Blame for benchmark/Belief/evaluation/validation_report.json

{
  "total_tied_samples": 20319,
  "by_split": {
    "test": 2949,
    "train": 14217,
    "validation": 3153
  },
  "outcome_sources": {
    "released_trade_source": 20066,
    "polymarket_clob_api": 113,
    "explicit_binary_complement": 105,
    "released_belief_observed_stances": 35
  },
  "checks": {
    "unique_sample_ids": true,
    "all_confidence_exact_zero": true,
    "all_have_two_valid_choices": true,
    "v1_label_is_valid": true
  }
}