{
  "best_metric": 0.20430698990821838,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.07931786634939521,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000396589331746976,
      "grad_norm": 6.674250602722168,
      "learning_rate": 1.001e-05,
      "loss": 4.3034,
      "step": 1
    },
    {
      "epoch": 0.000396589331746976,
      "eval_loss": 3.849916934967041,
      "eval_runtime": 46.9826,
      "eval_samples_per_second": 22.604,
      "eval_steps_per_second": 5.662,
      "step": 1
    },
    {
      "epoch": 0.000793178663493952,
      "grad_norm": 13.111907958984375,
      "learning_rate": 2.002e-05,
      "loss": 6.024,
      "step": 2
    },
    {
      "epoch": 0.001189767995240928,
      "grad_norm": 9.942648887634277,
      "learning_rate": 3.0029999999999995e-05,
      "loss": 5.3876,
      "step": 3
    },
    {
      "epoch": 0.001586357326987904,
      "grad_norm": 8.252339363098145,
      "learning_rate": 4.004e-05,
      "loss": 2.8694,
      "step": 4
    },
    {
      "epoch": 0.00198294665873488,
      "grad_norm": 10.045882225036621,
      "learning_rate": 5.005e-05,
      "loss": 2.5194,
      "step": 5
    },
    {
      "epoch": 0.002379535990481856,
      "grad_norm": 6.507627487182617,
      "learning_rate": 6.005999999999999e-05,
      "loss": 2.0718,
      "step": 6
    },
    {
      "epoch": 0.002776125322228832,
      "grad_norm": 7.497224807739258,
      "learning_rate": 7.006999999999998e-05,
      "loss": 2.275,
      "step": 7
    },
    {
      "epoch": 0.003172714653975808,
      "grad_norm": 5.831888675689697,
      "learning_rate": 8.008e-05,
      "loss": 1.198,
      "step": 8
    },
    {
      "epoch": 0.003569303985722784,
      "grad_norm": 6.313211917877197,
      "learning_rate": 9.009e-05,
      "loss": 1.3938,
      "step": 9
    },
    {
      "epoch": 0.00396589331746976,
      "grad_norm": 4.629086017608643,
      "learning_rate": 0.0001001,
      "loss": 1.3091,
      "step": 10
    },
    {
      "epoch": 0.004362482649216736,
      "grad_norm": 4.553199291229248,
      "learning_rate": 9.957315789473684e-05,
      "loss": 1.1133,
      "step": 11
    },
    {
      "epoch": 0.004759071980963712,
      "grad_norm": 5.859588623046875,
      "learning_rate": 9.904631578947367e-05,
      "loss": 0.9577,
      "step": 12
    },
    {
      "epoch": 0.005155661312710688,
      "grad_norm": 4.0679707527160645,
      "learning_rate": 9.851947368421052e-05,
      "loss": 0.8043,
      "step": 13
    },
    {
      "epoch": 0.005552250644457664,
      "grad_norm": 3.8273086547851562,
      "learning_rate": 9.799263157894736e-05,
      "loss": 0.6524,
      "step": 14
    },
    {
      "epoch": 0.00594883997620464,
      "grad_norm": 5.04181432723999,
      "learning_rate": 9.746578947368421e-05,
      "loss": 0.785,
      "step": 15
    },
    {
      "epoch": 0.006345429307951616,
      "grad_norm": 4.491847038269043,
      "learning_rate": 9.693894736842104e-05,
      "loss": 0.8275,
      "step": 16
    },
    {
      "epoch": 0.0067420186396985925,
      "grad_norm": 7.865641117095947,
      "learning_rate": 9.641210526315789e-05,
      "loss": 1.1221,
      "step": 17
    },
    {
      "epoch": 0.007138607971445568,
      "grad_norm": 5.287400722503662,
      "learning_rate": 9.588526315789473e-05,
      "loss": 0.7424,
      "step": 18
    },
    {
      "epoch": 0.007535197303192544,
      "grad_norm": 5.461188793182373,
      "learning_rate": 9.535842105263157e-05,
      "loss": 0.4901,
      "step": 19
    },
    {
      "epoch": 0.00793178663493952,
      "grad_norm": 3.19498348236084,
      "learning_rate": 9.483157894736841e-05,
      "loss": 0.5121,
      "step": 20
    },
    {
      "epoch": 0.008328375966686495,
      "grad_norm": 6.649216175079346,
      "learning_rate": 9.430473684210526e-05,
      "loss": 0.203,
      "step": 21
    },
    {
      "epoch": 0.008724965298433472,
      "grad_norm": 6.187960624694824,
      "learning_rate": 9.37778947368421e-05,
      "loss": 1.141,
      "step": 22
    },
    {
      "epoch": 0.009121554630180448,
      "grad_norm": 3.0972740650177,
      "learning_rate": 9.325105263157894e-05,
      "loss": 0.5017,
      "step": 23
    },
    {
      "epoch": 0.009518143961927425,
      "grad_norm": 4.420783042907715,
      "learning_rate": 9.272421052631578e-05,
      "loss": 0.7917,
      "step": 24
    },
    {
      "epoch": 0.009914733293674401,
      "grad_norm": 2.6894378662109375,
      "learning_rate": 9.219736842105263e-05,
      "loss": 0.2605,
      "step": 25
    },
    {
      "epoch": 0.010311322625421376,
      "grad_norm": 2.54012131690979,
      "learning_rate": 9.167052631578946e-05,
      "loss": 0.4831,
      "step": 26
    },
    {
      "epoch": 0.010707911957168352,
      "grad_norm": 2.170597791671753,
      "learning_rate": 9.114368421052632e-05,
      "loss": 0.3522,
      "step": 27
    },
    {
      "epoch": 0.011104501288915328,
      "grad_norm": 5.271856307983398,
      "learning_rate": 9.061684210526315e-05,
      "loss": 1.2457,
      "step": 28
    },
    {
      "epoch": 0.011501090620662305,
      "grad_norm": 2.632899522781372,
      "learning_rate": 9.009e-05,
      "loss": 0.4753,
      "step": 29
    },
    {
      "epoch": 0.01189767995240928,
      "grad_norm": 3.3594906330108643,
      "learning_rate": 8.956315789473683e-05,
      "loss": 0.6309,
      "step": 30
    },
    {
      "epoch": 0.012294269284156256,
      "grad_norm": 2.8951773643493652,
      "learning_rate": 8.903631578947368e-05,
      "loss": 0.439,
      "step": 31
    },
    {
      "epoch": 0.012690858615903232,
      "grad_norm": 4.174351215362549,
      "learning_rate": 8.850947368421052e-05,
      "loss": 0.5533,
      "step": 32
    },
    {
      "epoch": 0.013087447947650209,
      "grad_norm": 4.82841157913208,
      "learning_rate": 8.798263157894736e-05,
      "loss": 0.8201,
      "step": 33
    },
    {
      "epoch": 0.013484037279397185,
      "grad_norm": 4.0411858558654785,
      "learning_rate": 8.745578947368422e-05,
      "loss": 0.5246,
      "step": 34
    },
    {
      "epoch": 0.01388062661114416,
      "grad_norm": 3.1116089820861816,
      "learning_rate": 8.692894736842105e-05,
      "loss": 0.5387,
      "step": 35
    },
    {
      "epoch": 0.014277215942891136,
      "grad_norm": 3.7147281169891357,
      "learning_rate": 8.64021052631579e-05,
      "loss": 0.4417,
      "step": 36
    },
    {
      "epoch": 0.014673805274638112,
      "grad_norm": 3.4639275074005127,
      "learning_rate": 8.587526315789473e-05,
      "loss": 0.6997,
      "step": 37
    },
    {
      "epoch": 0.015070394606385089,
      "grad_norm": 5.578957557678223,
      "learning_rate": 8.534842105263157e-05,
      "loss": 1.3211,
      "step": 38
    },
    {
      "epoch": 0.015466983938132063,
      "grad_norm": 2.867260456085205,
      "learning_rate": 8.482157894736842e-05,
      "loss": 0.5126,
      "step": 39
    },
    {
      "epoch": 0.01586357326987904,
      "grad_norm": 3.506054401397705,
      "learning_rate": 8.429473684210525e-05,
      "loss": 0.4488,
      "step": 40
    },
    {
      "epoch": 0.016260162601626018,
      "grad_norm": 3.3142287731170654,
      "learning_rate": 8.376789473684211e-05,
      "loss": 0.373,
      "step": 41
    },
    {
      "epoch": 0.01665675193337299,
      "grad_norm": 1.8013468980789185,
      "learning_rate": 8.324105263157894e-05,
      "loss": 0.1376,
      "step": 42
    },
    {
      "epoch": 0.017053341265119967,
      "grad_norm": 2.2930386066436768,
      "learning_rate": 8.271421052631579e-05,
      "loss": 0.0869,
      "step": 43
    },
    {
      "epoch": 0.017449930596866944,
      "grad_norm": 0.5169830918312073,
      "learning_rate": 8.218736842105262e-05,
      "loss": 0.0196,
      "step": 44
    },
    {
      "epoch": 0.01784651992861392,
      "grad_norm": 1.5832732915878296,
      "learning_rate": 8.166052631578947e-05,
      "loss": 0.0804,
      "step": 45
    },
    {
      "epoch": 0.018243109260360896,
      "grad_norm": 2.339470386505127,
      "learning_rate": 8.113368421052631e-05,
      "loss": 0.4453,
      "step": 46
    },
    {
      "epoch": 0.018639698592107873,
      "grad_norm": 2.9261984825134277,
      "learning_rate": 8.060684210526315e-05,
      "loss": 0.2987,
      "step": 47
    },
    {
      "epoch": 0.01903628792385485,
      "grad_norm": 0.43078291416168213,
      "learning_rate": 8.008e-05,
      "loss": 0.0167,
      "step": 48
    },
    {
      "epoch": 0.019432877255601826,
      "grad_norm": 25.611412048339844,
      "learning_rate": 7.955315789473684e-05,
      "loss": 0.0732,
      "step": 49
    },
    {
      "epoch": 0.019829466587348802,
      "grad_norm": 2.956575393676758,
      "learning_rate": 7.902631578947368e-05,
      "loss": 0.1045,
      "step": 50
    },
    {
      "epoch": 0.019829466587348802,
      "eval_loss": 0.5711997747421265,
      "eval_runtime": 47.4495,
      "eval_samples_per_second": 22.382,
      "eval_steps_per_second": 5.606,
      "step": 50
    },
    {
      "epoch": 0.020226055919095775,
      "grad_norm": 8.624287605285645,
      "learning_rate": 7.849947368421052e-05,
      "loss": 1.3873,
      "step": 51
    },
    {
      "epoch": 0.02062264525084275,
      "grad_norm": 7.417449951171875,
      "learning_rate": 7.797263157894736e-05,
      "loss": 0.8735,
      "step": 52
    },
    {
      "epoch": 0.021019234582589728,
      "grad_norm": 4.000467300415039,
      "learning_rate": 7.744578947368421e-05,
      "loss": 0.8792,
      "step": 53
    },
    {
      "epoch": 0.021415823914336704,
      "grad_norm": 1.1016407012939453,
      "learning_rate": 7.691894736842104e-05,
      "loss": 0.1425,
      "step": 54
    },
    {
      "epoch": 0.02181241324608368,
      "grad_norm": 1.845973014831543,
      "learning_rate": 7.63921052631579e-05,
      "loss": 0.1722,
      "step": 55
    },
    {
      "epoch": 0.022209002577830657,
      "grad_norm": 2.5237021446228027,
      "learning_rate": 7.586526315789473e-05,
      "loss": 0.5188,
      "step": 56
    },
    {
      "epoch": 0.022605591909577633,
      "grad_norm": 1.7162082195281982,
      "learning_rate": 7.533842105263158e-05,
      "loss": 0.2142,
      "step": 57
    },
    {
      "epoch": 0.02300218124132461,
      "grad_norm": 3.1163854598999023,
      "learning_rate": 7.481157894736841e-05,
      "loss": 0.8212,
      "step": 58
    },
    {
      "epoch": 0.023398770573071586,
      "grad_norm": 1.6111253499984741,
      "learning_rate": 7.428473684210526e-05,
      "loss": 0.302,
      "step": 59
    },
    {
      "epoch": 0.02379535990481856,
      "grad_norm": 2.3448877334594727,
      "learning_rate": 7.375789473684209e-05,
      "loss": 0.5013,
      "step": 60
    },
    {
      "epoch": 0.024191949236565535,
      "grad_norm": 1.4863951206207275,
      "learning_rate": 7.323105263157895e-05,
      "loss": 0.0785,
      "step": 61
    },
    {
      "epoch": 0.02458853856831251,
      "grad_norm": 0.46337777376174927,
      "learning_rate": 7.270421052631578e-05,
      "loss": 0.0252,
      "step": 62
    },
    {
      "epoch": 0.024985127900059488,
      "grad_norm": 1.6508891582489014,
      "learning_rate": 7.217736842105263e-05,
      "loss": 0.2864,
      "step": 63
    },
    {
      "epoch": 0.025381717231806464,
      "grad_norm": 1.8168219327926636,
      "learning_rate": 7.165052631578947e-05,
      "loss": 0.1634,
      "step": 64
    },
    {
      "epoch": 0.02577830656355344,
      "grad_norm": 1.4724931716918945,
      "learning_rate": 7.11236842105263e-05,
      "loss": 0.1565,
      "step": 65
    },
    {
      "epoch": 0.026174895895300417,
      "grad_norm": 1.3949257135391235,
      "learning_rate": 7.059684210526315e-05,
      "loss": 0.2016,
      "step": 66
    },
    {
      "epoch": 0.026571485227047394,
      "grad_norm": 5.898593902587891,
      "learning_rate": 7.006999999999998e-05,
      "loss": 0.2504,
      "step": 67
    },
    {
      "epoch": 0.02696807455879437,
      "grad_norm": 1.9633386135101318,
      "learning_rate": 6.954315789473684e-05,
      "loss": 0.221,
      "step": 68
    },
    {
      "epoch": 0.027364663890541343,
      "grad_norm": 2.800323963165283,
      "learning_rate": 6.901631578947368e-05,
      "loss": 0.457,
      "step": 69
    },
    {
      "epoch": 0.02776125322228832,
      "grad_norm": 1.5002562999725342,
      "learning_rate": 6.848947368421052e-05,
      "loss": 0.0964,
      "step": 70
    },
    {
      "epoch": 0.028157842554035296,
      "grad_norm": 2.7115468978881836,
      "learning_rate": 6.796263157894737e-05,
      "loss": 0.4686,
      "step": 71
    },
    {
      "epoch": 0.028554431885782272,
      "grad_norm": 4.224088668823242,
      "learning_rate": 6.74357894736842e-05,
      "loss": 0.9701,
      "step": 72
    },
    {
      "epoch": 0.02895102121752925,
      "grad_norm": 4.742822647094727,
      "learning_rate": 6.690894736842105e-05,
      "loss": 0.7263,
      "step": 73
    },
    {
      "epoch": 0.029347610549276225,
      "grad_norm": 2.388258934020996,
      "learning_rate": 6.638210526315788e-05,
      "loss": 0.3153,
      "step": 74
    },
    {
      "epoch": 0.0297441998810232,
      "grad_norm": 2.4255616664886475,
      "learning_rate": 6.585526315789474e-05,
      "loss": 0.2614,
      "step": 75
    },
    {
      "epoch": 0.030140789212770178,
      "grad_norm": 1.9446436166763306,
      "learning_rate": 6.532842105263157e-05,
      "loss": 0.1088,
      "step": 76
    },
    {
      "epoch": 0.030537378544517154,
      "grad_norm": 2.5515191555023193,
      "learning_rate": 6.480157894736842e-05,
      "loss": 0.4212,
      "step": 77
    },
    {
      "epoch": 0.030933967876264127,
      "grad_norm": 1.997567892074585,
      "learning_rate": 6.427473684210526e-05,
      "loss": 0.2342,
      "step": 78
    },
    {
      "epoch": 0.0313305572080111,
      "grad_norm": 2.1094398498535156,
      "learning_rate": 6.37478947368421e-05,
      "loss": 0.2247,
      "step": 79
    },
    {
      "epoch": 0.03172714653975808,
      "grad_norm": 2.6329636573791504,
      "learning_rate": 6.322105263157894e-05,
      "loss": 0.431,
      "step": 80
    },
    {
      "epoch": 0.032123735871505056,
      "grad_norm": 2.7955644130706787,
      "learning_rate": 6.269421052631577e-05,
      "loss": 0.2616,
      "step": 81
    },
    {
      "epoch": 0.032520325203252036,
      "grad_norm": 2.9589076042175293,
      "learning_rate": 6.216736842105263e-05,
      "loss": 0.3463,
      "step": 82
    },
    {
      "epoch": 0.03291691453499901,
      "grad_norm": 2.4867746829986572,
      "learning_rate": 6.164052631578947e-05,
      "loss": 0.4058,
      "step": 83
    },
    {
      "epoch": 0.03331350386674598,
      "grad_norm": 12.15013313293457,
      "learning_rate": 6.111368421052631e-05,
      "loss": 0.845,
      "step": 84
    },
    {
      "epoch": 0.03371009319849296,
      "grad_norm": 1.4896100759506226,
      "learning_rate": 6.058684210526315e-05,
      "loss": 0.0909,
      "step": 85
    },
    {
      "epoch": 0.034106682530239935,
      "grad_norm": 3.909092426300049,
      "learning_rate": 6.005999999999999e-05,
      "loss": 0.7803,
      "step": 86
    },
    {
      "epoch": 0.034503271861986914,
      "grad_norm": 3.4637069702148438,
      "learning_rate": 5.953315789473684e-05,
      "loss": 0.2718,
      "step": 87
    },
    {
      "epoch": 0.03489986119373389,
      "grad_norm": 3.0589916706085205,
      "learning_rate": 5.9006315789473676e-05,
      "loss": 0.4081,
      "step": 88
    },
    {
      "epoch": 0.03529645052548087,
      "grad_norm": 0.2933317720890045,
      "learning_rate": 5.847947368421053e-05,
      "loss": 0.0098,
      "step": 89
    },
    {
      "epoch": 0.03569303985722784,
      "grad_norm": 2.7271687984466553,
      "learning_rate": 5.795263157894737e-05,
      "loss": 0.1337,
      "step": 90
    },
    {
      "epoch": 0.03608962918897482,
      "grad_norm": 0.08103972673416138,
      "learning_rate": 5.742578947368421e-05,
      "loss": 0.0037,
      "step": 91
    },
    {
      "epoch": 0.03648621852072179,
      "grad_norm": 2.7736382484436035,
      "learning_rate": 5.6898947368421046e-05,
      "loss": 0.2963,
      "step": 92
    },
    {
      "epoch": 0.036882807852468766,
      "grad_norm": 2.4266488552093506,
      "learning_rate": 5.6372105263157886e-05,
      "loss": 0.1306,
      "step": 93
    },
    {
      "epoch": 0.037279397184215746,
      "grad_norm": 1.565101981163025,
      "learning_rate": 5.584526315789473e-05,
      "loss": 0.0939,
      "step": 94
    },
    {
      "epoch": 0.03767598651596272,
      "grad_norm": 0.204191654920578,
      "learning_rate": 5.531842105263158e-05,
      "loss": 0.0056,
      "step": 95
    },
    {
      "epoch": 0.0380725758477097,
      "grad_norm": 1.7860132455825806,
      "learning_rate": 5.4791578947368424e-05,
      "loss": 0.2133,
      "step": 96
    },
    {
      "epoch": 0.03846916517945667,
      "grad_norm": 3.4425370693206787,
      "learning_rate": 5.426473684210526e-05,
      "loss": 0.2545,
      "step": 97
    },
    {
      "epoch": 0.03886575451120365,
      "grad_norm": 2.9134604930877686,
      "learning_rate": 5.37378947368421e-05,
      "loss": 0.282,
      "step": 98
    },
    {
      "epoch": 0.039262343842950624,
      "grad_norm": 2.0041611194610596,
      "learning_rate": 5.321105263157894e-05,
      "loss": 0.175,
      "step": 99
    },
    {
      "epoch": 0.039658933174697604,
      "grad_norm": 2.66727352142334,
      "learning_rate": 5.268421052631578e-05,
      "loss": 0.2079,
      "step": 100
    },
    {
      "epoch": 0.039658933174697604,
      "eval_loss": 0.2756502330303192,
      "eval_runtime": 47.4815,
      "eval_samples_per_second": 22.367,
      "eval_steps_per_second": 5.602,
      "step": 100
    },
    {
      "epoch": 0.04005552250644458,
      "grad_norm": 4.0711822509765625,
      "learning_rate": 5.2157368421052626e-05,
      "loss": 0.6342,
      "step": 101
    },
    {
      "epoch": 0.04045211183819155,
      "grad_norm": 1.8220709562301636,
      "learning_rate": 5.163052631578947e-05,
      "loss": 0.2614,
      "step": 102
    },
    {
      "epoch": 0.04084870116993853,
      "grad_norm": 0.1610630750656128,
      "learning_rate": 5.110368421052632e-05,
      "loss": 0.0077,
      "step": 103
    },
    {
      "epoch": 0.0412452905016855,
      "grad_norm": 1.660254955291748,
      "learning_rate": 5.057684210526316e-05,
      "loss": 0.2139,
      "step": 104
    },
    {
      "epoch": 0.04164187983343248,
      "grad_norm": 1.2034050226211548,
      "learning_rate": 5.005e-05,
      "loss": 0.1488,
      "step": 105
    },
    {
      "epoch": 0.042038469165179455,
      "grad_norm": 1.123096227645874,
      "learning_rate": 4.9523157894736836e-05,
      "loss": 0.1183,
      "step": 106
    },
    {
      "epoch": 0.042435058496926435,
      "grad_norm": 1.3603556156158447,
      "learning_rate": 4.899631578947368e-05,
      "loss": 0.2229,
      "step": 107
    },
    {
      "epoch": 0.04283164782867341,
      "grad_norm": 1.2517480850219727,
      "learning_rate": 4.846947368421052e-05,
      "loss": 0.1483,
      "step": 108
    },
    {
      "epoch": 0.04322823716042039,
      "grad_norm": 0.9318723082542419,
      "learning_rate": 4.794263157894737e-05,
      "loss": 0.0662,
      "step": 109
    },
    {
      "epoch": 0.04362482649216736,
      "grad_norm": 1.0203633308410645,
      "learning_rate": 4.7415789473684206e-05,
      "loss": 0.1234,
      "step": 110
    },
    {
      "epoch": 0.044021415823914334,
      "grad_norm": 1.8892359733581543,
      "learning_rate": 4.688894736842105e-05,
      "loss": 0.1998,
      "step": 111
    },
    {
      "epoch": 0.044418005155661314,
      "grad_norm": 0.6476733088493347,
      "learning_rate": 4.636210526315789e-05,
      "loss": 0.039,
      "step": 112
    },
    {
      "epoch": 0.04481459448740829,
      "grad_norm": 1.5910590887069702,
      "learning_rate": 4.583526315789473e-05,
      "loss": 0.2231,
      "step": 113
    },
    {
      "epoch": 0.045211183819155266,
      "grad_norm": 1.6052759885787964,
      "learning_rate": 4.530842105263158e-05,
      "loss": 0.2969,
      "step": 114
    },
    {
      "epoch": 0.04560777315090224,
      "grad_norm": 2.3403847217559814,
      "learning_rate": 4.4781578947368416e-05,
      "loss": 0.1473,
      "step": 115
    },
    {
      "epoch": 0.04600436248264922,
      "grad_norm": 3.6207292079925537,
      "learning_rate": 4.425473684210526e-05,
      "loss": 0.5701,
      "step": 116
    },
    {
      "epoch": 0.04640095181439619,
      "grad_norm": 0.6984104514122009,
      "learning_rate": 4.372789473684211e-05,
      "loss": 0.0504,
      "step": 117
    },
    {
      "epoch": 0.04679754114614317,
      "grad_norm": 2.5241734981536865,
      "learning_rate": 4.320105263157895e-05,
      "loss": 0.2509,
      "step": 118
    },
    {
      "epoch": 0.047194130477890145,
      "grad_norm": 1.5402919054031372,
      "learning_rate": 4.2674210526315786e-05,
      "loss": 0.1996,
      "step": 119
    },
    {
      "epoch": 0.04759071980963712,
      "grad_norm": 2.3560476303100586,
      "learning_rate": 4.2147368421052626e-05,
      "loss": 0.3034,
      "step": 120
    },
    {
      "epoch": 0.0479873091413841,
      "grad_norm": 1.9385439157485962,
      "learning_rate": 4.162052631578947e-05,
      "loss": 0.3015,
      "step": 121
    },
    {
      "epoch": 0.04838389847313107,
      "grad_norm": 1.4906829595565796,
      "learning_rate": 4.109368421052631e-05,
      "loss": 0.1302,
      "step": 122
    },
    {
      "epoch": 0.04878048780487805,
      "grad_norm": 2.4340176582336426,
      "learning_rate": 4.056684210526316e-05,
      "loss": 0.4329,
      "step": 123
    },
    {
      "epoch": 0.04917707713662502,
      "grad_norm": 2.110347032546997,
      "learning_rate": 4.004e-05,
      "loss": 0.3068,
      "step": 124
    },
    {
      "epoch": 0.049573666468372,
      "grad_norm": 1.2856132984161377,
      "learning_rate": 3.951315789473684e-05,
      "loss": 0.1518,
      "step": 125
    },
    {
      "epoch": 0.049970255800118976,
      "grad_norm": 2.3550281524658203,
      "learning_rate": 3.898631578947368e-05,
      "loss": 0.117,
      "step": 126
    },
    {
      "epoch": 0.050366845131865956,
      "grad_norm": 2.4217026233673096,
      "learning_rate": 3.845947368421052e-05,
      "loss": 0.4333,
      "step": 127
    },
    {
      "epoch": 0.05076343446361293,
      "grad_norm": 1.6932317018508911,
      "learning_rate": 3.7932631578947367e-05,
      "loss": 0.148,
      "step": 128
    },
    {
      "epoch": 0.0511600237953599,
      "grad_norm": 1.6000185012817383,
      "learning_rate": 3.7405789473684206e-05,
      "loss": 0.0932,
      "step": 129
    },
    {
      "epoch": 0.05155661312710688,
      "grad_norm": 1.805180311203003,
      "learning_rate": 3.6878947368421045e-05,
      "loss": 0.2301,
      "step": 130
    },
    {
      "epoch": 0.051953202458853855,
      "grad_norm": 2.063298463821411,
      "learning_rate": 3.635210526315789e-05,
      "loss": 0.1673,
      "step": 131
    },
    {
      "epoch": 0.052349791790600834,
      "grad_norm": 2.706456422805786,
      "learning_rate": 3.582526315789474e-05,
      "loss": 0.389,
      "step": 132
    },
    {
      "epoch": 0.05274638112234781,
      "grad_norm": 2.0284225940704346,
      "learning_rate": 3.5298421052631576e-05,
      "loss": 0.1468,
      "step": 133
    },
    {
      "epoch": 0.05314297045409479,
      "grad_norm": 1.6976109743118286,
      "learning_rate": 3.477157894736842e-05,
      "loss": 0.0762,
      "step": 134
    },
    {
      "epoch": 0.05353955978584176,
      "grad_norm": 1.7858450412750244,
      "learning_rate": 3.424473684210526e-05,
      "loss": 0.2102,
      "step": 135
    },
    {
      "epoch": 0.05393614911758874,
      "grad_norm": 1.9341448545455933,
      "learning_rate": 3.37178947368421e-05,
      "loss": 0.2661,
      "step": 136
    },
    {
      "epoch": 0.05433273844933571,
      "grad_norm": 3.936488628387451,
      "learning_rate": 3.319105263157894e-05,
      "loss": 0.6706,
      "step": 137
    },
    {
      "epoch": 0.054729327781082686,
      "grad_norm": 1.7080624103546143,
      "learning_rate": 3.2664210526315786e-05,
      "loss": 0.1062,
      "step": 138
    },
    {
      "epoch": 0.055125917112829666,
      "grad_norm": 1.5581916570663452,
      "learning_rate": 3.213736842105263e-05,
      "loss": 0.0934,
      "step": 139
    },
    {
      "epoch": 0.05552250644457664,
      "grad_norm": 1.3080987930297852,
      "learning_rate": 3.161052631578947e-05,
      "loss": 0.0814,
      "step": 140
    },
    {
      "epoch": 0.05591909577632362,
      "grad_norm": 2.823137044906616,
      "learning_rate": 3.108368421052632e-05,
      "loss": 0.4453,
      "step": 141
    },
    {
      "epoch": 0.05631568510807059,
      "grad_norm": 0.21419651806354523,
      "learning_rate": 3.0556842105263156e-05,
      "loss": 0.0073,
      "step": 142
    },
    {
      "epoch": 0.05671227443981757,
      "grad_norm": 0.22324123978614807,
      "learning_rate": 3.0029999999999995e-05,
      "loss": 0.0051,
      "step": 143
    },
    {
      "epoch": 0.057108863771564544,
      "grad_norm": 0.2457597404718399,
      "learning_rate": 2.9503157894736838e-05,
      "loss": 0.0057,
      "step": 144
    },
    {
      "epoch": 0.057505453103311524,
      "grad_norm": 3.4714629650115967,
      "learning_rate": 2.8976315789473684e-05,
      "loss": 0.5372,
      "step": 145
    },
    {
      "epoch": 0.0579020424350585,
      "grad_norm": 0.028041277080774307,
      "learning_rate": 2.8449473684210523e-05,
      "loss": 0.0009,
      "step": 146
    },
    {
      "epoch": 0.05829863176680547,
      "grad_norm": 1.089476466178894,
      "learning_rate": 2.7922631578947366e-05,
      "loss": 0.0318,
      "step": 147
    },
    {
      "epoch": 0.05869522109855245,
      "grad_norm": 2.4500927925109863,
      "learning_rate": 2.7395789473684212e-05,
      "loss": 0.4794,
      "step": 148
    },
    {
      "epoch": 0.05909181043029942,
      "grad_norm": 0.0174604132771492,
      "learning_rate": 2.686894736842105e-05,
      "loss": 0.0006,
      "step": 149
    },
    {
      "epoch": 0.0594883997620464,
      "grad_norm": 2.791302442550659,
      "learning_rate": 2.634210526315789e-05,
      "loss": 0.4109,
      "step": 150
    },
    {
      "epoch": 0.0594883997620464,
      "eval_loss": 0.23389868438243866,
      "eval_runtime": 47.3967,
      "eval_samples_per_second": 22.407,
      "eval_steps_per_second": 5.612,
      "step": 150
    },
    {
      "epoch": 0.059884989093793375,
      "grad_norm": 4.477522373199463,
      "learning_rate": 2.5815263157894736e-05,
      "loss": 0.5733,
      "step": 151
    },
    {
      "epoch": 0.060281578425540355,
      "grad_norm": 3.334212303161621,
      "learning_rate": 2.528842105263158e-05,
      "loss": 0.3337,
      "step": 152
    },
    {
      "epoch": 0.06067816775728733,
      "grad_norm": 2.710191011428833,
      "learning_rate": 2.4761578947368418e-05,
      "loss": 0.1571,
      "step": 153
    },
    {
      "epoch": 0.06107475708903431,
      "grad_norm": 1.1137654781341553,
      "learning_rate": 2.423473684210526e-05,
      "loss": 0.1697,
      "step": 154
    },
    {
      "epoch": 0.06147134642078128,
      "grad_norm": 0.505480945110321,
      "learning_rate": 2.3707894736842103e-05,
      "loss": 0.0237,
      "step": 155
    },
    {
      "epoch": 0.061867935752528254,
      "grad_norm": 1.122707724571228,
      "learning_rate": 2.3181052631578946e-05,
      "loss": 0.0803,
      "step": 156
    },
    {
      "epoch": 0.062264525084275234,
      "grad_norm": 0.782159149646759,
      "learning_rate": 2.265421052631579e-05,
      "loss": 0.063,
      "step": 157
    },
    {
      "epoch": 0.0626611144160222,
      "grad_norm": 2.235445499420166,
      "learning_rate": 2.212736842105263e-05,
      "loss": 0.2019,
      "step": 158
    },
    {
      "epoch": 0.06305770374776919,
      "grad_norm": 3.1838295459747314,
      "learning_rate": 2.1600526315789474e-05,
      "loss": 0.3564,
      "step": 159
    },
    {
      "epoch": 0.06345429307951617,
      "grad_norm": 1.417310118675232,
      "learning_rate": 2.1073684210526313e-05,
      "loss": 0.1137,
      "step": 160
    },
    {
      "epoch": 0.06385088241126313,
      "grad_norm": 1.3451471328735352,
      "learning_rate": 2.0546842105263155e-05,
      "loss": 0.1321,
      "step": 161
    },
    {
      "epoch": 0.06424747174301011,
      "grad_norm": 3.381690740585327,
      "learning_rate": 2.002e-05,
      "loss": 0.3365,
      "step": 162
    },
    {
      "epoch": 0.06464406107475709,
      "grad_norm": 1.9907605648040771,
      "learning_rate": 1.949315789473684e-05,
      "loss": 0.1682,
      "step": 163
    },
    {
      "epoch": 0.06504065040650407,
      "grad_norm": 1.1349059343338013,
      "learning_rate": 1.8966315789473683e-05,
      "loss": 0.0735,
      "step": 164
    },
    {
      "epoch": 0.06543723973825104,
      "grad_norm": 1.4078447818756104,
      "learning_rate": 1.8439473684210522e-05,
      "loss": 0.1848,
      "step": 165
    },
    {
      "epoch": 0.06583382906999802,
      "grad_norm": 2.635918617248535,
      "learning_rate": 1.791263157894737e-05,
      "loss": 0.288,
      "step": 166
    },
    {
      "epoch": 0.066230418401745,
      "grad_norm": 2.0039432048797607,
      "learning_rate": 1.738578947368421e-05,
      "loss": 0.4102,
      "step": 167
    },
    {
      "epoch": 0.06662700773349196,
      "grad_norm": 2.923654794692993,
      "learning_rate": 1.685894736842105e-05,
      "loss": 0.8078,
      "step": 168
    },
    {
      "epoch": 0.06702359706523894,
      "grad_norm": 2.5879900455474854,
      "learning_rate": 1.6332105263157893e-05,
      "loss": 0.3793,
      "step": 169
    },
    {
      "epoch": 0.06742018639698592,
      "grad_norm": 1.7998006343841553,
      "learning_rate": 1.5805263157894735e-05,
      "loss": 0.1745,
      "step": 170
    },
    {
      "epoch": 0.0678167757287329,
      "grad_norm": 2.5468177795410156,
      "learning_rate": 1.5278421052631578e-05,
      "loss": 0.1744,
      "step": 171
    },
    {
      "epoch": 0.06821336506047987,
      "grad_norm": 2.898380994796753,
      "learning_rate": 1.4751578947368419e-05,
      "loss": 0.379,
      "step": 172
    },
    {
      "epoch": 0.06860995439222685,
      "grad_norm": 3.7562787532806396,
      "learning_rate": 1.4224736842105262e-05,
      "loss": 0.5993,
      "step": 173
    },
    {
      "epoch": 0.06900654372397383,
      "grad_norm": 1.9443773031234741,
      "learning_rate": 1.3697894736842106e-05,
      "loss": 0.1486,
      "step": 174
    },
    {
      "epoch": 0.0694031330557208,
      "grad_norm": 2.690262794494629,
      "learning_rate": 1.3171052631578945e-05,
      "loss": 0.2218,
      "step": 175
    },
    {
      "epoch": 0.06979972238746777,
      "grad_norm": 2.491971492767334,
      "learning_rate": 1.264421052631579e-05,
      "loss": 0.1893,
      "step": 176
    },
    {
      "epoch": 0.07019631171921475,
      "grad_norm": 3.372171401977539,
      "learning_rate": 1.211736842105263e-05,
      "loss": 0.5962,
      "step": 177
    },
    {
      "epoch": 0.07059290105096173,
      "grad_norm": 2.8883213996887207,
      "learning_rate": 1.1590526315789473e-05,
      "loss": 0.3897,
      "step": 178
    },
    {
      "epoch": 0.0709894903827087,
      "grad_norm": 2.1615426540374756,
      "learning_rate": 1.1063684210526316e-05,
      "loss": 0.1384,
      "step": 179
    },
    {
      "epoch": 0.07138607971445568,
      "grad_norm": 3.077579975128174,
      "learning_rate": 1.0536842105263156e-05,
      "loss": 0.6491,
      "step": 180
    },
    {
      "epoch": 0.07178266904620266,
      "grad_norm": 1.228853464126587,
      "learning_rate": 1.001e-05,
      "loss": 0.06,
      "step": 181
    },
    {
      "epoch": 0.07217925837794964,
      "grad_norm": 1.8013077974319458,
      "learning_rate": 9.483157894736842e-06,
      "loss": 0.1734,
      "step": 182
    },
    {
      "epoch": 0.0725758477096966,
      "grad_norm": 0.8816391229629517,
      "learning_rate": 8.956315789473684e-06,
      "loss": 0.0388,
      "step": 183
    },
    {
      "epoch": 0.07297243704144359,
      "grad_norm": 1.7668546438217163,
      "learning_rate": 8.429473684210525e-06,
      "loss": 0.217,
      "step": 184
    },
    {
      "epoch": 0.07336902637319057,
      "grad_norm": 3.179753541946411,
      "learning_rate": 7.902631578947368e-06,
      "loss": 0.28,
      "step": 185
    },
    {
      "epoch": 0.07376561570493753,
      "grad_norm": 3.8374545574188232,
      "learning_rate": 7.3757894736842095e-06,
      "loss": 0.4457,
      "step": 186
    },
    {
      "epoch": 0.07416220503668451,
      "grad_norm": 3.593108892440796,
      "learning_rate": 6.848947368421053e-06,
      "loss": 1.1263,
      "step": 187
    },
    {
      "epoch": 0.07455879436843149,
      "grad_norm": 3.4441983699798584,
      "learning_rate": 6.322105263157895e-06,
      "loss": 0.4889,
      "step": 188
    },
    {
      "epoch": 0.07495538370017847,
      "grad_norm": 2.782144546508789,
      "learning_rate": 5.7952631578947365e-06,
      "loss": 0.3363,
      "step": 189
    },
    {
      "epoch": 0.07535197303192544,
      "grad_norm": 0.27459532022476196,
      "learning_rate": 5.268421052631578e-06,
      "loss": 0.0108,
      "step": 190
    },
    {
      "epoch": 0.07574856236367242,
      "grad_norm": 2.1841888427734375,
      "learning_rate": 4.741578947368421e-06,
      "loss": 0.3438,
      "step": 191
    },
    {
      "epoch": 0.0761451516954194,
      "grad_norm": 0.02204926311969757,
      "learning_rate": 4.2147368421052626e-06,
      "loss": 0.0007,
      "step": 192
    },
    {
      "epoch": 0.07654174102716636,
      "grad_norm": 1.9425251483917236,
      "learning_rate": 3.6878947368421047e-06,
      "loss": 0.2348,
      "step": 193
    },
    {
      "epoch": 0.07693833035891334,
      "grad_norm": 0.04739471897482872,
      "learning_rate": 3.1610526315789474e-06,
      "loss": 0.0012,
      "step": 194
    },
    {
      "epoch": 0.07733491969066032,
      "grad_norm": 0.06455172598361969,
      "learning_rate": 2.634210526315789e-06,
      "loss": 0.0017,
      "step": 195
    },
    {
      "epoch": 0.0777315090224073,
      "grad_norm": 3.44230318069458,
      "learning_rate": 2.1073684210526313e-06,
      "loss": 0.4222,
      "step": 196
    },
    {
      "epoch": 0.07812809835415427,
      "grad_norm": 2.755849838256836,
      "learning_rate": 1.5805263157894737e-06,
      "loss": 0.4239,
      "step": 197
    },
    {
      "epoch": 0.07852468768590125,
      "grad_norm": 1.2250853776931763,
      "learning_rate": 1.0536842105263156e-06,
      "loss": 0.0493,
      "step": 198
    },
    {
      "epoch": 0.07892127701764823,
      "grad_norm": 2.685635566711426,
      "learning_rate": 5.268421052631578e-07,
      "loss": 0.3463,
      "step": 199
    },
    {
      "epoch": 0.07931786634939521,
      "grad_norm": 4.368965148925781,
      "learning_rate": 0.0,
      "loss": 0.6687,
      "step": 200
    },
    {
      "epoch": 0.07931786634939521,
      "eval_loss": 0.20430698990821838,
      "eval_runtime": 47.6165,
      "eval_samples_per_second": 22.303,
      "eval_steps_per_second": 5.586,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.39474174476288e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}