In an ablation across five LLMs and two Rocq projects, informed prompts with dependencies and in-file context produced the highest proof success (up to 52% of hs-to-coq theorems), and success fell sharply without context.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.PL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Case Study on the Effectiveness of LLMs in Verification with Proof Assistants
In an ablation across five LLMs and two Rocq projects, informed prompts with dependencies and in-file context produced the highest proof success (up to 52% of hs-to-coq theorems), and success fell sharply without context.