Papers by Xiaocheng Zhang

12 papers
Beyond Reasoning Gains: Mitigating General-Capability Forgetting in Large Reasoning Models (2026.findings-acl)

Copied to clipboard

Challenge: Reinforcement learning with verifiable rewards (RLVR) has delivered impressive gains in mathematical and multimodal reasoning . however, the recipe introduces a significant risk of capability regression, where models forget foundational skills after prolonged training without employing regularization strategies.
Approach: They propose a replay strategy with dynamic objective reweighting for general knowledge preservation using short-horizon signals of convergence and instability.
Outcome: The proposed method preserves general capabilities and improves reasoning . it can be applied to existing RLVR pipelines without training additional models or tuning .
CLAIM: Mitigating Multilingual Object Hallucination in Large Vision-Language Models with Cross-Lingual Attention Intervention (2025.acl-long)

Copied to clipboard

Challenge: Large Vision-Language Models (LVLMs) have impressive multimodal abilities but remain prone to multilingual object hallucination.
Approach: They propose a cross-lingual attention intervention method to mitigate multilingual object hallucination in LVLMs by aligning attention patterns.
Outcome: The proposed method improves 13.56% (up to 30%) on the POPE and 21.75% on the hallucination subsets across languages.
EscapeBench: Towards Advancing Creative Intelligence of Language Model Agents (2025.acl-long)

Copied to clipboard

Challenge: Existing language model agents excel in planning and reasoning, but lack creativity in unfamiliar environments.
Approach: They propose a benchmark suite of room escape game environments to challenge agents with creative reasoning, unconventional tool use and iterative problem-solving to uncover implicit goals.
Outcome: The proposed framework can perform with 40% fewer steps and hints and performs robustly across difficulty levels.
LI4: Label-Infused Iterative Information Interacting Based Fact Verification in Question-answering Dialogue (2024.lrec-main)

Copied to clipboard

Challenge: Existing studies on fact verification have failed to fully exploit question structures and ignoring relevant label information during the verification process.
Approach: They propose a new approach for question-answering dialogue based fact verification using label-infused iterative information interacting.
Outcome: The proposed approach achieves remarkable performance on HEALTHVER, FAVIQ, and COLLOQUIAL.
One for All: Update Parameterized Knowledge Across Multiple Models with Once Edit (2025.acl-long)

Copied to clipboard

Challenge: Existing methods for modifying large language models focus on individual models, resulting in errors and hallucinations.
Approach: They propose an ensemble-based approach that employs a plug-in model as the editing module and a dynamic weight mechanism to enhance its effectiveness.
Outcome: The proposed approach outperforms existing methods while achieving superior editing efficiency.
Controllable Text Generation via Probability Density Estimation in the Latent Space (2023.acl-long)

Copied to clipboard

Challenge: Existing control approaches cannot effectively model complex space with diverse attributes, high dimensionality, and asymmetric structure, leaving subsequent controls unsatisfactory.
Approach: They propose a control framework using probability density estimation in the latent space and an invertible transformation function that maps the complex distributions to simple Gaussian distributions in the prior space.
Outcome: The proposed method outperforms baselines on attribute relevance and text quality, achieving a new SOTA.
WebAnchor: Anchoring Agent Planning to Stabilize Long-Horizon Web Reasoning (2026.findings-acl)

Copied to clipboard

Challenge: Existing methods for reinforcement learning (RL)-based agents struggle with long-horizon planning and strategy coherence.
Approach: They propose a reinforcement learning framework that decouples planning and execution.
Outcome: The proposed framework outperforms baseline and first-step RL frameworks on four benchmarks.
Learning Which Features Matter: RoBERTa Acquires a Preference for Linguistic Generalizations (Eventually) (2020.emnlp-main)

Copied to clipboard

Challenge: Pretraining on self-supervised linguistic tasks is effective for learning features helpful for language understanding, but it requires more data to learn to prefer linguistic generalizations over surface ones.
Approach: They propose a set of 20 ambiguous binary classification tasks to test whether a pretrained model prefers linguistic or surface generalizations.
Outcome: The proposed model can learn to represent linguistic features with little pretraining data, but requires far more data to learn to prefer linguistic generalizations over surface ones.
A Distributional Lens for Multi-Aspect Controllable Text Generation (2022.emnlp-main)

Copied to clipboard

Challenge: Existing methods for multi-aspect control suffer from attribute degeneration due to mutual interference of these controllers.
Approach: They propose to use attribute fusion to find the intersections of multiple attributes as their combination for generation.
Outcome: The proposed method outperforms baselines on attribute relevance and text quality and achieves the SOTA.
When Do You Need Billions of Words of Pretraining Data? (2021.acl-long)

Copied to clipboard

Challenge: Pretrained language models (LMs) are dominated by models that can encode billions of words.
Approach: They use classifier probing, information-theoretic probing and unsupervised relative acceptability judgments to evaluate model ability.
Outcome: The proposed models require only about 10M to 100M words to learn to encode most syntactic and semantic features.
CC-Tuning: A Cross-Lingual Connection Mechanism for Improving Joint Multilingual Supervised Fine-Tuning (2025.acl-long)

Copied to clipboard

Challenge: Existing fine-tuning approaches that focus on English-centric training corpora often introduce implicit cross-lingual alignment, overlooking the potential for more profound, latent-level cross-linguistic interactions.
Approach: They propose a multilingual fine-tuning paradigm that explicitly establishes a cross-lingual connection mechanism at the latent level.
Outcome: The proposed model outperforms vanilla SFT and offers a strong latent-level alternative to data-level augmentation methods.
TrendFact: A Benchmark Towards Hotspot Perception in Automatic Fact-Checking (2026.acl-long)

Copied to clipboard

Challenge: Existing benchmarks lack social metadata and evaluation framework to meet this urgent evaluation needs.
Approach: They propose a benchmark capable of evaluating HPA and three fact-checking tasks.
Outcome: The proposed framework improves HPA and computational efficiency for RLM-driven systems.

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