Profiles LLM reasoning by turning chain-of-thought into thought graphs, with no generative calls. 22 structural metrics. arXiv:2606.29067
benchmark graph-analysis interpretability llm llms chain-of-thought chain-of-thought-reasoning llm-evaluation llm-evaluation-toolkit llm-evaluation-framework llm-evaluation-metrics llm-evaluation-benchmark llm-evaluations thought-graphs
-
Updated
Oct 6, 2026 - Python