{
  "best_metric": 3.975489854812622,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/existential-there-quantifier/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": 10.9613,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 6.8339,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 6.2109,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.0062,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 5.8426,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 5.7432,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 5.6318,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 5.5712,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 5.4912,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 5.4285,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 5.3855,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 5.3466,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 5.3034,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 5.2471,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 5.2143,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 5.1748,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 5.1393,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.1181,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.0901,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.058,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.0471,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9823927859886547e-05,
      "loss": 5.0125,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 4.9945,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 4.9666,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 4.9558,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790400448648195e-05,
      "loss": 4.9259,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 4.9074,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 4.8931,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 4.8721,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756873037409844e-05,
      "loss": 4.8469,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748487089899324e-05,
      "loss": 4.842,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740117521192533e-05,
      "loss": 4.8308,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731731573682013e-05,
      "loss": 4.8162,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.97233456261715e-05,
      "loss": 4.7985,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.971495967866098e-05,
      "loss": 4.7952,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 4.7685,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969820416244367e-05,
      "loss": 4.7618,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 4.7329,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 4.7346,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 4.7207,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 4.7176,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 4.7052,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 4.7026,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 4.6832,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 4.6759,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 4.6692,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 4.6532,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 4.6615,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9597605549924894e-05,
      "loss": 4.6353,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95892359812181e-05,
      "loss": 4.6231,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958085003370758e-05,
      "loss": 4.6346,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957246408619706e-05,
      "loss": 4.6118,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956407813868654e-05,
      "loss": 4.6059,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9555692191176016e-05,
      "loss": 4.585,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9547306243665496e-05,
      "loss": 4.5971,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9538920296154976e-05,
      "loss": 4.5731,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 4.6011,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522148401133936e-05,
      "loss": 4.565,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951376245362342e-05,
      "loss": 4.5783,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 4.5685,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 4.5436,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 4.5362,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948023504238507e-05,
      "loss": 4.5417,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947186547367828e-05,
      "loss": 4.5256,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946347952616776e-05,
      "loss": 4.5186,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 4.536,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944670763114672e-05,
      "loss": 4.5204,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94383216836362e-05,
      "loss": 4.5067,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 4.4903,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9421582546222625e-05,
      "loss": 4.4887,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413196598712105e-05,
      "loss": 4.5002,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404810651201585e-05,
      "loss": 4.5,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396424703691065e-05,
      "loss": 4.49,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388038756180545e-05,
      "loss": 4.4871,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379652808670025e-05,
      "loss": 4.4841,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371266861159505e-05,
      "loss": 4.4779,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362897292452714e-05,
      "loss": 4.4645,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354511344942194e-05,
      "loss": 4.463,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346125397431674e-05,
      "loss": 4.4578,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337739449921154e-05,
      "loss": 4.4467,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329353502410634e-05,
      "loss": 4.4489,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9320967554900114e-05,
      "loss": 4.4486,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9312581607389594e-05,
      "loss": 4.4504,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9304195659879074e-05,
      "loss": 4.4323,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929582609117229e-05,
      "loss": 4.4128,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92874565224655e-05,
      "loss": 4.4326,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927907057495498e-05,
      "loss": 4.4276,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927068462744446e-05,
      "loss": 4.4264,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.926229867993394e-05,
      "loss": 4.4162,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925392911122715e-05,
      "loss": 4.4187,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924554316371663e-05,
      "loss": 4.4077,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923715721620611e-05,
      "loss": 4.3976,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922877126869559e-05,
      "loss": 4.4022,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922038532118507e-05,
      "loss": 4.3986,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921199937367455e-05,
      "loss": 4.4049,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920361342616403e-05,
      "loss": 4.4033,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919522747865351e-05,
      "loss": 4.3976,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918684153114299e-05,
      "loss": 4.3773,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917845558363247e-05,
      "loss": 4.3796,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917006963612195e-05,
      "loss": 4.3823,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 4.3795,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.37,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9144928172394116e-05,
      "loss": 4.3676,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136542224883596e-05,
      "loss": 4.3571,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128172656176805e-05,
      "loss": 4.3719,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119786708666285e-05,
      "loss": 4.3425,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91114171399595e-05,
      "loss": 4.3551,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910303119244898e-05,
      "loss": 4.3543,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909464524493846e-05,
      "loss": 4.3446,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908625929742794e-05,
      "loss": 4.3427,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907787334991742e-05,
      "loss": 4.3478,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90694874024069e-05,
      "loss": 4.337,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906110145489638e-05,
      "loss": 4.343,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905271550738586e-05,
      "loss": 4.3389,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904434593867907e-05,
      "loss": 4.3308,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903595999116855e-05,
      "loss": 4.333,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.902757404365803e-05,
      "loss": 4.3387,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019220853754975e-05,
      "loss": 4.3284,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010834906244455e-05,
      "loss": 4.3176,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9002448958733935e-05,
      "loss": 4.3227,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994063011223415e-05,
      "loss": 4.3245,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985677063712895e-05,
      "loss": 4.3158,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977291116202375e-05,
      "loss": 4.3047,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968905168691855e-05,
      "loss": 4.3122,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960519221181335e-05,
      "loss": 4.3176,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952133273670815e-05,
      "loss": 4.3069,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943747326160294e-05,
      "loss": 4.3023,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935361378649774e-05,
      "loss": 4.2953,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926991809942984e-05,
      "loss": 4.2986,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918605862432463e-05,
      "loss": 4.2934,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910219914921943e-05,
      "loss": 4.3059,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8901833967411423e-05,
      "loss": 4.2984,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88934480199009e-05,
      "loss": 4.2776,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.2877,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887667612487986e-05,
      "loss": 4.2918,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886829017736934e-05,
      "loss": 4.2836,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885992060866256e-05,
      "loss": 4.2898,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885155103995577e-05,
      "loss": 4.2744,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884316509244525e-05,
      "loss": 4.2834,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883477914493473e-05,
      "loss": 4.2869,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882639319742421e-05,
      "loss": 4.2753,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881800724991369e-05,
      "loss": 4.2707,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880962130240316e-05,
      "loss": 4.2722,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880123535489264e-05,
      "loss": 4.2612,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879286578618586e-05,
      "loss": 4.26,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878447983867534e-05,
      "loss": 4.2728,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877609389116482e-05,
      "loss": 4.2535,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87677079436543e-05,
      "loss": 4.2583,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875933837494751e-05,
      "loss": 4.2611,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8750952427436986e-05,
      "loss": 4.2578,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.220180988311768,
      "eval_runtime": 302.6371,
      "eval_samples_per_second": 1260.886,
      "eval_steps_per_second": 39.404,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8742566479926466e-05,
      "loss": 4.2403,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8734180532415946e-05,
      "loss": 4.2417,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872581096370916e-05,
      "loss": 4.2581,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717425016198635e-05,
      "loss": 4.2436,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709039068688115e-05,
      "loss": 4.2538,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700653121177595e-05,
      "loss": 4.2272,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.869228355247081e-05,
      "loss": 4.2454,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.868389760496029e-05,
      "loss": 4.2313,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867551165744977e-05,
      "loss": 4.2303,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866712570993925e-05,
      "loss": 4.232,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865873976242873e-05,
      "loss": 4.2331,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865035381491821e-05,
      "loss": 4.2345,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8642000625015155e-05,
      "loss": 4.2182,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633614677504635e-05,
      "loss": 4.2164,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862522872999411e-05,
      "loss": 4.2129,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861684278248359e-05,
      "loss": 4.2115,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860845683497307e-05,
      "loss": 4.21,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860007088746255e-05,
      "loss": 4.2128,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859168493995203e-05,
      "loss": 4.2085,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858329899244151e-05,
      "loss": 4.2334,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574913044930995e-05,
      "loss": 4.206,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566527097420475e-05,
      "loss": 4.2123,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558157528713684e-05,
      "loss": 4.2057,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549771581203164e-05,
      "loss": 4.2113,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541385633692644e-05,
      "loss": 4.2003,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532999686182124e-05,
      "loss": 4.1999,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524613738671604e-05,
      "loss": 4.1976,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8516227791161084e-05,
      "loss": 4.1897,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8507841843650564e-05,
      "loss": 4.1765,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8499455896140044e-05,
      "loss": 4.1925,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849110270623698e-05,
      "loss": 4.1908,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848271675872646e-05,
      "loss": 4.1949,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847433081121595e-05,
      "loss": 4.1859,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846594486370543e-05,
      "loss": 4.1886,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845757529499864e-05,
      "loss": 4.1785,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844918934748812e-05,
      "loss": 4.1776,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84408033999776e-05,
      "loss": 4.1654,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843241745246708e-05,
      "loss": 4.1733,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842403150495656e-05,
      "loss": 4.1702,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841566193624977e-05,
      "loss": 4.1703,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840727598873925e-05,
      "loss": 4.169,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839889004122873e-05,
      "loss": 4.1775,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839050409371821e-05,
      "loss": 4.1668,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8382118146207687e-05,
      "loss": 4.1676,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83737485775009e-05,
      "loss": 4.169,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836536262999038e-05,
      "loss": 4.1552,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835697668247986e-05,
      "loss": 4.1664,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834859073496934e-05,
      "loss": 4.154,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834022116626255e-05,
      "loss": 4.1464,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833183521875203e-05,
      "loss": 4.1648,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832344927124151e-05,
      "loss": 4.149,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831506332373099e-05,
      "loss": 4.1468,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83066937550242e-05,
      "loss": 4.1398,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829830780751368e-05,
      "loss": 4.1479,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828992186000316e-05,
      "loss": 4.1315,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828153591249264e-05,
      "loss": 4.1612,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273166343785856e-05,
      "loss": 4.1378,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264780396275336e-05,
      "loss": 4.1526,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256394448764816e-05,
      "loss": 4.151,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8248008501254296e-05,
      "loss": 4.1301,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8239638932547505e-05,
      "loss": 4.1265,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8231252985036985e-05,
      "loss": 4.1371,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222867037526465e-05,
      "loss": 4.1266,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214481090015945e-05,
      "loss": 4.1214,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820609514250542e-05,
      "loss": 4.1401,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197725573798634e-05,
      "loss": 4.1358,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189339626288114e-05,
      "loss": 4.1225,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180953678777594e-05,
      "loss": 4.1159,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172567731267074e-05,
      "loss": 4.1183,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816419816256029e-05,
      "loss": 4.1304,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.815581221504977e-05,
      "loss": 4.1289,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814742626753924e-05,
      "loss": 4.128,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813904032002872e-05,
      "loss": 4.1283,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813067075132194e-05,
      "loss": 4.128,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812228480381141e-05,
      "loss": 4.1292,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811389885630089e-05,
      "loss": 4.1117,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810551290879037e-05,
      "loss": 4.1226,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809714334008359e-05,
      "loss": 4.1171,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808875739257306e-05,
      "loss": 4.1062,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808037144506255e-05,
      "loss": 4.1167,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807198549755203e-05,
      "loss": 4.1175,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8063615928845243e-05,
      "loss": 4.122,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8055229981334717e-05,
      "loss": 4.1038,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046844033824197e-05,
      "loss": 4.0931,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8038458086313677e-05,
      "loss": 4.1094,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8030072138803156e-05,
      "loss": 4.1072,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8021702570096366e-05,
      "loss": 4.111,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8013316622585846e-05,
      "loss": 4.1045,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8004930675075325e-05,
      "loss": 4.1022,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7996561106368535e-05,
      "loss": 4.1029,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7988175158858015e-05,
      "loss": 4.0905,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7979789211347494e-05,
      "loss": 4.0981,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797140326383698e-05,
      "loss": 4.0988,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796303369513019e-05,
      "loss": 4.1067,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795464774761967e-05,
      "loss": 4.1104,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794626180010915e-05,
      "loss": 4.1027,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793787585259863e-05,
      "loss": 4.0891,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792950628389184e-05,
      "loss": 4.0907,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792112033638132e-05,
      "loss": 4.0939,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79127343888708e-05,
      "loss": 4.1012,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790434844136028e-05,
      "loss": 4.087,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789596249384976e-05,
      "loss": 4.0865,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788759292514297e-05,
      "loss": 4.08,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787920697763245e-05,
      "loss": 4.0988,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870821030121935e-05,
      "loss": 4.0684,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862435082611415e-05,
      "loss": 4.0835,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7854065513904624e-05,
      "loss": 4.0878,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845679566394104e-05,
      "loss": 4.077,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837293618883584e-05,
      "loss": 4.0771,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7828907671373064e-05,
      "loss": 4.0888,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7820521723862544e-05,
      "loss": 4.0769,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7812135776352024e-05,
      "loss": 4.0805,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803749828841504e-05,
      "loss": 4.0846,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779538026013471e-05,
      "loss": 4.0742,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778699431262419e-05,
      "loss": 4.0827,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777860836511367e-05,
      "loss": 4.0886,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777023879640689e-05,
      "loss": 4.0785,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776185284889637e-05,
      "loss": 4.0689,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775346690138585e-05,
      "loss": 4.075,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774508095387533e-05,
      "loss": 4.0752,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773669500636481e-05,
      "loss": 4.0692,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772830905885429e-05,
      "loss": 4.0645,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771992311134377e-05,
      "loss": 4.0662,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771153716383325e-05,
      "loss": 4.0819,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770316759512646e-05,
      "loss": 4.0659,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769478164761594e-05,
      "loss": 4.0678,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768639570010542e-05,
      "loss": 4.0595,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76780097525949e-05,
      "loss": 4.0621,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7669640183888106e-05,
      "loss": 4.0604,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7661254236377586e-05,
      "loss": 4.0757,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765286828886707e-05,
      "loss": 4.0708,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764448234135655e-05,
      "loss": 4.0498,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763611277264976e-05,
      "loss": 4.0574,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762772682513924e-05,
      "loss": 4.0661,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761934087762872e-05,
      "loss": 4.0621,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7610954930118195e-05,
      "loss": 4.0686,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760258536141141e-05,
      "loss": 4.0543,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759419941390089e-05,
      "loss": 4.0614,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758581346639037e-05,
      "loss": 4.0664,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7577427518879844e-05,
      "loss": 4.0571,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756905795017306e-05,
      "loss": 4.056,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756067200266254e-05,
      "loss": 4.0567,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755228605515202e-05,
      "loss": 4.0496,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7543916486445236e-05,
      "loss": 4.0487,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7535530538934716e-05,
      "loss": 4.0614,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7527144591424196e-05,
      "loss": 4.0436,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751875864391367e-05,
      "loss": 4.0497,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751037269640315e-05,
      "loss": 4.0536,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7502003127696365e-05,
      "loss": 4.0491,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.049336910247803,
      "eval_runtime": 302.6955,
      "eval_samples_per_second": 1260.643,
      "eval_steps_per_second": 39.396,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7493617180185845e-05,
      "loss": 4.0332,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748523123267532e-05,
      "loss": 4.0386,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.74768452851648e-05,
      "loss": 4.0556,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746845933765428e-05,
      "loss": 4.0374,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7460073390143764e-05,
      "loss": 4.0554,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7451687442633244e-05,
      "loss": 4.0324,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7443301495122724e-05,
      "loss": 4.0484,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7434915547612204e-05,
      "loss": 4.0305,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7426529600101684e-05,
      "loss": 4.0414,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7418143652591164e-05,
      "loss": 4.0365,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7409757705080644e-05,
      "loss": 4.0367,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740138813637385e-05,
      "loss": 4.045,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.739303494647079e-05,
      "loss": 4.0259,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.738464899896027e-05,
      "loss": 4.0282,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.737626305144975e-05,
      "loss": 4.0239,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736787710393923e-05,
      "loss": 4.0221,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735949115642871e-05,
      "loss": 4.0258,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73511052089182e-05,
      "loss": 4.0289,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.734271926140768e-05,
      "loss": 4.0229,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733433331389716e-05,
      "loss": 4.0484,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732594736638664e-05,
      "loss": 4.0264,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.731756141887612e-05,
      "loss": 4.0341,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73091754713656e-05,
      "loss": 4.0254,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730080590265881e-05,
      "loss": 4.0319,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.729241995514829e-05,
      "loss": 4.0227,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7284050386441496e-05,
      "loss": 4.0212,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7275664438930976e-05,
      "loss": 4.0228,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7267278491420456e-05,
      "loss": 4.0148,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7258892543909936e-05,
      "loss": 4.0023,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7250506596399416e-05,
      "loss": 4.0164,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7242120648888896e-05,
      "loss": 4.0261,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.723373470137838e-05,
      "loss": 4.0201,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722536513267159e-05,
      "loss": 4.0196,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721697918516107e-05,
      "loss": 4.0199,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720860961645428e-05,
      "loss": 4.0096,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720022366894376e-05,
      "loss": 4.0091,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719183772143324e-05,
      "loss": 4.0029,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718345177392272e-05,
      "loss": 4.0039,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71750658264122e-05,
      "loss": 4.0026,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716667987890168e-05,
      "loss": 4.0053,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7158293931391154e-05,
      "loss": 4.0057,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7149907983880634e-05,
      "loss": 4.0155,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714153841517385e-05,
      "loss": 4.0029,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713315246766333e-05,
      "loss": 4.0071,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.712476652015281e-05,
      "loss": 4.0105,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711638057264229e-05,
      "loss": 3.9933,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7108011003935505e-05,
      "loss": 4.0071,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709962505642498e-05,
      "loss": 4.0004,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7091255487718194e-05,
      "loss": 3.9879,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7082869540207674e-05,
      "loss": 4.0056,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7074483592697154e-05,
      "loss": 3.9965,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706609764518663e-05,
      "loss": 3.9932,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.705771169767611e-05,
      "loss": 3.9901,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704932575016559e-05,
      "loss": 3.9935,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70409561814588e-05,
      "loss": 3.9755,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.703257023394828e-05,
      "loss": 4.0045,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702418428643776e-05,
      "loss": 3.9888,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701579833892724e-05,
      "loss": 4.0,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.700741239141672e-05,
      "loss": 4.0026,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69990264439062e-05,
      "loss": 3.9831,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699065687519941e-05,
      "loss": 3.9771,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.698227092768889e-05,
      "loss": 3.9922,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.697388498017837e-05,
      "loss": 3.9791,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.696549903266785e-05,
      "loss": 3.9728,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.695711308515733e-05,
      "loss": 3.9952,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694872713764681e-05,
      "loss": 3.9924,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694034119013629e-05,
      "loss": 3.9768,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.693197162142951e-05,
      "loss": 3.9702,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.692358567391899e-05,
      "loss": 3.9749,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.691519972640847e-05,
      "loss": 3.9854,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.690681377889795e-05,
      "loss": 3.9875,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.689842783138743e-05,
      "loss": 3.9834,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6890058262680636e-05,
      "loss": 3.9916,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6881672315170116e-05,
      "loss": 3.9852,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6873286367659596e-05,
      "loss": 3.992,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6864900420149076e-05,
      "loss": 3.9739,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6856514472638556e-05,
      "loss": 3.9804,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6848128525128036e-05,
      "loss": 3.9821,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6839758956421245e-05,
      "loss": 3.9677,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6831373008910725e-05,
      "loss": 3.979,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6822987061400205e-05,
      "loss": 3.9858,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.681460111388969e-05,
      "loss": 3.9861,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6806215166379165e-05,
      "loss": 3.9699,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6797829218868645e-05,
      "loss": 3.9568,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678945965016186e-05,
      "loss": 3.9748,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678107370265134e-05,
      "loss": 3.9691,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6772687755140814e-05,
      "loss": 3.9803,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6764301807630294e-05,
      "loss": 3.9727,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6755915860119774e-05,
      "loss": 3.9695,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674754629141298e-05,
      "loss": 3.97,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673916034390246e-05,
      "loss": 3.9602,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673077439639194e-05,
      "loss": 3.9676,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672238844888143e-05,
      "loss": 3.9679,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671400250137091e-05,
      "loss": 3.9802,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.670563293266412e-05,
      "loss": 3.9835,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66972469851536e-05,
      "loss": 3.9736,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668886103764308e-05,
      "loss": 3.9567,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668047509013256e-05,
      "loss": 3.9659,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.667208914262204e-05,
      "loss": 3.9659,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.666371957391525e-05,
      "loss": 3.9742,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.665533362640473e-05,
      "loss": 3.961,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664694767889421e-05,
      "loss": 3.9624,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663856173138369e-05,
      "loss": 3.9544,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663017578387317e-05,
      "loss": 3.9742,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6621806215166383e-05,
      "loss": 3.9441,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6613420267655863e-05,
      "loss": 3.9626,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660503432014534e-05,
      "loss": 3.9561,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.659664837263482e-05,
      "loss": 3.9587,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65882624251243e-05,
      "loss": 3.9521,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657987647761378e-05,
      "loss": 3.9656,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657149053010326e-05,
      "loss": 3.9607,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.656312096139647e-05,
      "loss": 3.9545,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.655473501388595e-05,
      "loss": 3.964,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654634906637543e-05,
      "loss": 3.9496,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.653796311886491e-05,
      "loss": 3.9676,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652959355015812e-05,
      "loss": 3.9713,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65212076026476e-05,
      "loss": 3.9564,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.651282165513708e-05,
      "loss": 3.9535,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.650443570762657e-05,
      "loss": 3.9537,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.649604976011605e-05,
      "loss": 3.9563,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.648768019140926e-05,
      "loss": 3.9484,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647929424389874e-05,
      "loss": 3.95,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647090829638822e-05,
      "loss": 3.9476,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64625223488777e-05,
      "loss": 3.9646,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6454152780170906e-05,
      "loss": 3.9523,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6445766832660386e-05,
      "loss": 3.9512,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6437380885149866e-05,
      "loss": 3.9431,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6428994937639346e-05,
      "loss": 3.9492,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642060899012882e-05,
      "loss": 3.9432,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6412223042618306e-05,
      "loss": 3.9615,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640385347391152e-05,
      "loss": 3.9581,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6395467526401e-05,
      "loss": 3.9406,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6387081578890475e-05,
      "loss": 3.9377,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6378695631379955e-05,
      "loss": 3.9534,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6370309683869435e-05,
      "loss": 3.9484,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6361923736358915e-05,
      "loss": 3.9592,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6353537788848395e-05,
      "loss": 3.9377,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6345151841337874e-05,
      "loss": 3.9535,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6336782272631084e-05,
      "loss": 3.9509,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6328396325120564e-05,
      "loss": 3.9502,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6320010377610043e-05,
      "loss": 3.9466,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631164080890326e-05,
      "loss": 3.9454,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630325486139274e-05,
      "loss": 3.9399,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629486891388222e-05,
      "loss": 3.9409,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.62864829663717e-05,
      "loss": 3.9497,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.627809701886118e-05,
      "loss": 3.9412,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626971107135066e-05,
      "loss": 3.9372,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626132512384014e-05,
      "loss": 3.9426,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625293917632962e-05,
      "loss": 3.9403,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 3.975489854812622,
      "eval_runtime": 305.2702,
      "eval_samples_per_second": 1250.011,
      "eval_steps_per_second": 39.064,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.5848446258584576e+17,
  "trial_name": null,
  "trial_params": null
}