Papers by Zhengyu Chen

7 papers
Revisiting Scaling Laws for Language Models: The Role of Data Quality and Training Strategies (2025.acl-long)

Copied to clipboard

Challenge: Existing scaling laws suggest augmenting model size and training data results in enhanced performance, but recent studies reveal deviations, particularly in large language models, where performance improvements decelerate—a phenomenon known as sub-scaling.
Approach: They propose a sub-optimal scaling law that better predicts performance in sub-scaling regimes by examining data quality and training strategies.
Outcome: The proposed scaling law better predicts performance in sub-scaling regimes, highlighting the importance of data quality and diversity.
Mitigating Structural Knowledge Collapse in Domain-Specific LLMs via Morpheme-Aware KV-Aggregation (2026.acl-long)

Copied to clipboard

Challenge: Existing tokenizers over-fragment domain terms, disrupting morpheme semantics.
Approach: They propose a lightweight tokenizer that dynamically consolidates fragments without tokenizer changes.
Outcome: The proposed adapter outperforms vocabulary adaptation baselines on medical and legal terms by 3.2–4.6% and 7.9% on high-fragmentation terms.
MAPO: Boosting Large Language Model Performance with Model-Adaptive Prompt Optimization (2023.findings-emnlp)

Copied to clipboard

Challenge: Existing research emphasizes the importance of adapting prompts to specific tasks, rather than specific LLMs.
Approach: They propose a model-adaptive prompt optimizer method that optimizes original prompts for each LLM in downstream tasks.
Outcome: The proposed method can optimize prompts for an LLM in downstream tasks.
Let’s Ask GNN: Empowering Large Language Model for Graph In-Context Learning (2024.findings-emnlp)

Copied to clipboard

Challenge: Textual Attributed Graphs (TAGs) are crucial for modeling complex real-world systems, yet leveraging large language models (LLMs) for TAGs presents unique challenges due to the gap between sequential text processing and graph-structured data.
Approach: They propose a novel approach that leverages In-Context Learning to integrate graph data and task-specific information into large language models (LLMs) they employ a Graph Neural Network-powered structure-enhanced retriever to select labeled nodes across graphs, incorporating complex graph structures and their supervision signals.
Outcome: Experiments on three tasks and seven LLMs show that AskGNN performs better than existing methods.
Scaling Laws Across Model Architectures: A Comparative Analysis of Dense and MoE Models in Large Language Models (2024.emnlp-main)

Copied to clipboard

Challenge: a study of large language models (LLMs) reveals the transferability and discrepancies of scaling laws between Dense and MoE models.
Approach: They investigate the transferability and discrepancies of scaling laws between Dense Models and Mixture of Experts models.
Outcome: The results show that the power-law scaling framework also applies to MoE Models .
SampleMix: A Sample-wise Pre-training Data Mixing Strategy by Coordinating Data Quality and Diversity (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing methods for pretraining data mixing for large language models neglect significant inter-domain overlaps and commonalities, failing to control the global diversity of the constructed training dataset.
Approach: They propose a sample-wise data mixture approach that performs global cross-domain sampling by systematically evaluating the quality and diversity of each sample.
Outcome: The proposed method exceeds existing domain-based methods in multiple downstream tasks and perplexity assessments.
Explaining Length Bias in LLM-Based Preference Evaluations (2025.findings-emnlp)

Copied to clipboard

Challenge: a preference evaluation metric is often biased towards longer responses, revealing a reliability problem . a decomposition of the preference evaluation into two components is needed to understand this bias.
Approach: They propose to decompose the preference evaluation metric into two key components . the first component is length-dependent and related to trustworthiness .
Outcome: The proposed evaluation metric is based on two components: desirability and information mass.

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