Papers by Zeng Zihao

9 papers
AdaMoE: Token-Adaptive Routing with Null Experts for Mixture-of-Experts Language Models (2024.findings-emnlp)

Copied to clipboard

Challenge: Existing MoE methods require a constant top-k routing for all tokens, which is restrictive because of the number of experts required for feature abstraction.
Approach: They propose a token-adaptive routing method that allows different tokens to select a different number of experts.
Outcome: a new method can reduce average expert load while achieving superior performance.
Beyond Noise: Characterizing Creative Potential in Unverifiable LLM Hallucinations (2026.acl-long)

Copied to clipboard

Challenge: Large Language Models generate outputs that extend beyond established knowledge . prior work does not characterize the unverifiable space as a whole .
Approach: They propose a novelty-verifiability characterization that distinguishes Creative Synthesis from Groundless Fabrication by a conceptual creation task.
Outcome: The proposed model distinguishes Creative Synthesis (Region A) from Groundless Fabrication (Regium B) it shows that Region A is non-negligible and robust, persisting across generation strategies, models, domains, and embedding choices.
Improving Automatic Grammatical Error Annotation for Chinese Through Linguistically-Informed Error Typology (2025.coling-main)

Copied to clipboard

Challenge: In educational settings, GEC systems provide immediate and consistent feedback to both native (L1) and non-native (L2) language learners.
Approach: They propose a framework that provides detailed feedback on 12-16% of all errors by identifying them under a new error typology, specific enough to uncover subtle differences in error patterns between L1 and L2 writings.
Outcome: The proposed framework can provide detailed feedback on 12-16% of all errors, revealing subtle differences in error patterns between L1 and L2 writings.
SIFT: Grounding LLM Reasoning in Contexts via Stickers (2025.findings-emnlp)

Copied to clipboard

Challenge: Using a new approach, we can improve the pass@1 accuracy of LLM reasoning in large language models.
Approach: They propose a method that leverages increasing inference-time compute to ground LLM reasoning in contexts.
Outcome: The proposed approach improves pass@1 accuracy of DeepSeek-R1 on AIME2024 from 78.33% to **85.67%** and that on Aime2025 from 69.8% to **77.33%**.
Reinforcement Learning on Pre-Training Data (2026.acl-long)

Copied to clipboard

Challenge: Recent progress in large language models is driven by scaling of training compute through pre-training with nexttoken prediction (NTP) or post-training (RL) Pre-training using NTP enables models to acquire extensive knowledge and skills from general data, but it suffers from data inefficiency and catastrophic forgetting in continual learning settings.
Approach: They propose to scale training compute through pre-training with next-token prediction (NTP) or post-training by scaling reinforcement learning (RL) to improve learning from general data.
Outcome: Experiments on multiple benchmarks and models show that the proposed approach improves continual pre-training and provides a strong foundation for post-training on Qwen3-8B-Base.
Attention Basin: Why Contextual Position Matters in Large Language Models (2026.acl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) are sensitive to the contextual position of information in input.
Approach: They introduce Attention-Driven Reranking (AttnRank) which estimates a model’s intrinsic positional attention preferences using a small calibration set and reorders retrieved documents or few-shot examples to align the most salient content with these high-attention positions.
Outcome: Experiments on multi-hop QA and few-shot in-context learning tasks show that AttnRank achieves substantial improvements across 10 large language models of varying architectures and scales, without modifying model parameters or training procedures.
Detecting AI-Generated Content on Social Media with Multi-modal Language Models (2026.acl-industry)

Copied to clipboard

Challenge: Existing methods for AI-generated content detection face poor generalization to newer models, reliance on single modalities, and lack of interpretable explanations.
Approach: They propose a model that curates diverse social media data and trains a vision-language model for detection and explanation.
Outcome: The proposed model achieves state-of-the-art detection performance on public benchmarks and observes positive downstream impacts on user engagement.
SEMIROUTER: Sparse-Data Enhanced Routing for Adaptive Multi-LLM System (2026.eacl-long)

Copied to clipboard

Challenge: Existing routing methods suffer from poor scalability and dependence on datasets for training . energy footprint is also considered in the decision to implement our new LLM routing framework .
Approach: They propose a new LLM routing framework that dynamically allocates queries to the most appropriate LLM.
Outcome: The proposed method improves data efficiency, adaptability, and routing accuracy compared to existing methods.
Activation-Guided Local Editing for Jailbreaking Attacks (2026.acl-long)

Copied to clipboard

Challenge: Existing methods for jailbreaking Large Language Models (LLMs) are limited and produce incoherent or unreadable inputs.
Approach: They propose a two-stage framework that performs a one-shot, scenario-based generation of context and rephrases the original malicious query to obscure its harmful intent.
Outcome: The proposed framework achieves state-of-the-art Attack Success Rate, with gains of up to 37.74% over the strongest baseline, and excellent transferability to black-box and large-scale models.

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