ML Newsnew | past | best | rss | submit about
Stories from July 28, 2026 (UTC)
Go back a day, month, or year. Go forward a day.
1. Kimi Linear: An Expressive, Efficient Attention Architecture (2025) (arxiv.org)
306 points by ronfriedhaber 67 days ago | hide | past | pdf | 133 comments
2. "Uncensored" open LLMs are measurably more optimistic than their base models (arxiv.org)
43 points by oleczek 67 days ago | hide | past | pdf | 22 comments
3. Show HN: A 6M-token movable window on a single 46GB GPU (arxiv.org)
7 points by Wetime 68 days ago | hide | past | pdf | 16 comments
4. Distilling proprietary model reasoning into open-source search agents (arxiv.org)
5 points by cpard 67 days ago | hide | past | pdf | discuss
5. CryptanalysisBench: Can LLMs Do Cryptanalysis? (arxiv.org)
2 points by rvz 67 days ago | hide | past | pdf | discuss
6. Agentic Permissions Policy Algebra for Taint Confinement in LLM Agents (arxiv.org)
2 points by zhinit 67 days ago | hide | past | pdf | discuss
7. LoopLynx: A Scalable Dataflow Architecture for Efficient LLM Inference (arxiv.org)
2 points by binyu 67 days ago | hide | past | pdf | discuss
8. Tag Questions and the Generational Reversal of Sycophancy Across 45 Language (arxiv.org)
2 points by sbulaev 67 days ago | hide | past | pdf | discuss
9. Neural Representation of Minimal Surfaces (arxiv.org)
2 points by E-Reverance 67 days ago | hide | past | pdf | discuss
10. Protocol-Level Attacks on Agentic Commerce Platforms: Taxonomy and Defense (arxiv.org)
2 points by sbulaev 68 days ago | hide | past | pdf | discuss
11. The SpiNNaker2 chip: a many-core platform for brain-inspired computing (arxiv.org)
1 point by Jimmc414 67 days ago | hide | past | pdf | discuss
About | RSS | RSS (all) | HN arXiv