10 papers
Understanding Reasoning from Pretraining to Post-Training
Jingyan Shen, Ang Li, Salman Rahman +4
Reinforcement learning (RL) has become central to improving large language models (LLMs) on complex reasoning tasks, yet RL post-training is largely studied in isolation from the p…
Towards Privacy-Preserving Federated Prompt Tuning under Data Heterogeneity: A Subspace-Decomposed Expert Approach
Yuhua Wang, Xiaodong Li, Yihao Guo +6
Federated prompt tuning (FPT) enables collaborative adaptation of vision--language models (VLMs) using lightweight prompts. Existing methods often address heterogeneity and privacy…
Taming Noise-Induced Prototype Degradation for Privacy-Preserving Personalized Federated Fine-Tuning
Yuhua Wang, Qinnan Zhang, Xiaodong Li +6
Prototype-based Personalized Federated Learning (ProtoPFL) enables efficient multi-domain adaptation by communicating compact class prototypes, but directly sharing them poses priv…
Mask-Free Privacy Extraction and Rewriting: A Domain-Aware Approach via Prototype Learning
Xiaodong Li, Yuhua Wang, Qingchen Yu +5
Client-side privacy rewriting is crucial for deploying LLMs in privacy-sensitive domains. However, existing approaches struggle to balance privacy and utility. Full-text methods of…
MonitorBench: A Comprehensive Benchmark for Chain-of-Thought Monitorability in Large Language Models
Han Wang, Yifan Sun, Brian Ko +8
Large language models (LLMs) can generate chains of thought (CoTs) that are not always causally responsible for their final outputs. When such a mismatch occurs, the CoT no longer…
When Reasoning Meets Its Laws
Junyu Zhang, Yifan Sun, Tianang Leng +4
Despite the superior performance of Large Reasoning Models (LRMs), their reasoning behaviors are often counterintuitive, leading to suboptimal reasoning capabilities. To theoretica…
Learning a Zeroth-Order Optimizer for Fine-Tuning LLMs
Kairun Zhang, Haoyu Li, Yanjun Zhao +2
Zeroth-order optimizers have recently emerged as an attractive approach for fine-tuning large language models (LLMs), as they avoid backpropagation and can substantially reduce mem…
Improving Data Efficiency for LLM Reinforcement Fine-tuning Through Difficulty-targeted Online Data Selection and Rollout Replay
Yifan Sun, Jingyan Shen, Yibin Wang +4
Reinforcement learning (RL) has become an effective approach for fine-tuning large language models (LLMs), particularly to enhance their reasoning capabilities. However, RL fine-tu…
The Emperor's New Clothes in Benchmarking? A Rigorous Examination of Mitigation Strategies for LLM Benchmark Data Contamination
Yifan Sun, Han Wang, Dongbai Li +2
Benchmark Data Contamination (BDC)-the inclusion of benchmark testing samples in the training set-has raised increasing concerns in Large Language Model (LLM) evaluation, leading t…
SVIP: Towards Verifiable Inference of Open-source Large Language Models
Yifan Sun, Yuhang Li, Yue Zhang +2
The ever-increasing size of open-source Large Language Models (LLMs) renders local deployment impractical for individual users. Decentralized computing has emerged as a cost-effect…