October 3, 2025 front

ML Newsnew | past | best | rss | submit

Stories from October 3, 2025 (UTC)

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

1.	Thoughtbubbles: An Unsupervised Method for Parallel Thinking in Latent Space (arxiv.org)
	4 points by shetaye on Oct 3, 2025 \| hide \| past \| pdf \| 1 comment
2.	Pretraining Under Infinite Compute (arxiv.org)
	3 points by jedharris on Oct 3, 2025 \| hide \| past \| pdf \| 1 comment
3.	Aristotle: IMO-Level Automated Theorem Proving (arxiv.org)
	3 points by jasondavies on Oct 3, 2025 \| hide \| past \| pdf \| discuss
4.	Pretraining Large Language Models with NVFP4 (arxiv.org)
	1 point by aportnoy on Oct 3, 2025 \| hide \| past \| pdf \| discuss
5.	The AI Productivity Index (Apex) (arxiv.org)
	1 point by paulpauper on Oct 3, 2025 \| hide \| past \| pdf \| discuss
6.	xLSTM Scaling Laws: Competitive Performance with Linear Time-Complexity (arxiv.org)
	1 point by lairv on Oct 3, 2025 \| hide \| past \| pdf \| discuss
7.	Delta-Code: How Does RL Unlock and Transfer New Programming Algorithms in LLMs? (arxiv.org)
	1 point by sonabinu on Oct 3, 2025 \| hide \| past \| pdf \| discuss
8.	Security Degradation in Iterative AI Code Generation (arxiv.org)
	1 point by chillax on Oct 3, 2025 \| hide \| past \| pdf \| discuss