May 22, 2026 front

ML Newsnew | past | best | rss | submit

Stories from May 22, 2026 (UTC)

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

1.	CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs (arxiv.org)
	105 points by matt_d 135 days ago \| hide \| past \| pdf \| 12 comments
2.	Domain-Camouflaged Injection Attacks Evade Detection in Multi-Agent LLM Systems (arxiv.org)
	38 points by sbulaev 134 days ago \| hide \| past \| pdf \| 4 comments
3.	Moss: Self-Evolution Through Source-Level Rewriting in Autonomous Agent Systems (arxiv.org)
	11 points by Timofeibu 135 days ago \| hide \| past \| pdf \| discuss
4.	Event Tensor: A Unified Abstraction for Compiling Dynamic Megakernel (arxiv.org)
	6 points by matt_d 134 days ago \| hide \| past \| pdf \| discuss
5.	Measuring Security Without Fooling Ourselves: Why Benchmarking Agents Is Hard (arxiv.org)
	3 points by Timofeibu 134 days ago \| hide \| past \| pdf \| discuss
6.	Meow-Omni 1: a multi-modal feline LLM (arxiv.org)
	3 points by Otterly99 135 days ago \| hide \| past \| pdf \| discuss
7.	An AI system to help scientists write expert-level empirical software (arxiv.org)
	2 points by cyco130 135 days ago \| hide \| past \| pdf \| discuss
8.	Matching Principle: Adversarial, augmentation, etc. are estimators of one matrix (arxiv.org)
	2 points by visark_42 135 days ago \| hide \| past \| pdf \| discuss
9.	Negation Neglect: When models fail to learn negations in training (arxiv.org)
	2 points by chr15m 135 days ago \| hide \| past \| pdf \| discuss
10.	Is Capability a Liability? More Capable Language Models Make Worse Forecasts (arxiv.org)
	1 point by sbulaev 135 days ago \| hide \| past \| pdf \| discuss
11.	PopPy: Opportunistically Exploiting Parallelism in Python Compound AI Apps (arxiv.org)
	1 point by matt_d 135 days ago \| hide \| past \| pdf \| discuss