October 18, 2024 front

ML Newsnew | past | best | rss | submit

Stories from October 18, 2024 (UTC)

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

1.	Numerical Precision Affects Mathematical Reasoning Capabilities of LLMs (arxiv.org)
	66 points by belter on Oct 18, 2024 \| hide \| past \| pdf \| 49 comments
2.	LLMD: A Large Language Model for Interpreting Longitudinal Medical Records (arxiv.org)
	48 points by troyastorino on Oct 18, 2024 \| hide \| past \| pdf \| 19 comments
3.	Agents Thinking Fast and Slow: A Talker-Reasoner Architecture (arxiv.org)
	13 points by robg on Oct 18, 2024 \| hide \| past \| pdf \| discuss
4.	NGPT: Normalized Transformer with Representation Learning on the Hypersphere (arxiv.org)
	4 points by programd on Oct 18, 2024 \| hide \| past \| pdf \| discuss
5.	Does Refusal Training in LLMs Generalize to the Past Tense? (arxiv.org)
	2 points by fzliu on Oct 18, 2024 \| hide \| past \| pdf \| discuss
6.	Decoupling Visual Encoding for Unified Multimodal Understanding and Generation (arxiv.org)
	2 points by jinqueeny on Oct 18, 2024 \| hide \| past \| pdf \| discuss
7.	Reducing the Transformer Architecture to a Minimum (arxiv.org)
	2 points by belter on Oct 18, 2024 \| hide \| past \| pdf \| discuss
8.	State-space models can learn in-context by gradient descent (arxiv.org)
	2 points by trextrex on Oct 18, 2024 \| hide \| past \| pdf \| discuss
9.	Understanding Slides and User Interfaces via Synthetic Data Generation (arxiv.org)
	1 point by PaulHoule on Oct 18, 2024 \| hide \| past \| pdf \| discuss