Papers by Xiaohui Song

4 papers
BiLD: Bi-directional Logits Difference Loss for Large Language Model Distillation (2025.coling-main)

Copied to clipboard

Challenge: Knowledge distillation (KD) is a method for reducing model size while preserving performance.
Approach: They propose a method to distill large language models at the logit level by transferring knowledge from a large teacher model to a smaller student model.
Outcome: The proposed method outperforms supervised fine-tuning, vanilla KL loss and five other distillation methods on 13 datasets.
Awakening Dormant Experts:Counterfactual Routing to Mitigate MoE Hallucinations (2026.acl-long)

Copied to clipboard

Challenge: Sparse Mixture-of-Experts models are vulnerable to hallucinations, authors say . static Top-k routing leaves "specialist experts" under-prioritized for specific tokens .
Approach: They propose a training-free inference framework to awaken dormant experts . they propose 'counterfactual routing' to shift computational resources from syntax-dominant to knowledge-intensive layers .
Outcome: Experiments show that CoR improves factual accuracy by 3.1% without increasing the inference budget.
Align Attention Heads Before Merging Them: An Effective Way for Converting MHA to GQA (2025.findings-emnlp)

Copied to clipboard

Challenge: Large language models (LLMs) have demonstrated exceptional performance across diverse natural language processing tasks.
Approach: They propose a method for converting multi-head attention into grouped-query attention with any compression ratio of KV heads.
Outcome: The proposed method can compress up to 87.5% KV heads of LLaMA2-7B model and 75% Kv heads of Sheared-LLa MA-1.3B with acceptable performance degradation.
Supervised Prototypical Contrastive Learning for Emotion Recognition in Conversation (2022.emnlp-main)

Copied to clipboard

Challenge: Existing methods to capture emotions in conversation (ERC) lack the correlation between emotions and semantics, resulting in many challenges.
Approach: They propose a Supervised Prototypical Contrastive Learning (SPCL) loss for the ERC task . they use a Prototype Network to leverage the supervised contrastive learning approach .
Outcome: The proposed approach outperforms CoG-BART's proposed approach on three widely used benchmarks and shows that it is effective on multiple scenarios.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations