Language models contain interpretable computational structure encoded in their parameter weights, not irreducibly impenetrable complexity

{
  "doc": "paper.md",
  "context": "Core empirical hypothesis of the paper, supported by successful VPD decomposition yielding ~10,000 interpretable subcomponents across 24 weight matrices.",
  "category": "ai",
  "norm_label": "Language models contain interpretable computational structure encoded in their parameter weights, not irreducibly impenetrable complexity",
  "graphify_id": "parameter_structure_hypothesis",
  "source_file": "paper.md",
  "imported_from": "/Users/antonborzov/Documents/Research.nosync/papers/extract_typed_out/paper/graph.json",
  "extracted_type": "hypothesis",
  "source_location": "§3",
  "graphify_file_type": "hypothesis"
}