Do LLMs leverage architectural capacity for introspection on internal computations and prior token generation?

{
  "doc": "janus-information-flow-transformers-2025.md",
  "author": null,
  "context": "Central empirical question separating architectural possibility from actual model behavior; gates introspection research.",
  "enrichment": {
    "is_stale": true
  },
  "norm_label": "Do LLMs leverage architectural capacity for introspection on internal computations and prior token generation?",
  "source_url": null,
  "graphify_id": "can_llms_introspect",
  "source_file": "janus-information-flow-transformers-2025.md",
  "imported_from": "/Users/antonborzov/Documents/Research.nosync/papers/extract_typed_out/janus-information-flow-transformers-2025/graph.json",
  "extracted_type": "question",
  "graphify_file_type": "question"
}