Papers by Xiaotian Chen

12 papers
Refusal-Aware Red Teaming: Exposing Inconsistency in Safety Evaluations (2025.emnlp-main)

Copied to clipboard

Challenge: Large Language Models (LLMs) require rigorous safety evaluations to be effective.
Approach: They propose a red teaming framework that detects internal model refusals and contrasts them with judgments from an external safety evaluator to generate test cases that expose such discrepancies.
Outcome: The proposed framework outperforms existing reinforcement learning-based approaches in generating diverse test cases and achieves a substantially higher discovery rate of refusal gaps.
Towards Proactive Personalization through Profile Customization for Individual Users in Dialogues (2026.findings-acl)

Copied to clipboard

Challenge: Existing alignment methods focus on universal human values or static, single-turn preferences, thereby failing to address the critical needs of long-term personalization and the initial user cold-start problem.
Approach: They propose a user-centric lifelong agent that continuously infers and adapts to user preferences.
Outcome: The proposed agent achieves superior performance over strong prompt-based and policy optimization baselines, not only in idealized but also in noisy conversational contexts.
KELE: Residual Knowledge Erasure for Enhanced Multi-hop Reasoning in Knowledge Editing (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing knowledge editing techniques show limitations when applied to multi-hop reasoning . residual single-hop knowledge causes edited models to revert to original answers .
Approach: They propose a knowledge editing method that incorporates a Knowledge Erasure mechanism for Large language model Editing (KELE) they propose an erasure function for residual knowledge and an injection function for new knowledge .
Outcome: The proposed method significantly improves multi-hop reasoning capability of edited models.
Persona-judge: Personalized Alignment of Large Language Models via Token-level Self-judgment (2025.findings-acl)

Copied to clipboard

Challenge: Existing methods for aligning language models with human preferences rely on reward signals and additional annotated data, limiting their scalability and adaptability to diverse human values.
Approach: They propose a discriminative paradigm that leverages the intrinsic preference judgment capabilities of the model to align language models with human preferences.
Outcome: The proposed model is scalable and efficient, paving the way for more adaptive personalized alignment.
ACRM: Multi-Agent Trajectory Learning for Automated Credit Risk Model Refreshing in Production (2026.acl-industry)

Copied to clipboard

Challenge: Credit risk models suffer from rapid performance decay due to distribution shifts, requiring frequent updates to meet strict operational guardrails.
Approach: They propose a multi-agent framework that treats model refreshing as a learnable trajectory of agent interactions.
Outcome: The proposed framework reduces the average model refresh cycle from weeks to 1.1 days and iteration rounds by 65% while maintaining superior stability metrics.
InfiMM: Advancing Multimodal Understanding with an Open-Sourced Visual Language Model (2024.findings-acl)

Copied to clipboard

Challenge: InfiMM is a multimodal large language model that adapts to complex vision-language tasks.
Approach: They present a Multimodal Large Language Model that adapts to intricate vision-language tasks using large-scale training data and comprehensive training strategies.
Outcome: Empirical evaluations across a variety of benchmarks underscore InfiMM’s remarkable capability in multimodal understanding.
Spectral Characterization and Mitigation of Sequential Knowledge Editing Collapse (2026.acl-long)

Copied to clipboard

Challenge: Existing approaches to reducing the effects of knowledge editing are insufficiently understood.
Approach: They propose a plug-and-play framework that preserves the dominant subspace of the original weights and analyzes parameter updates in the spectral basis of the weights.
Outcome: The proposed framework improves editing efficacy while preserving general abilities under long-horizon sequential editing, including extreme settings with up to 20,000 edits.
GOBench: Stage-Wise Diagnostics and the Visual Paradox in Multimodal Graph Optimization (2026.findings-acl)

Copied to clipboard

Challenge: Existing benchmarks fail to represent multimodal problem specifications, score outcomes only and cannot localize where failures occur along the modeling pipeline.
Approach: They propose a Graph Optimization benchmark that aligns multiple modalities with solver-derived oracles and a diagnostic protocol that evaluates intermediate artifacts as well as end results.
Outcome: Graph Optimization benchmark (GOBench) evaluates intermediate artifacts as well as end results . vision reliably increases inference cost, while reliability impact is regime-dependent . current benchmarks fail to represent multimodal problem specifications, fail to localize failures .
Quantize What Counts: More for Keys, Less for Values (2026.findings-acl)

Copied to clipboard

Challenge: Empirical evaluations across various prominent LLMs and benchmarks show that key-favored allocations retain up to 98.3% accuracy compared to uniform allocations (e.g., 4-bit keys, 2-bit values).
Approach: They propose two theorems that anchor mixed-precision KV quantization in the intrinsic geometry of Transformer models.
Outcome: Empirical evaluations show that key-favored allocations retain up to 98.3% accuracy while conserving memory.
Knowledge Graph Enhanced Large Language Model Editing (2024.emnlp-main)

Copied to clipboard

Challenge: Existing methods for editing large language models struggle to track and incorporate changes in knowledge associated with edits, which limits the generalization ability of post-edit LLMs in processing edited knowledge.
Approach: They propose a model editing method that leverages knowledge graphs to enhance LLM editing by capturing changes in associated knowledge by constructing an external graph.
Outcome: The proposed method improves the generalization ability of LLMs in processing edited knowledge.
UIPE: Enhancing LLM Unlearning by Removing Knowledge Related to Forgetting Targets (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing methods to unlearning large language models focus on forgetting target data while overlooking the impact of logically related knowledge on the effectiveness of unlearning.
Approach: They propose a method that removes knowledge highly correlated with the forgetting targets and a technique that remove logically related knowledge from the model.
Outcome: The proposed method significantly improves the performance of the proposed method on the TOFU and WMDP benchmarks.
DiffPO: Diffusion-styled Preference Optimization for Inference Time Alignment of Large Language Models (2025.acl-long)

Copied to clipboard

Challenge: Inference-time alignment approaches still face limitations due to policy-specific value functions and latency during the inference phase.
Approach: They propose an efficient and policy-agnostic preference optimization method that avoids time latency associated with token generation.
Outcome: The proposed method achieves a favorable trade-off between alignment quality and inference-time latency.

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