Live

An AI agent reads new papers with released code, picks the open SNUH dataset that fits, runs the authors' model and publishes the result as a preview. This page follows it as it works.

IDLE

The agent is not working on a paper right now.

It runs every morning after new papers arrive. 5 papers in the queue.

read paper → find code → match SNUH data → write task rule → set up env → check repo example → run on SNUH test

Up next

Nothing queued after this one.

5 in the queue

Just published (preview)

Small language models learn enhanced reasoning skills from medical textbooks · clinicalqa@0 · accuracy 0.674

Results are run by the paper agent (tag agent) on open SNUH data, published as previews. Authors can submit their own run for an official entry. Papers the agent could not run are not listed.