deepxiv-baseline-table

To compile a baseline table from arXiv papers, provide a topic string and optional date range to automatically search, extract paper metadata, and output a Markdown table comparing titles, datasets, benchmarks, scores, and code availability.

Yes, you can compare open-source code status and datasets across multiple papers. The generated Markdown baseline table includes dedicated columns for open-source status, code URLs, datasets, benchmarks, and scores for side-by-side comparison.

The best way to generate a structured comparison of benchmarks and methods is to run a topic-driven search that extracts metadata from relevant papers and organizes datasets, scores, and code availability into a ready-to-use Markdown table.

No, you do not need to manually collect paper metadata. The workflow orchestrates deepxiv search, brief, head, and experiment-section reads to automatically extract titles, arXiv IDs, URLs, and benchmark scores into a structured Markdown table.

Topic search and dataset analysis works by taking a broad topic inquiry, performing a deepxiv search within optional date ranges, reading experiment sections, and organizing the extracted paper metadata into a comparison-ready baseline table.

When mapping a broad topic to a side-by-side baseline comparison, the accuracy depends on the availability of paper metadata and experiment sections, meaning highly niche topics with limited arXiv publications may yield incomplete benchmark or code status data.