272,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636195649396638e-05,
      "loss": 3.9389,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635357054645586e-05,
      "loss": 3.9228,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.634520097774907e-05,
      "loss": 3.9218,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633681503023855e-05,
      "loss": 3.9223,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632842908272803e-05,
      "loss": 3.9205,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632004313521751e-05,
      "loss": 3.9166,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631165718770699e-05,
      "loss": 3.9181,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6303287619000204e-05,
      "loss": 3.9232,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6294901671489684e-05,
      "loss": 3.9146,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6286515723979164e-05,
      "loss": 3.925,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6278129776468644e-05,
      "loss": 3.9244,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626974382895812e-05,
      "loss": 3.917,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626137426025133e-05,
      "loss": 3.9226,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625298831274081e-05,
      "loss": 3.9099,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 3.9505887031555176,
      "eval_runtime": 308.0876,
      "eval_samples_per_second": 1238.579,
      "eval_steps_per_second": 38.707,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.593503783284009e+17,
  "trial_name": null,
  "trial_params": null
}