November 6, 2025 front

ML Newsnew | past | best | rss | submit

Stories from November 6, 2025 (UTC)

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

1.	LLMs encode how difficult problems are (arxiv.org)
	174 points by stansApprentice 331 days ago \| hide \| past \| pdf \| 38 comments
2.	Deep sequence models tend to memorize geometrically; it is unclear why (arxiv.org)
	4 points by amichail 332 days ago \| hide \| past \| pdf \| discuss
3.	Whisper Leak: a side-channel attack on Large Language Models (arxiv.org)
	3 points by bikenaga 332 days ago \| hide \| past \| pdf \| discuss
4.	HaluMem: Evaluating Hallucinations in Memory Systems of Agents (arxiv.org)
	2 points by timini 331 days ago \| hide \| past \| pdf \| 1 comment
5.	Mathematical Exploration and Discovery at Scale (arxiv.org)
	2 points by SweetSoftPillow 332 days ago \| hide \| past \| pdf \| discuss
6.	The Curved Spacetime of Transformer Architectures (arxiv.org)
	2 points by luis_likes_math 332 days ago \| hide \| past \| pdf \| 1 comment
7.	Sequential Scaling Beats Majority Voting at Matched Compute (arxiv.org)
	2 points by paraschopra 332 days ago \| hide \| past \| pdf \| discuss
8.	Accumulating Context Changes the Beliefs of Language Models (arxiv.org)
	2 points by Anon84 332 days ago \| hide \| past \| pdf \| discuss
9.	Evaluating Control Protocols for Untrusted AI Agents (arxiv.org)
	1 point by timini 331 days ago \| hide \| past \| pdf \| 1 comment
10.	The OpenHands Software Agent SDK: Composable and Extensible (arxiv.org)
	1 point by timini 332 days ago \| hide \| past \| pdf \| 1 comment