1 citations · 2 across the 18 of their papers we have counts for
20 papers
Is Better Teacher Supervision Enough? Unlocking Student-side Learning in Multimodal On-Policy Distillation
Siyuan Liu, Kanghui Tian, Yue Duan +4
On-policy distillation (OPD) improves reasoning by providing token-level supervision from a teacher on a student's own trajectories. Existing methods primarily focus on enhancing t…
What Should a Self-Teacher See? Privileged Context Design for On-Policy Self-Distillation
Kanghui Tian, Siyuan Liu, Tianxiang Jiang +8
More privileged information does not always make a better teacher. We study this tension in on-policy self-distillation (OPSD), where a self-teacher scores the student's own rollou…
A*-Thought-V2: Efficient Latent Reasoning via Geometric Dynamics of LLM
Xiaoang Xu, Siyuan Liu, Shuo Wang +13
Chain-of-Thought (CoT) improves the reasoning ability of Large Language Models (LLMs) but incurs substantial computation and context costs. Existing methods either lose intermediat…
How Data Shapes RoPE Frequency Usage: From Positional Scale Matching to Length Generalization
Xinyi Wu, Siyuan Liu, Ali Jadbabaie
Rotary Position Embeddings (RoPE) provide transformers with a fixed grid of positional frequencies, yet trained models use these frequencies highly non-uniformly. We study what det…
Late-Layer Fusion is Enough: Dual-Path Vision Token Routing for Multimodal Large Language Models under Visual Saturation
Siyuan Liu, Jinyang Wu
Multimodal large language models (MLLMs) commonly inherit the deep, symmetric Transformer backbone designed for unimodal text modeling, and apply the same computation uniformly to…
Defending Against Malicious Finetuning by Scaling Train-time Adversarial Attacks
Haoming Wen, Shi Chen, Qingyu Shi +4
Malicious finetuning attacks pose a major safety threat against open-weight large language models (LLMs). However, existing alignment-stage defenses provide limited protection agai…
ViCuR: Visual Cues as Recoverable Privilege for Multimodal On-Policy Distillation
Kanghui Tian, Siyuan Liu, Ziang Yan +3
On-policy distillation (OPD) improves reasoning by training a student on trajectories sampled from its own policy under supervision from a teacher. In multimodal reasoning, a commo…
OpenSkillEval: Automatically Auditing the Open Skill Ecosystem for LLM Agents
Jiahao Ying, Boxian Ai, Wei Tang +2
Skills, i.e., structured workflow instructions distilled for large language models (LLMs), are becoming an increasingly important mechanism for improving agent performance on real-…
Data Difficulty and the Generalization--Extrapolation Tradeoff in LLM Fine-Tuning
Siyuan Liu, Tinghong Chen, Xinghan Li +2
Data selection during supervised fine-tuning (SFT) can critically change the behavior of large language models (LLMs). Although existing work has studied the effect of selecting da…
Do We Still Need GraphRAG? Benchmarking RAG and GraphRAG for Agentic Search Systems
Dongzhe Fan, Zheyi Xue, Siyuan Liu +1
Retrieval-augmented generation (RAG) and its graph-based extensions (GraphRAG) are effective paradigms for improving large language model (LLM) reasoning by grounding generation in…
Enhancing Persona Following at Decoding Time via Dynamic Importance Estimation for Role-Playing Agents
Yuxin Liu, Mingye Zhu, Siyuan Liu +2
The utility of Role-Playing Language Agents in sociological research is growing alongside the adoption of Large Language Models. For realism in social simulation, these agents must…
MiniCPM-SALA: Hybridizing Sparse and Linear Attention for Efficient Long-Context Modeling
MiniCPM Team, Wenhao An, Yingfa Chen +44
The evolution of large language models (LLMs) towards applications with ultra-long contexts faces challenges posed by the high computational and memory costs of the Transformer arc…
Discourse-Aware Dual-Track Streaming Response for Low-Latency Spoken Dialogue Systems
Siyuan Liu, Jiahui Xu, Feng Jiang +6
Achieving human-like responsiveness is a critical yet challenging goal for cascaded spoken dialogue systems. Conventional ASR-LLM-TTS pipelines follow a strictly sequential paradig…
What Do LLM Agents Know About Their World? Task2Quiz: A Paradigm for Studying Environment Understanding
Siyuan Liu, Hongbang Yuan, Xinze Li +3
Large language model (LLM) agents have demonstrated remarkable capabilities in complex decision-making and tool-use tasks, yet their ability to generalize across varying environmen…
EMemBench: Interactive Benchmarking of Episodic Memory for VLM Agents
Xinze Li, Ziyue Zhu, Siyuan Liu +4
We introduce EMemBench, a programmatic benchmark generator for evaluating long-term episodic memory of agents through interactive games. Rather than using a fixed set of questions,…
SimRPD: Optimizing Recruitment Proactive Dialogue Agents through Simulator-Based Data Evaluation and Selection
Zhiyong Cao, Dunqiang Liu, Qi Dai +9
Task-oriented proactive dialogue agents play a pivotal role in recruitment, particularly for steering conversations towards specific business outcomes, such as acquiring social-med…
Fast LLM Post-training via Decoupled and Fastest-of-N Speculation
Rongxin Cheng, Kai Zhou, Xingda Wei +8
Rollout dominates the training time in large language model (LLM) post-training, where the trained model is used to generate tokens given a batch of prompts. This work, SpecActor,…
I Know What You Said: Unveiling Hardware Cache Side-Channels in Local Large Language Model Inference
Zibo Gao, Junjie Hu, Feng Guo +5
Large Language Models (LLMs) that can be deployed locally have recently gained popularity for privacy-sensitive tasks, with companies such as Meta, Google, and Intel playing signif…
InfiMed: Low-Resource Medical MLLMs with Advancing Understanding and Reasoning
Zeyu Liu, Zhitian Hou, Guanghao Zhu +3
Multimodal Large Language Models (MLLMs) have achieved remarkable progress in domains such as visual understanding and mathematical reasoning. However, their application in the med…
Two-Stage Regularization-Based Structured Pruning for LLMs
Mingkuan Feng, Jinyang Wu, Siyuan Liu +7
The deployment of large language models (LLMs) is largely hindered by their large number of parameters. Structural pruning has emerged as a promising solution. Prior structured pru…