February 21, 2025 front

ML Newsnew | past | best | rss | submit

Stories from February 21, 2025 (UTC)

Go back a day, month, or year. Go forward a day.

1.	Some critical issues with the SWE-bench dataset (arxiv.org)
	350 points by joshwa on Feb 21, 2025 \| hide \| past \| pdf \| 116 comments
2.	Performance of Zero-Shot Time Series Foundation Models on Cloud Data (arxiv.org)
	4 points by wanderingmind on Feb 21, 2025 \| hide \| past \| pdf \| discuss
3.	NoLiMa: Long-Context Evaluation Beyond Literal Matching (arxiv.org)
	2 points by azhenley on Feb 21, 2025 \| hide \| past \| pdf \| discuss
4.	Soup-of-Experts: Pretraining Specialist Models via Parameters Averaging (arxiv.org)
	2 points by PaulHoule on Feb 21, 2025 \| hide \| past \| pdf \| discuss
5.	The Case for Cognitive-Dissonance-Aware Knowledge Updates in LLMs (arxiv.org)
	2 points by PaulHoule on Feb 21, 2025 \| hide \| past \| pdf \| discuss
6.	Fire-Flyer AI-HPC: Cost-Effective Software-Hardware Co-Design for Deep Learning (arxiv.org)
	2 points by doener on Feb 21, 2025 \| hide \| past \| pdf \| discuss
7.	Too Noisy to Learn: Enhancing Data Quality for Code Review (arxiv.org)
	1 point by PaulHoule on Feb 21, 2025 \| hide \| past \| pdf \| discuss
8.	MLGym: A New Framework and Benchmark for Advancing AI Research Agents (arxiv.org)
	1 point by jonbaer on Feb 21, 2025 \| hide \| past \| pdf \| discuss
9.	Presumed Cultural Identity: How Names Shape LLM Responses (arxiv.org)
	1 point by SerCe on Feb 21, 2025 \| hide \| past \| pdf \| discuss