{
  "best_global_step": 6000,
  "best_metric": 0.8207234712908512,
  "best_model_checkpoint": "./results/run-2/checkpoint-6000",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 7000,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "grad_norm": 3.6001648902893066,
      "learning_rate": 3.1887409174919234e-06,
      "loss": 0.0412,
      "step": 100
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6239789128303528,
      "learning_rate": 6.409691339200936e-06,
      "loss": 0.1091,
      "step": 200
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6479267477989197,
      "learning_rate": 9.63064176090995e-06,
      "loss": 0.0,
      "step": 300
    },
    {
      "epoch": 0.4,
      "grad_norm": 20.527082443237305,
      "learning_rate": 1.054959200884952e-05,
      "loss": 0.0,
      "step": 400
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8853731751441956,
      "learning_rate": 1.0481970753745042e-05,
      "loss": 0.0,
      "step": 500
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.839735507965088,
      "learning_rate": 1.0414349498640566e-05,
      "loss": 0.0,
      "step": 600
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.7137391567230225,
      "learning_rate": 1.0346728243536088e-05,
      "loss": 0.0284,
      "step": 700
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6026535630226135,
      "learning_rate": 1.027910698843161e-05,
      "loss": 0.1612,
      "step": 800
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5444966554641724,
      "learning_rate": 1.0211485733327134e-05,
      "loss": 0.0,
      "step": 900
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2413469552993774,
      "learning_rate": 1.0143864478222655e-05,
      "loss": 0.0986,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "eval_economic_inequality_accuracy": 0.8047858942065491,
      "eval_economic_inequality_f1": 0.7538941327354425,
      "eval_economic_policy_benefits_accuracy": 0.8282828282828283,
      "eval_economic_policy_benefits_f1": 0.764129475069552,
      "eval_ethnic_boundaries_accuracy": 0.9493464052287581,
      "eval_ethnic_boundaries_f1": 0.9281293325410972,
      "eval_language_policy_accuracy": 0.7058823529411765,
      "eval_language_policy_f1": 0.6071428571428571,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.8888888888888888,
      "eval_mother_tongue_education_f1": 0.837037037037037,
      "eval_overall_accuracy": 0.8390088178776338,
      "eval_overall_f1": 0.7839408748610676,
      "eval_religion_ethnic_identity_accuracy": 0.8568665377176016,
      "eval_religion_ethnic_identity_f1": 0.8133124146404199,
      "eval_runtime": 4.4593,
      "eval_samples_per_second": 897.004,
      "eval_steps_per_second": 56.063,
      "step": 1000
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.5792437195777893,
      "learning_rate": 1.0076243223118178e-05,
      "loss": 0.0,
      "step": 1100
    },
    {
      "epoch": 1.2,
      "grad_norm": 8.802961349487305,
      "learning_rate": 1.0008621968013701e-05,
      "loss": 0.0297,
      "step": 1200
    },
    {
      "epoch": 1.3,
      "grad_norm": 3.981771469116211,
      "learning_rate": 9.941000712909224e-06,
      "loss": 0.0,
      "step": 1300
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9215607047080994,
      "learning_rate": 9.873379457804745e-06,
      "loss": 0.263,
      "step": 1400
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5474179983139038,
      "learning_rate": 9.805758202700268e-06,
      "loss": 0.0,
      "step": 1500
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1813520193099976,
      "learning_rate": 9.738136947595791e-06,
      "loss": 0.0,
      "step": 1600
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.511495113372803,
      "learning_rate": 9.670515692491313e-06,
      "loss": 0.0037,
      "step": 1700
    },
    {
      "epoch": 1.8,
      "grad_norm": 3.1629152297973633,
      "learning_rate": 9.602894437386836e-06,
      "loss": 0.0027,
      "step": 1800
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.6996293663978577,
      "learning_rate": 9.535273182282359e-06,
      "loss": 19.1531,
      "step": 1900
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3346213102340698,
      "learning_rate": 9.467651927177882e-06,
      "loss": 2.9418,
      "step": 2000
    },
    {
      "epoch": 2.0,
      "eval_economic_inequality_accuracy": 0.7896725440806045,
      "eval_economic_inequality_f1": 0.7951291586358689,
      "eval_economic_policy_benefits_accuracy": 0.8148148148148148,
      "eval_economic_policy_benefits_f1": 0.8131577832835695,
      "eval_ethnic_boundaries_accuracy": 0.9264705882352942,
      "eval_ethnic_boundaries_f1": 0.9324831159457648,
      "eval_language_policy_accuracy": 0.7058823529411765,
      "eval_language_policy_f1": 0.6071428571428571,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.7777777777777778,
      "eval_mother_tongue_education_f1": 0.7407407407407408,
      "eval_overall_accuracy": 0.8103022392805538,
      "eval_overall_f1": 0.7884819842273197,
      "eval_religion_ethnic_identity_accuracy": 0.8471953578336557,
      "eval_religion_ethnic_identity_f1": 0.8422382496151178,
      "eval_runtime": 4.499,
      "eval_samples_per_second": 889.087,
      "eval_steps_per_second": 55.568,
      "step": 2000
    },
    {
      "epoch": 2.1,
      "grad_norm": 133.0632781982422,
      "learning_rate": 9.400030672073403e-06,
      "loss": 0.0,
      "step": 2100
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1462996006011963,
      "learning_rate": 9.332409416968926e-06,
      "loss": 0.0,
      "step": 2200
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.11461023986339569,
      "learning_rate": 9.26478816186445e-06,
      "loss": 0.0,
      "step": 2300
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.37900322675704956,
      "learning_rate": 9.197166906759972e-06,
      "loss": 0.0046,
      "step": 2400
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.0005897283554077,
      "learning_rate": 9.129545651655495e-06,
      "loss": 0.0,
      "step": 2500
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.15003152191638947,
      "learning_rate": 9.061924396551017e-06,
      "loss": 0.0,
      "step": 2600
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9842479228973389,
      "learning_rate": 8.99430314144654e-06,
      "loss": 0.0017,
      "step": 2700
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.747731685638428,
      "learning_rate": 8.926681886342063e-06,
      "loss": 0.0,
      "step": 2800
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.4671611189842224,
      "learning_rate": 8.859060631237586e-06,
      "loss": 0.0,
      "step": 2900
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9774578213691711,
      "learning_rate": 8.791439376133107e-06,
      "loss": 0.0,
      "step": 3000
    },
    {
      "epoch": 3.0,
      "eval_economic_inequality_accuracy": 0.8173803526448362,
      "eval_economic_inequality_f1": 0.8105151543481738,
      "eval_economic_policy_benefits_accuracy": 0.8686868686868687,
      "eval_economic_policy_benefits_f1": 0.8544859369266805,
      "eval_ethnic_boundaries_accuracy": 0.9330065359477124,
      "eval_ethnic_boundaries_f1": 0.9342620574580529,
      "eval_language_policy_accuracy": 0.7058823529411765,
      "eval_language_policy_f1": 0.6071428571428571,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.7777777777777778,
      "eval_mother_tongue_education_f1": 0.7407407407407408,
      "eval_overall_accuracy": 0.8266000709526621,
      "eval_overall_f1": 0.8006341340755991,
      "eval_religion_ethnic_identity_accuracy": 0.8568665377176016,
      "eval_religion_ethnic_identity_f1": 0.85665805783709,
      "eval_runtime": 4.4861,
      "eval_samples_per_second": 891.646,
      "eval_steps_per_second": 55.728,
      "step": 3000
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.6768048405647278,
      "learning_rate": 8.72381812102863e-06,
      "loss": 0.0002,
      "step": 3100
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.13458865880966187,
      "learning_rate": 8.656196865924153e-06,
      "loss": 0.0006,
      "step": 3200
    },
    {
      "epoch": 3.3,
      "grad_norm": 2.0157546997070312,
      "learning_rate": 8.588575610819674e-06,
      "loss": 0.0,
      "step": 3300
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.07812555134296417,
      "learning_rate": 8.520954355715199e-06,
      "loss": 0.0,
      "step": 3400
    },
    {
      "epoch": 3.5,
      "grad_norm": 149.58753967285156,
      "learning_rate": 8.45333310061072e-06,
      "loss": 0.0032,
      "step": 3500
    },
    {
      "epoch": 3.6,
      "grad_norm": 3.429203748703003,
      "learning_rate": 8.385711845506243e-06,
      "loss": 0.0,
      "step": 3600
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.6850093603134155,
      "learning_rate": 8.318090590401766e-06,
      "loss": 0.0003,
      "step": 3700
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.06944789737462997,
      "learning_rate": 8.250469335297288e-06,
      "loss": 0.0,
      "step": 3800
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.11943838745355606,
      "learning_rate": 8.18284808019281e-06,
      "loss": 0.0,
      "step": 3900
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.308050274848938,
      "learning_rate": 8.115226825088332e-06,
      "loss": 0.0,
      "step": 4000
    },
    {
      "epoch": 4.0,
      "eval_economic_inequality_accuracy": 0.8085642317380353,
      "eval_economic_inequality_f1": 0.810531309630811,
      "eval_economic_policy_benefits_accuracy": 0.8653198653198653,
      "eval_economic_policy_benefits_f1": 0.8539054809077693,
      "eval_ethnic_boundaries_accuracy": 0.9362745098039216,
      "eval_ethnic_boundaries_f1": 0.9400162685161402,
      "eval_language_policy_accuracy": 0.6764705882352942,
      "eval_language_policy_f1": 0.6150075414781296,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.7777777777777778,
      "eval_mother_tongue_education_f1": 0.7735042735042735,
      "eval_overall_accuracy": 0.8169885251374339,
      "eval_overall_f1": 0.8070145214695552,
      "eval_religion_ethnic_identity_accuracy": 0.8375241779497099,
      "eval_religion_ethnic_identity_f1": 0.8491222547802076,
      "eval_runtime": 4.4658,
      "eval_samples_per_second": 895.686,
      "eval_steps_per_second": 55.98,
      "step": 4000
    },
    {
      "epoch": 4.1,
      "grad_norm": 0.2866123616695404,
      "learning_rate": 8.047605569983857e-06,
      "loss": 0.0,
      "step": 4100
    },
    {
      "epoch": 4.2,
      "grad_norm": 0.26467469334602356,
      "learning_rate": 7.979984314879378e-06,
      "loss": 0.0,
      "step": 4200
    },
    {
      "epoch": 4.3,
      "grad_norm": 0.06363595277070999,
      "learning_rate": 7.912363059774901e-06,
      "loss": 0.0,
      "step": 4300
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.2771783173084259,
      "learning_rate": 7.844741804670424e-06,
      "loss": 0.0002,
      "step": 4400
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.022821573540568352,
      "learning_rate": 7.777120549565947e-06,
      "loss": 0.0006,
      "step": 4500
    },
    {
      "epoch": 4.6,
      "grad_norm": 0.7363903522491455,
      "learning_rate": 7.709499294461468e-06,
      "loss": 0.0,
      "step": 4600
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.3803585767745972,
      "learning_rate": 7.641878039356991e-06,
      "loss": 0.0,
      "step": 4700
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.14331550896167755,
      "learning_rate": 7.5742567842525145e-06,
      "loss": 0.0019,
      "step": 4800
    },
    {
      "epoch": 4.9,
      "grad_norm": 0.06504154205322266,
      "learning_rate": 7.506635529148037e-06,
      "loss": 0.0,
      "step": 4900
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.09633997082710266,
      "learning_rate": 7.43901427404356e-06,
      "loss": 0.0,
      "step": 5000
    },
    {
      "epoch": 5.0,
      "eval_economic_inequality_accuracy": 0.8249370277078085,
      "eval_economic_inequality_f1": 0.823342228819045,
      "eval_economic_policy_benefits_accuracy": 0.8451178451178452,
      "eval_economic_policy_benefits_f1": 0.8327902682741393,
      "eval_ethnic_boundaries_accuracy": 0.9395424836601307,
      "eval_ethnic_boundaries_f1": 0.9429587969409783,
      "eval_language_policy_accuracy": 0.6470588235294118,
      "eval_language_policy_f1": 0.5939135835329606,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.7777777777777778,
      "eval_mother_tongue_education_f1": 0.7735042735042735,
      "eval_overall_accuracy": 0.8165062399029553,
      "eval_overall_f1": 0.8063046374512969,
      "eval_religion_ethnic_identity_accuracy": 0.8646034816247582,
      "eval_religion_ethnic_identity_f1": 0.8713186736363845,
      "eval_runtime": 4.4849,
      "eval_samples_per_second": 891.887,
      "eval_steps_per_second": 55.743,
      "step": 5000
    },
    {
      "epoch": 5.1,
      "grad_norm": 2.563802480697632,
      "learning_rate": 7.371393018939082e-06,
      "loss": 0.0049,
      "step": 5100
    },
    {
      "epoch": 5.2,
      "grad_norm": 10.922028541564941,
      "learning_rate": 7.303771763834604e-06,
      "loss": 0.0004,
      "step": 5200
    },
    {
      "epoch": 5.3,
      "grad_norm": 0.9929112792015076,
      "learning_rate": 7.236150508730128e-06,
      "loss": 0.0827,
      "step": 5300
    },
    {
      "epoch": 5.4,
      "grad_norm": 0.13412432372570038,
      "learning_rate": 7.16852925362565e-06,
      "loss": 0.0,
      "step": 5400
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.0726282000541687,
      "learning_rate": 7.100907998521172e-06,
      "loss": 0.0,
      "step": 5500
    },
    {
      "epoch": 5.6,
      "grad_norm": 12.0023832321167,
      "learning_rate": 7.033286743416695e-06,
      "loss": 0.0,
      "step": 5600
    },
    {
      "epoch": 5.7,
      "grad_norm": 1.0574413537979126,
      "learning_rate": 6.9656654883122174e-06,
      "loss": 0.0,
      "step": 5700
    },
    {
      "epoch": 5.8,
      "grad_norm": 0.3997838795185089,
      "learning_rate": 6.89804423320774e-06,
      "loss": 0.0,
      "step": 5800
    },
    {
      "epoch": 5.9,
      "grad_norm": 2.0076065063476562,
      "learning_rate": 6.830422978103262e-06,
      "loss": 0.0,
      "step": 5900
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.676760673522949,
      "learning_rate": 6.762801722998786e-06,
      "loss": 0.0,
      "step": 6000
    },
    {
      "epoch": 6.0,
      "eval_economic_inequality_accuracy": 0.8350125944584383,
      "eval_economic_inequality_f1": 0.8243306188572417,
      "eval_economic_policy_benefits_accuracy": 0.8619528619528619,
      "eval_economic_policy_benefits_f1": 0.840430611429002,
      "eval_ethnic_boundaries_accuracy": 0.9477124183006536,
      "eval_ethnic_boundaries_f1": 0.9385901883316666,
      "eval_language_policy_accuracy": 0.7058823529411765,
      "eval_language_policy_f1": 0.6172459893048128,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.8888888888888888,
      "eval_mother_tongue_education_f1": 0.837037037037037,
      "eval_overall_accuracy": 0.8529320416673837,
      "eval_overall_f1": 0.8207234712908512,
      "eval_religion_ethnic_identity_accuracy": 0.8781431334622823,
      "eval_religion_ethnic_identity_f1": 0.8667063827853473,
      "eval_runtime": 4.4702,
      "eval_samples_per_second": 894.82,
      "eval_steps_per_second": 55.926,
      "step": 6000
    },
    {
      "epoch": 6.1,
      "grad_norm": 0.18037013709545135,
      "learning_rate": 6.695180467894308e-06,
      "loss": 0.0,
      "step": 6100
    },
    {
      "epoch": 6.2,
      "grad_norm": 22.60749626159668,
      "learning_rate": 6.62755921278983e-06,
      "loss": 0.0,
      "step": 6200
    },
    {
      "epoch": 6.3,
      "grad_norm": 0.8915765881538391,
      "learning_rate": 6.559937957685353e-06,
      "loss": 0.0,
      "step": 6300
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.5682930946350098,
      "learning_rate": 6.492316702580875e-06,
      "loss": 0.0,
      "step": 6400
    },
    {
      "epoch": 6.5,
      "grad_norm": 0.08639243990182877,
      "learning_rate": 6.424695447476398e-06,
      "loss": 0.0001,
      "step": 6500
    },
    {
      "epoch": 6.6,
      "grad_norm": 0.060448452830314636,
      "learning_rate": 6.357074192371921e-06,
      "loss": 0.3897,
      "step": 6600
    },
    {
      "epoch": 6.7,
      "grad_norm": 0.7599291801452637,
      "learning_rate": 6.289452937267443e-06,
      "loss": 0.0,
      "step": 6700
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.19008946418762207,
      "learning_rate": 6.2218316821629656e-06,
      "loss": 0.0012,
      "step": 6800
    },
    {
      "epoch": 6.9,
      "grad_norm": 0.2047979235649109,
      "learning_rate": 6.1542104270584894e-06,
      "loss": 0.0,
      "step": 6900
    },
    {
      "epoch": 7.0,
      "grad_norm": 4.452627658843994,
      "learning_rate": 6.086589171954012e-06,
      "loss": 0.0,
      "step": 7000
    },
    {
      "epoch": 7.0,
      "eval_economic_inequality_accuracy": 0.8047858942065491,
      "eval_economic_inequality_f1": 0.8098052998323155,
      "eval_economic_policy_benefits_accuracy": 0.8686868686868687,
      "eval_economic_policy_benefits_f1": 0.8627489757007829,
      "eval_ethnic_boundaries_accuracy": 0.9313725490196079,
      "eval_ethnic_boundaries_f1": 0.9331205006157053,
      "eval_language_policy_accuracy": 0.6470588235294118,
      "eval_language_policy_f1": 0.6161764705882353,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.7777777777777778,
      "eval_mother_tongue_education_f1": 0.7735042735042735,
      "eval_overall_accuracy": 0.8137799964973732,
      "eval_overall_f1": 0.8086761873509373,
      "eval_religion_ethnic_identity_accuracy": 0.8529980657640233,
      "eval_religion_ethnic_identity_f1": 0.8567016038643116,
      "eval_runtime": 4.489,
      "eval_samples_per_second": 891.066,
      "eval_steps_per_second": 55.692,
      "step": 7000
    }
  ],
  "logging_steps": 100,
  "max_steps": 16000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 16,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 2,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.9472671563776e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": {
    "gradient_accumulation_steps": 2,
    "learning_rate": 1.0596926887422654e-05,
    "num_train_epochs": 16,
    "per_device_train_batch_size": 8,
    "warmup_steps": 329,
    "weight_decay": 0.059185379025811405
  }
}