May 26, 2026 front

ML Newsnew | past | best | rss | submit

Stories from May 26, 2026 (UTC)

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

1.	A sleep-like consolidation mechanism for LLMs (arxiv.org)
	212 points by juxtapose 130 days ago \| hide \| past \| pdf \| 140 comments
2.	Investigating how prompt politeness affects LLM accuracy (2025) (arxiv.org)
	156 points by KnuthIsGod 131 days ago \| hide \| past \| pdf \| 208 comments
3.	Barriers to Complexity-Theoretic Proofs That "AGI" Using ML Is Impossible (arxiv.org)
	4 points by mike_uoftdcs 130 days ago \| hide \| past \| pdf \| discuss
4.	SkillOpt: Executive Strategy for Self-Evolving Agent Skills (arxiv.org)
	4 points by theaniketmaurya 130 days ago \| hide \| past \| pdf \| discuss
5.	ThriftAttention: Selective Mixed Precision for Long-Context FP4 Attention (arxiv.org)
	4 points by joesharratt29 131 days ago \| hide \| past \| pdf \| 1 comment
6.	Agentic Harness Engineering (arxiv.org)
	3 points by cobblr_mosaic 130 days ago \| hide \| past \| pdf \| discuss
7.	Polar: Agentic RL on Any Harness at Scale (arxiv.org)
	3 points by Brajeshwar 130 days ago \| hide \| past \| pdf \| discuss
8.	Quest: Training Frontier Deep Research Agents with Synthetic Tasks (arxiv.org)
	2 points by Brajeshwar 130 days ago \| hide \| past \| pdf \| discuss
9.	Contrastive Decoding Diffing: Recovering Finetuning Data Without Weight Access (arxiv.org)
	2 points by Timofeibu 131 days ago \| hide \| past \| pdf \| discuss
10.	Continual Speaker Identity Unlearning with Minimal Interference (arxiv.org)
	2 points by berlianta 131 days ago \| hide \| past \| pdf \| discuss
11.	MileStone: A Multi-Objective Compiler Phase Ordering Framework (arxiv.org)
	1 point by matt_d 131 days ago \| hide \| past \| pdf \| discuss