activity
20242026
most citedDeepSeek-V3.2: Pushing the Frontier of Open Large Language Models

8 citations · 10 across the 6 of their papers we have counts for

collaborators

11 papers

cs.LG2026

Why GRPO Needs Normalization: A Local-Curvature Perspective on Adaptive Gradients

Cheng Ge, Caitlyn Heqi Yin, Hao Liang +1

Reinforcement learning (RL) has become a key driver of language model reasoning. Among RL algorithms, Group Relative Policy Optimization (GRPO) is the de facto standard, avoiding t…

cs.CL20258 cited

DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models

DeepSeek-AI, Aixin Liu, Aoxue Mei +260

We introduce DeepSeek-V3.2, a model that harmonizes high computational efficiency with superior reasoning and agent performance. The key technical breakthroughs of DeepSeek-V3.2 ar…

cs.AI2025

Jarvis: Towards Personalized AI Assistant via Personal KV-Cache Retrieval

Binxiao Xu, Junyu Feng, Shaolin Lu +5

The rapid development of Vision-language models (VLMs) enables open-ended perception and reasoning. Recent works have started to investigate how to adapt general-purpose VLMs into…

cs.CV2025

LongInsightBench: A Comprehensive Benchmark for Evaluating Omni-Modal Models on Human-Centric Long-Video Understanding

ZhaoYang Han, Qihan Lin, Hao Liang +3

We introduce \textbf{LongInsightBench}, the first benchmark designed to assess models' ability to understand long videos, with a focus on human language, viewpoints, actions, and o…

cs.CV2025

Native Visual Understanding: Resolving Resolution Dilemmas in Vision-Language Models

Junbo Niu, Yuanhong Zheng, Ziyang Miao +8

Vision-Language Models (VLMs) face significant challenges when dealing with the diverse resolutions and aspect ratios of real-world images, as most existing models rely on fixed, l…

cs.CV2025

UniCTokens: Boosting Personalized Understanding and Generation via Unified Concept Tokens

Ruichuan An, Sihan Yang, Renrui Zhang +10

Personalized models have demonstrated remarkable success in understanding and generating concepts provided by users. However, existing methods use separate concept tokens for under…