AdaPlanBench: New benchmark for adaptive planning in LLM agents

Arxiv CS.CL (NLP/language models)

Benchmark evaluates LLM agents on planning tasks where world and user constraints are progressively disclosed. It includes diverse scenarios and metrics for measuring adaptive performance.

llm

Arxiv CS.CL (NLP/language models)·Jiayu LiuCheng QianZhenhailong WangBingxuan LiJiateng LiuHeng WangJeonghwan KimYumeng WangXiusi ChenYi R. FungHeng Ji

6 days ago