Papers by Zeyu Zheng

7 papers
Mechanistic Interpretability Should Prioritize Feature Consistency in Sparse Autoencoders (2026.acl-long)

Copied to clipboard

Challenge: Sparse Autoencoders (SAEs) are a tool in mechanistic interpretability (MI) but the aspiration to identify a canonical set of features is challenged by the observed inconsistency of learned SAE features across different training runs.
Approach: They propose to use the Pairwise Dictionary Mean Correlation Coefficient to quantify SAE feature consistency as an evaluation axis alongside reconstruction and sparsity.
Outcome: The proposed measure is based on the pairwise dictionary mean correlation coefficient (PW-MCC) on LLM activations.
PaddleSpeech: An Easy-to-Use All-in-One Speech Toolkit (2022.naacl-demo)

Copied to clipboard

Challenge: PaddleSpeech is an open-source speech toolkit that supports speech-to-text and text-to speech tasks.
Approach: They describe the design philosophy and core architecture of PaddleSpeech to support several essential speech-to-text and text-to speech tasks.
Outcome: The proposed framework achieves competitive or state-of-the-art performance on various speech datasets and implements the most popular methods.
Towards Hierarchical Multi-Step Reward Models for Enhanced Reasoning in Large Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Existing Process Reward Models (PRMs) are vulnerable to reward hacking and require expensive, large-scale annotation of reasoning steps.
Approach: They propose a reward model approach which evaluates both individual and consecutive reasoning steps from fine-grained and coarse-grounded level.
Outcome: Empirical results show that the proposed model performs better than existing PRMs and is more robust than existing models.
Demons in the Detail: On Implementing Load Balancing Loss for Training Specialized Mixture-of-Expert Models (2025.acl-long)

Copied to clipboard

Challenge: Existing Mixture-of-Experts training frameworks use a micro-batch to calculate LBL . micro-batches are restricted to a single sequence, preventing expert specialization .
Approach: They propose to use a global-batch to loosen the load balance constraint for MoEs models . they propose to synchronize fi across micro-batches and then use it to calculate the LBL .
Outcome: The proposed global-batch LBL improves the domain specialization of experts . the micro-battery LBL is almost at the sequence level, and the router is pushed to distribute the token evenly .
Grammar-Based Code Representation: Is It a Worthy Pursuit for LLMs? (2025.findings-acl)

Copied to clipboard

Challenge: Existing research demonstrates the effectiveness of grammar-based code representations in small-scale models, showing their ability to reduce syntax errors and enhance performance.
Approach: They develop a series of billion-scale grammar-based code representations that incorporate grammar rules into the code generation process.
Outcome: Experiments on HumanEval and MBPP show that grammar-based representations reduce syntax errors and improve performance even in billion-scale models.
CypherSmith: Transforming Text-to-Cypher Generation for LLMs with Synthetic Data (2026.acl-long)

Copied to clipboard

Challenge: Existing datasets are small, domain-limited, and lack diversity, constraining LLM progress.
Approach: They propose a knowledge Graph retrieval tool that can translate natural language questions into structured queries.
Outcome: Extensive experiments show that CypherSmith achieves state-of-the-art performance.
A Comprehensive Survey of Process Reward Models: Data Generation, Model Construction, and Usage (2026.acl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) have advanced reasoning ability, yet conventional alignment remains dominated by outcome reward models that judge only final answers.
Approach: They summarize applications across math, code, text, multimodal reasoning, robotics, and agents . goal is to clarify design spaces, reveal open challenges, and guide future research toward fine-grained, robust reasoning alignment.
Outcome: The proposed model enables finer credit assignment, richer diagnostics, and improved robustness.

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