{
  "results": {
    "race": {
      "acc": 0.3320574162679426,
      "acc_stderr": 0.014575582129545914
    },
    "drop": {
      "em": 0.006921140939597316,
      "em_stderr": 0.0008490247804930383,
      "f1": 0.06890939597315439,
      "f1_stderr": 0.001599412572762237
    },
    "coqa": {
      "f1": 0.53547064794564,
      "f1_stderr": 0.018874554692662703,
      "em": 0.4089999999999999,
      "em_stderr": 0.020308360980593812
    }
  },
  "versions": {
    "race": 1,
    "drop": 1,
    "coqa": 1
  },
  "config": {
    "model": "hf-causal-experimental",
    "model_args": "pretrained=bigscience/bloom-1b7,use_accelerate=True",
    "num_fewshot": 0,
    "batch_size": "auto",
    "device": "cuda:0",
    "no_cache": true,
    "limit": null,
    "bootstrap_iters": 100000,
    "description_dict": {}
  }
}