eval_result.json

{
  "model": "runs/large_v4",
  "threshold": 0.5,
  "hatecheck_recall": 0.9462,
  "hatecheck_fp": 0.018,
  "dialect_fp_high_aae": 0.0853,
  "dialect_fp_low_aae": 0.036,
  "dialect_fp_gap": 0.0493,
  "n_high": 1500,
  "n_low": 1500,
  "GATE_PASSED": true
}