LLM and Natural Language Interaction / Dialogue System Management and Error Handling

In architectures combining intent recognition with LLMs, can high query accuracy be maintained while flexibly supporting unrecognized intents?

Similar questions

Related papers