What We Learned by Reproducing 2,200 papers from ICML
Content body was empty; no extractable signal beyond the title.
The content field was blank, so no substantive claims, findings, or quotes could be extracted. The title suggests a large-scale ML reproducibility study covering 2,200 ICML papers by Hugging Face, which would normally score high, but the body must be re-fetched before scoring meaningfully.