outputs/claim2.json

{
  "claim": 2,
  "literal_claim": "On ColorMNIST+, PGCM matches CBM performance with concept accuracy 99.2±0.0 vs 99.2±0.1 and task accuracy 99.7±0.0 vs 99.6±0.1, while on CelebA CBM's task accuracy (84.0±0.3) modestly exceeds PGCM's (83.0±0.0) (Table 4).",
  "native_result": {
    "cbm": {
      "concept_accuracy": 0.9913396835327148,
      "task_accuracy": 0.9959578947368422
    },
    "cbm_permuted_label_control": {
      "concept_accuracy": 0.5023798942565918,
      "task_accuracy": 0.9023157894736842
    },
    "pgcm": {
      "concept_accuracy": 0.9880701899528503,
      "task_accuracy": 0.9938
    },
    "pgcm_permuted_label_control": {
      "concept_accuracy": 0.502457320690155,
      "task_accuracy": 0.9023263157894736
    },
    "primary_table_values_present": true
  }
}