February 26, 2025 front

ML Newsnew | past | best | rss | submit

Stories from February 26, 2025 (UTC)

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

1.	The FFT Strikes Back: An Efficient Alternative to Self-Attention (arxiv.org)
	456 points by iNic on Feb 26, 2025 \| hide \| past \| pdf \| 168 comments
2.	Improving Consistency in Large Language Models Through Chain of Guidance (arxiv.org)
	11 points by Anon84 on Feb 26, 2025 \| hide \| past \| pdf \| discuss
3.	MITRE's Offensive Security Evaluation Framework for LLMs (arxiv.org)
	5 points by fourierslide on Feb 26, 2025 \| hide \| past \| pdf \| 1 comment
4.	Dataflow Execution in GPU (arxiv.org)
	3 points by ban-lan-gen on Feb 26, 2025 \| hide \| past \| pdf \| discuss
5.	Fractal Generative Models (arxiv.org)
	2 points by lnyan on Feb 26, 2025 \| hide \| past \| pdf \| discuss
6.	Twilight: Adaptive Attention Sparsity with Hierarchical Top-$P$ Pruning (arxiv.org)
	2 points by PaulHoule on Feb 26, 2025 \| hide \| past \| pdf \| discuss
7.	Prompt-to-Leaderboard (arxiv.org)
	1 point by CrypticShift on Feb 26, 2025 \| hide \| past \| pdf \| discuss
8.	Demonstrating specification gaming in reasoning models (arxiv.org)
	1 point by wluk on Feb 26, 2025 \| hide \| past \| pdf \| 1 comment