Tomesphere ends arxiv context-switching with inline TLDRs, reviews, and citation graphs
3 million arxiv papers with LLM-curated summaries and one-click GitHub links.
A Reddit user spent months building Tomesphere to eliminate the frustration of context-switching between arxiv, OpenReview, GitHub, and HuggingFace when researching papers. The tool provides a unified page for each of 3M arxiv papers, featuring an LLM-curated TLDR and key findings, public OpenReview reviews from venues like NeurIPS, ICLR, ICML, TMLR, and COLM, plus linked GitHub repositories and HuggingFace models. It also includes a citation graph showing both incoming and outgoing references, and a SPECTER2-based semantic neighbor graph to discover related papers by topic similarity.
Accessible via a Chrome extension using the MV3 side panel API, the same context renders inline on any arxiv page, or users can browse directly at tomesphere.com. The tool is currently free and requires no signup. Caveats: blind-review venues like CVPR, AAAI, and ECCV are excluded until community contributions fill them in, and GitHub/HuggingFace matches are best-effort. The creator is actively seeking feedback on missing features and which papers users check first.
- SPECTER2 semantic neighbor graph helps discover related papers by topic similarity, not just citation links.
- Covers 3M arxiv papers with inline LLM-curated TLDRs and multi-source context (OpenReview, GitHub, HuggingFace).
- Free Chrome extension with MV3 side panel renders context directly on arxiv pages—no signup needed.
Why It Matters
Saves AI researchers hours per week by centralizing paper context, reviews, and repositories into one interface.