Papers by Xinyi Yu

16 papers
NL-Debugging: Exploiting Natural Language as an Intermediate Representation for Code Debugging (2025.emnlp-main)

Copied to clipboard

Challenge: Early debugging efforts focused on code-level analysis, which often fails when addressing complex programming errors.
Approach: They propose a framework that employs natural language as an intermediate representation to improve code debugging by debuggating at a natural language level.
Outcome: The proposed framework outperforms traditional debugging methods and enables a broader modification space through direct refinement guided by execution feedback.
MentalSeek-Dx: Towards Progressive Hypothetico-Deductive Reasoning for Real-world Psychiatric Diagnosis (2026.acl-long)

Copied to clipboard

Challenge: Mental health disorders represent a burgeoning global public health challenge . lack of ecological validity and fine-grained diagnostic supervision limits their utility .
Approach: They propose a medical-specialized LLM trained to internalize clinical reasoning process through supervised trajectory construction and curriculum-based reinforcement learning.
Outcome: The proposed model achieves state-of-the-art with only 14B parameters, establishing a clinically grounded framework for reliable psychiatric diagnosis.
DebateCoder: Towards Collective Intelligence of LLMs via Test Case Driven LLM Debate for Code Generation (2025.acl-long)

Copied to clipboard

Challenge: Existing debate-based approaches to code generation are limited due to several reasons: 1) Reliance on different instances of the same LLM for debate, 2) under-utilization of test cases, and 3) reliance on third-party moderators for result consolidation and decision-making.
Approach: They propose to use test cases to analyze code and identify bugs while opposing models generate test cases for each other to challenge each other's code during the debate process.
Outcome: The proposed model collects intelligence of LLMs via test case-driven debate for code generation.
SafeSteer: A Decoding-level Defense Mechanism for Multimodal Large Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Existing defense methods rely on fine-tuning or inefficient post-hoc interventions, limiting their ability to address novel attacks.
Approach: They propose a decoding-level defense mechanism that employs a lightweight discriminator to iteratively steer the decoding process toward safety.
Outcome: The proposed method improves safety performance by up to 33.40% without fine-tuning on multiple MLLMs.
LoRAPrune: Structured Pruning Meets Low-Rank Parameter-Efficient Fine-Tuning (2024.findings-acl)

Copied to clipboard

Challenge: Low-rank adaption (LoRA) is a low-level pruning method that can be expensive and slow to deploy.
Approach: They propose a low-rank adaption pruning framework that provides an accurate structured pruned model in a memory-efficient manner.
Outcome: The proposed pruning framework reduces perplexity and memory usage by 52.6% on LLaMA and T5 models while reducing memory usage.
Boost, Disentangle, and Customize: A Robust System2-to-System1 Pipeline for Code Generation (2025.findings-acl)

Copied to clipboard

Challenge: Existing systems 2 methods for code generation are difficult to implement due to the complex hidden reasoning process and heterogeneous data distribution.
Approach: They propose a framework that Boosts reasoning exploration via multi-agent collaboration and Disentangles heterogeneous data into specialized experts.
Outcome: The proposed framework outperforms state-of-the-art methods on APPS and CodeContest benchmarks and achieves 73.8% accuracy on hard problems.
Don’t Tell the Answer, Truly Guide the Reasoning During RL Rollouts (2026.findings-acl)

Copied to clipboard

Challenge: Existing methods such as GRPO often break down when task difficulty exceeds the model’s capacity, resulting in sparse rewards and inefficient training.
Approach: They propose to measure the compatibility between external guidance and a model's intrinsic policy by introducing an adaptive framework to enhance reasoning performance while explicitly preserving high Affinity.
Outcome: The proposed framework outperforms baseline models while maintaining high Affinity.
ADaPT: Token-Level Decoupling for Efficient Large Reasoning Models (2026.findings-acl)

Copied to clipboard

Challenge: Existing efficiency-oriented methods attempt to shorten or mix reasoning strategies, yet often degrade reasoning capability.
Approach: They propose a token-level dual-process framework that explicitly decouples efficiency and correctness signals during training.
Outcome: The proposed framework reduces inference cost while maintaining strong reasoning ability across multiple benchmarks.
DPN-LE: Dual Personality Neuron Localization and Editing for Large Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Current methods for editing personality traits in large language models can change personalities but reduce performance.
Approach: They propose a novel paradigm for personality editing that locates and edits LLM neurons and enables competitive personality control at inference time.
Outcome: Experiments on LLaMA-3-8B-Instruct and Qwen2.5-7B-instruct show that the proposed approach can improve performance and improve performance.
RethinkMCTS: Refining Erroneous Thoughts in Monte Carlo Tree Search for Code Generation (2025.emnlp-main)

Copied to clipboard

Challenge: Existing tree search methods neglect the underlying reasoning process, resulting in poor search quality.
Approach: They propose a framework that systematically explores and refines the reasoning process for code generation by using a tree search engine and a reflection mechanism.
Outcome: The proposed framework outperforms existing methods in the code generation domain.
CodePRM: Execution Feedback-enhanced Process Reward Model for Code Generation (2025.findings-acl)

Copied to clipboard

Challenge: Recent advances in code generation focus on optimizing the thought process, but lack effective process supervision, making it difficult to optimize the thoughts.
Approach: They propose a method that leverages the code execution feedback to build a code PRM by collecting a large dataset of thought traces and then training it to take both the reasoning process and code execution as input.
Outcome: The proposed approach outperforms baselines and strong LLMs in the inference stage.
Me-Agent: A Personalized Mobile Agent with Two-Level User Habit Learning for Enhanced Interaction (2026.findings-acl)

Copied to clipboard

Challenge: Existing Large Language Model (LLM)-based mobile agents follow explicit user instructions without personalized needs.
Approach: They propose a user preference learning strategy enhanced with a Personal Reward Model to improve personalization performance.
Outcome: The proposed agent achieves state-of-the-art performance while maintaining competitive instruction execution performance.
Exploring Fine-Grained Human Motion Video Captioning (2025.coling-main)

Copied to clipboard

Challenge: Existing video captioning models fail to capture nuanced semantics of videos . existing models generate coarse descriptions of human motions, resulting in poor quality .
Approach: They construct a fine-grained human motion video captioning dataset named BoFiT and a model that generates fine-grain descriptions of human motions via prompting.
Outcome: The proposed model outperforms existing models on comprehensive metrics.
DiffER: Diffusion Entity-Relation Modeling for Reversal Curse in Diffusion Large Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Existing large language models exhibit unidirectional behavior when processing bidirectional relationships . authors propose a solution to alleviate the reversal curse in Diffusion LLMs .
Approach: They propose a model that addresses the "reversal curse" of bidirectional behavior in large language models . they propose 'entity-aware training' and balanced data construction to alleviate asymmetry and missing relations .
Outcome: The proposed model alleviates the "reversal curse" in Diffusion LLMs . the proposed model employs whole-entity masking to mitigate entity fragmentation .
Root Defense Strategies: Ensuring Safety of LLM at the Decoding Level (2025.acl-long)

Copied to clipboard

Challenge: Existing methods to detect harmful outputs from prefill-level lacks utilization of the model’s decoding outputs, leading to relatively lower effectiveness and robustness.
Approach: They propose a robust decoding mechanism that corrects harmful queries directly rather than rejecting them outright.
Outcome: The proposed model improves model security without compromising reasoning speed.
LLM×MapReduce: Simplified Long-Sequence Processing using Large Language Models (2025.acl-long)

Copied to clipboard

Challenge: Existing studies have focused on extending the context length of large language models (LLMs) due to their quadratic computational complexity and a lack of high-quality long training examples, most LLMs are trained with a limited window size.
Approach: They propose a training-free framework that enables large language models to effectively process long texts using a divide-and-conquer strategy for comprehensive document understanding.
Outcome: The proposed framework outperforms open-source and commercial long-context LLMs and is compatible with several 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