Papers by Yiyang Zhou

9 papers
Tracking the Limits of Knowledge Propagation: How LLMs Fail at Multi-Step Reasoning with Conflicting Knowledge (2026.eacl-long)

Copied to clipboard

Challenge: Existing benchmarks for analyzing the performance of Large Language Models (LLMs) focus on single knowledge updates and fact recall, but do not consider how these updates affect downstream reasoning.
Approach: They propose a benchmark to study how LLMs propagate new knowledge when it conflicts with the model's parametric knowledge.
Outcome: The proposed benchmark compared models with no updated facts to show that the new methods worsen performance and improve reasoning performance.
Improving Alignment in LVLMs with Debiased Self-Judgment (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing methods for aligning LVLMs rely on external datasets, human annotations or complex post-processing.
Approach: They propose a method that generates a debiased self-judgment score for LVLMs . this self-evaluation metric is created internally by the model without external resources .
Outcome: The proposed approach outperforms existing methods in reducing hallucinations and safety concerns.
Conditional Dichotomy Quantification via Geometric Embedding (2025.acl-long)

Copied to clipboard

Challenge: Existing methods that rely on semantic similarity fail to capture the nuanced oppositional dynamics essential for these applications.
Approach: They propose a task that formalizes the measurement of conditional dichotomy by using a dichotomian framework.
Outcome: The proposed framework provides carefully constructed datasets covering debate, defeasible inference, and causal reasoning scenarios.
Unraveling Misinformation Propagation in LLM Reasoning (2025.findings-emnlp)

Copied to clipboard

Challenge: Large Language Models (LLMs) have demonstrated impressive capabilities in reasoning, but how they propagate within their reasoning process remains underexplored.
Approach: They propose a practical approach to mitigating misinformation propagation in LLMs by applying factual corrections early in the reasoning process and fine-tuning on synthesized data with early-stage corrections significantly improves reasoning factuality.
Outcome: The proposed model can correct misinformation when explicitly instructed, but fails to correct misinformation less than half the time even with explicit instructions.
SAPGraph: Structure-aware Extractive Summarization for Scientific Papers with Heterogeneous Graph (2022.aacl-main)

Copied to clipboard

Challenge: Abstractive and extractive methods are used to condense long text into concise summaries while retaining essential information.
Approach: They propose to use paper structure to extract paper summaries from long text . they provide a large-scale dataset of COVID-19-related papers .
Outcome: The proposed framework generates more comprehensive and valuable summaries compared to previous work on COVID-19-related papers.
SimpleOCR: Rendering Visual Questions to Teach MLLMs to Read (2026.findings-acl)

Copied to clipboard

Challenge: MLLMs lack visual grounding mechanism to read text embedded in images, or rely on parametric shortcuts . despite strong OCR capabilities, models suffer performance degradation of 12.7% in the VQ setting .
Approach: They propose a plug-and-play training strategy that invalidates shortcuts in text prompts . they propose 'vq' setting where text queries are rendered directly onto images .
Outcome: The proposed training strategy surpasses the base model by 5.4% and GRPO based on original images by 2.7% on four representative OOD benchmarks.
GLIMPSE: Do Large Vision-Language Models Truly Think With Videos or Just Glimpse at Them? (2025.emnlp-main)

Copied to clipboard

Challenge: Existing video benchmarks often resemble image-based questions with scans of only a few key frames, without deep temporal reasoning.
Approach: They propose a video benchmark to assess whether large vision-language models can genuinely think with videos rather than perform superficial frame-level analysis.
Outcome: The proposed benchmark consists of 3,269 videos and over 4,342 highly visual-centric questions across 11 categories, including Trajectory Analysis, Temporal Reasoning, and Forensics Detection.
Improving Factual Consistency in Abstractive Summarization with Sentence Structure Pruning (2024.lrec-main)

Copied to clipboard

Challenge: Abstractive summarization models suffer from factual inconsistency problem . post-editing methods focus on replacing suspicious entities, failing to modify incorrect content hidden in sentence structures.
Approach: They propose to use sentence pruning operation to correct possible errors . they propose to apply sentence pruning operations to the syntactic dependency tree .
Outcome: The proposed method improves factual consistency on the FRANK dataset compared with baselines . it is model-independent and can serve as the final step in ensuring factual consistentness.
Enhancing Visual-Language Modality Alignment in Large Vision Language Models via Self-Improvement (2025.findings-naacl)

Copied to clipboard

Challenge: Existing methods for visual and language alignment depend on external models or data, leading to uncontrollable and unstable results.
Approach: They propose a framework that enhances visual and language alignment without external dependencies by incorporating an in-context self-critic mechanism that constructs preference pairs for tuning.
Outcome: The proposed framework outperforms existing methods and improves performance on 14 hallucination and comprehensive benchmarks.

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