September 19, 2025 front

ML Newsnew | past | best | rss | submit

Stories from September 19, 2025 (UTC)

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

1.	The Common Pile v0.1: An 8TB Dataset of Public Domain and Openly Licensed Text (arxiv.org)
	68 points by djoldman on Sep 19, 2025 \| hide \| past \| pdf \| 17 comments
2.	Why most AI coding benchmarks are misleading (COMPASS paper) (arxiv.org)
	13 points by jmeaden on Sep 19, 2025 \| hide \| past \| pdf \| 2 comments
3.	Is Chain-of-Thought Reasoning of LLMs a Mirage? (arxiv.org)
	3 points by SomaticPirate on Sep 19, 2025 \| hide \| past \| pdf \| discuss
4.	Generalizable Geometric Image Caption Synthesis (arxiv.org)
	3 points by ceolin on Sep 19, 2025 \| hide \| past \| pdf \| discuss
5.	3D and 4D World Modeling: A Survey (arxiv.org)
	3 points by PaulHoule on Sep 19, 2025 \| hide \| past \| pdf \| discuss
6.	Internalizing Self-Consistency in LMs: Multi-Agent Consensus Alignment (arxiv.org)
	2 points by simonpure on Sep 19, 2025 \| hide \| past \| pdf \| discuss
7.	Igniting VLMs Toward the Embodied Space (arxiv.org)
	2 points by NeoInHacker on Sep 19, 2025 \| hide \| past \| pdf \| discuss
8.	Verbalized Algorithms (arxiv.org)
	1 point by PaulHoule on Sep 19, 2025 \| hide \| past \| pdf \| discuss
9.	VCBench: Benchmarking LLMs in Venture Capital (arxiv.org)
	1 point by simonpure on Sep 19, 2025 \| hide \| past \| pdf \| discuss