Search papers, labs, and topics across Lattice.
This paper introduces PaperCompiler, a framework designed to translate research papers into repository-level code implementations by compiling explicit specifications from paper-grounded evidence. It addresses the common pitfalls of existing paper-to-code agents, which often produce vague outputs that lead to inconsistent implementations and loss of methodological fidelity. The results demonstrate a significant improvement in reference-based fidelity and a reduction in high-severity critiques, showcasing PaperCompiler's effectiveness in preserving the integrity of the original research while allowing for flexible engineering choices.
Achieving a 13.8% boost in fidelity, PaperCompiler transforms how research papers are converted into reliable code implementations, tackling the pitfalls of existing methods.
Faithfully translating research papers into repository-level implementations remains challenging because papers often describe methods at a high level, leave implementation assumptions implicit, and require generated repositories to preserve method logic, evaluation protocols, and cross-file consistency. Despite recent advances in paper-to-code agents, their intermediate outputs are often presented as free-form plans or summaries that downstream coding agents may ignore, reinterpret, or compress, leading to algorithmic simplification and inconsistent repository structure. To address these challenges, we introduce PaperCompiler, a paper-to-code generation framework that compiles paper-grounded evidence into explicit repository-level implementation specifications. PaperCompiler grounds implementation-relevant evidence while preserving source provenance and distinguishing paper-supported, inferred, externally delegated, and unresolved information. The resulting specifications encode non-degradation requirements, ownership assignments, cross-file dependencies, and file-level constraints. Repository generation proceeds under these compiled specifications while retaining flexibility over local engineering choices not fixed by the paper. PaperCompiler outperforms strong baselines on Paper2CodeBench, achieving a 13.8% relative improvement in reference-based fidelity (from 3.64 to 4.15) and reducing high-severity evaluator critiques (from 13.2% to 6.1%).
