11 papers
PRPO: Aligning Process Reward with Outcome Reward in Policy Optimization
Ruiyi Ding, Yongxuan Lv, Xianhui Meng +4
Policy optimization for large language models often suffers from sparse reward signals in multi-step reasoning tasks. Critic-free methods like GRPO assign a single normalized outco…
Window-Diffusion: Accelerating Diffusion Language Model Inference with Windowed Token Pruning and Caching
Fengrui Zuo, Zhiwei Ke, Yiming Liu +3
Diffusion language models (DLMs) generate text through iterative denoising, but inference requires full-sequence attention at every iteration, resulting in substantial redundant co…
KlingAvatar 2.0 Technical Report
Kling Team, Jialu Chen, Yikang Ding +25
Avatar video generation models have achieved remarkable progress in recent years. However, prior work exhibits limited efficiency in generating long-duration high-resolution videos…
VideoMiner: Iteratively Grounding Key Frames of Hour-Long Videos via Tree-based Group Relative Policy Optimization
Xinye Cao, Hongcan Guo, Jiawen Qian +6
Understanding hour-long videos with multi-modal large language models (MM-LLMs) enriches the landscape of human-centered AI applications. However, for end-to-end video understandin…
Thinking with Camera: A Unified Multimodal Model for Camera-Centric Understanding and Generation
Kang Liao, Size Wu, Zhonghua Wu +6
Camera-centric understanding and generation are two cornerstones of spatial intelligence, yet they are typically studied in isolation. We present Puffin, a unified camera-centric m…
Understanding Textual Capability Degradation in Speech LLMs via Parameter Importance Analysis
Chao Wang, Rui-Chen Zheng, Yang Ai +1
The integration of speech into Large Language Models (LLMs) has substantially expanded their capabilities, but often at the cost of weakening their core textual competence. This de…
DAMR: Efficient and Adaptive Context-Aware Knowledge Graph Question Answering with LLM-Guided MCTS
Yingxu Wang, Shiqi Fan, Mengzhu Wang +3
Knowledge Graph Question Answering (KGQA) aims to interpret natural language queries and perform structured reasoning over knowledge graphs by leveraging their relational and seman…
CIDER: A Causal Cure for Brand-Obsessed Text-to-Image Models
Fangjian Shen, Zifeng Liang, Chao Wang +1
Text-to-image (T2I) models exhibit a significant yet under-explored "brand bias", a tendency to generate contents featuring dominant commercial brands from generic prompts, posing…
LLIA -- Enabling Low-Latency Interactive Avatars: Real-Time Audio-Driven Portrait Video Generation with Diffusion Models
Haojie Yu, Zhaonian Wang, Yihan Pan +7
Diffusion-based models have gained wide adoption in the virtual human generation due to their outstanding expressiveness. However, their substantial computational requirements have…
Accelerating LLM Inference Throughput via Asynchronous KV Cache Prefetching
Yanhao Dong, Yubo Miao, Weinan Li +4
Large Language Models (LLMs) exhibit pronounced memory-bound characteristics during inference due to High Bandwidth Memory (HBM) bandwidth constraints. In this paper, we propose an…
GeoDecoder: Empowering Multimodal Map Understanding
Feng Qi, Mian Dai, Zixian Zheng +1
This paper presents GeoDecoder, a dedicated multimodal model designed for processing geospatial information in maps. Built on the BeitGPT architecture, GeoDecoder incorporates spec…