651 research papers on mechanistic interpretability, including circuits, superposition, and feature visualization. The collection is continuously updated and deduplicated by FineSet from arXiv and Semantic Scholar. Records include a citation-normalized quality score ranging from 0 to 1.
Use Cases
- Literature review on mechanistic interpretability based on the assembled paper collection.
- Tracking research trends in circuits and feature visualization based on the paper metadata.
- Identifying highly cited papers for foundational reading based on the provided quality scoring.
Strengths
- 651 records provide a substantial corpus for analysis.
- Papers are deduplicated and cross-referenced from arXiv and Semantic Scholar, suggesting reduced redundancy.
- Includes a citation-normalized quality score (p50=0.119, p90=0.358) for filtering.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Freshness should be verified as the last update was 2026-06-12.
Provenance
- Source
- arXiv and Semantic Scholar.
- Collection Method
- Assembled and deduplicated by FineSet.
- Time Range
- 2022–2026
- Freshness
- Continuously updated; last recorded update was 2026-06-12 00:44:03.