ToM2Vec Gives Agents Theory of Mind via Continuous Thought Vectors, Generalizing to Unseen Partners
EliasEskin · x · 2026-10-23
Researchers introduce ToM2Vec, which lets LLM-based agents build theory of mind about partners from interaction history to guide coordination.
Key points:
- Instead of holding many discrete hypotheses, partners' reasoning is represented as continuous thought vectors capturing multiple possible intentions at once
- These ToM tokens then steer the agent's own reasoning
- Improved coordination across 5 multi-agent tasks, transferring to unseen partners — including frontier-model partners
Motivation: LLM agents often ignore their partner's next steps, causing redundant or conflicting actions.
More from coding & agent
- Testing Hermes Agent with n8n: Letting Agents Execute Real Automation Workflows — techlatest_net · 2026-10-23
- OpenAI: Keep MCP Server Instructions Under ~1,000 Characters or ChatGPT May Truncate Them — bibryam · 2026-10-23
- Only ~56% of MCP Servers Provide Instructions, OpenAI Data Shows — bibryam · 2026-10-23
- OpenAI: Don't Hide MCP Tools Behind listCommand/executeCommand Dispatchers — bibryam · 2026-10-23
- Resend logs 1M monthly MCP calls as OpenAI frames MCP as a distribution channel — bibryam · 2026-10-23
- MCP Elicitation lands in Claude: servers can ask users questions mid-action — bibryam · 2026-10-23