Yulan He

When Thinking Backfires: Mechanistic Insights into Reason-induced Misalignment

Latent Refinement Decoding: Enhancing Diffusion-Based Language Models by Refining Belief States

Towards Causal Fine-Tuning under Latent-Confounded Shift

MULTI-FACETED MULTIMODAL MONOSEMANTICITY

Soft Reasoning: Navigating Solution Spaces in Large Language Models through Controlled Embedding Exploration

SciReplicate-Bench: Benchmarking LLMs in Agent-driven Algorithmic Reproduction from Research Papers

Direct Preference Optimization Using Sparse Feature-level Constraints

Fine-Tuning Pre-trained Language Models for Robust Causal Representation Learning

Constrain Alignment with Sparse Autoencoders

GARLIC: LLM-Guided Dynamic Progress Control with Hierarchical Weighted Graph for Long Document QA

Enhancing LLM Faithfulness in Rationale Generation via Dual-Reward Probabilistic Inference