概念导读
ROME and MEMIT are model-editing methods for changing factual knowledge stored in model weights.
- Related
ROME / MEMIT
ROME and MEMIT are model-editing methods for changing factual knowledge stored in model weights.
They are useful contrast points for agent-memory systems: model editing changes parameters, while external memory systems update retrievable state.
Related
- memoryllm — Memory-aware LLM updates.
- contextual-agentic-memory — Contrasts external memory with weight editing.
- knowledge — Knowledge representation and update concerns.