Sparse Attention Mismatch Degrades Multi-Hop Reasoning in Repository-Level Code Understanding

Abstract

Abstract: This report synthesises findings from 12 peer-reviewed papers addressing the following research question: Does training-inference mismatch in sparse attention mechanisms cause measurable degradation in multi-hop reasoning accuracy for repository-level code understanding. 8 claims were extracted from source literature; 8 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 9.0/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research Question

Does training-inference mismatch in sparse attention mechanisms cause measurable degradation in multi-hop reasoning accuracy for repository-level code understanding?

Verification Level

Descriptive public verification status only; aggregate claim counts are public, but individual claim records are not exposed here.

Quality Tier

Descriptive public triage only; this tier does not alter current publication or DOI behavior.

Quality Dimensions

Automated triage signals derived from public fields; not human peer review or independent validation.

Correction Record

Public corrections are additive records. Current status does not claim the synthesis is error-free.

Provenance