ZICQ curates daily AI news and deep dives on models, open-source infrastructure, and developer field reports.
Submit an articleCurrent filter: Tag: #专家混合架构 Clear filters
LLMs & Foundation Models· 2026-10-21#DScout#Diffusion Models#Speculative Decoding#Large Language Models#Inference Acceleration
By Mr.Xu0 views0 discussions
arXiv introduces DScout, a novel method for factorization-guided diffusion-drafted specula…
Research & Papers· 2026-10-21
Mr.Xu0 views0 discussions
This study investigates the diffusion of social interaction norms (e.g., sharing space, ta…
LLMs & Foundation Models· 2026-10-21
Mr.Xu0 views0 discussions
This research provides a theoretical breakthrough in prefix-tuning, a parameter-efficient …
Research & Papers· 2026-10-21
Mr.Xu0 views0 discussions
This research investigates the use of large language models (LLMs) and sparse autoencoders…
LLMs & Foundation Models· 2026-10-21
Mr.Xu0 views0 discussions
arXiv introduces OR-License, a novel benchmark for evaluating the performance of Large Lan…
LLMs & Foundation Models· 2026-10-21
Mr.Xu0 views0 discussions
A new arXiv study investigates the decision-making mechanisms of Large Language Model (LLM…
LLMs & Foundation Models· 2026-10-21
Mr.Xu0 views0 discussions
Huvyuan-MT-7B is an open-weight translation model designed as an alternative to GPT-4 for …
LLMs & Foundation Models· 2026-10-21
Mr.Xu0 views0 discussions
arXiv introduces PARADE, a novel framework for simulating heterogeneous response distribut…
Research & Papers· 2026-10-21
Mr.Xu0 views0 discussions
arXiv introduces a novel approach for automated theorem proving that leverages a bounded, …
LLMs & Foundation Models· 2026-10-21
Mr.Xu0 views0 discussions
This research investigates how to combine individual behavioral histories using Large Lang…
LLMs & Foundation Models· 2026-10-21
Mr.Xu0 views0 discussions
arXiv introduces miniTE, a lightweight framework designed to address the execution consist…
Research & Papers· 2026-10-21
Mr.Xu0 views0 discussions
arXiv has announced a novel research method that separates conformal reliability from freq…
