Yiqi Liu

Yiqi Liu (刘 溢祺)

PhD Student · University of Manchester

I am a final-year PhD student at the University of Manchester, supervised by Prof. Chenghua Lin. My research focuses on LLM-as-a-Judge, LLM Agent, and Mechanistic Interpretability, with current interests in long-horizon agents and medical agents. I began my PhD at the University of Sheffield before transferring to Manchester. Previously, I received a BEng in Engineering Mechanics from Chongqing University.

Google Scholar GitHub
Yiqi Liu hiking in the hills

News

Preprints

2026

World Embedding Benchmark

Yiqi Liu, R Yuan, Y Wang, L Li, F Cai, H P Chan, J Yu, H Zhang, C Lin, C Xiao

arXiv 2026, submitted to ICLR

World Embedding Benchmark, figure 1World Embedding Benchmark, figure 2

Publications

2026

Broken Symmetry in LLM Refusal: Answer Release Is More Local Than Refusal Restoration

Yiqi Liu, Y Wang, S Wang, C Xiao, C Lin

EMNLP 2026 Main (Acceptance rate 15.4%)

2026

JudgeMoE: Distributional Aggregation for LLM-as-a-Judge

Yiqi Liu, J James, Y Wang, K Zhao, C Xiao, C Lin

IJCNLP–AACL 2026 Main

2026

When Tool-Backed Skill Retrieval Fails: Source-Style Collapse in Executable Capability Retrieval

Yiqi Liu, J James, Y Wang, C Xiao, C Lin

IJCNLP–AACL 2026 Findings

2026

Feedback as an Interface: Controlling Form and Detail in LLM Translation Refinement

Yiqi Liu, F Tomori, N. S. Moosavi, S Eger, C Lin

IJCNLP–AACL 2026 Findings

2026

Rate-Resolved Text Evaluation with Masked Diffusion Language Models

Yiqi Liu, K Zhao, C Lin, N. S. Moosavi

INLG 2026

2026

The Achilles' Heel of Angular Margins: A Chebyshev Polynomial Fix for Speaker Verification

Y Wang, Yiqi Liu, C Xiao, C Lin

ICASSP 2026

2026

Reading Between the Frames: Interpreting Implicit and Non-literal Meaning in Social Media Videos

Y Wang, Y Ma, Yiqi Liu, Z Y Chang, C-L Chen, C-Y Hsiao, T Loakman, A Villavicencio, C Xiao, C Lin

arXiv 2026

2026

An Approach for the Comparative Evaluation of Requirements Formalisation Approaches

S Kolahdouz Rahimi, K Lano, S Yassipour Tehrani, C Lin, Yiqi Liu, M A Umar

Model-Based Software and Systems Engineering 2026 · MODELSWARD 2024 selected papers

2025

Beyond One-Size-Fits-All: Inversion Learning for Highly Effective NLG Evaluation Prompts

H Hong, C Xiao, Y Wang, Yiqi Liu, W Rong, C Lin

TACL 2025

2025

ContrastScore: Towards Higher Quality, Less Biased, More Efficient Evaluation Metrics with Contrastive Evaluation

X Wang, D Larionov, S Wu, Yiqi Liu, S Eger, N. S. Moosavi, C Lin

IJCNLP–AACL 2025

2024

LLMs as Narcissistic Evaluators: When Ego Inflates

Yiqi Liu, N. S. Moosavi, C Lin

ACL 2024 Findings (Citations: 100+)

2024

Comparative Evaluation of NLP Approaches for Requirements Formalisation

S Kolahdouz Rahimi, K Lano, S Yassipour Tehrani, C Lin, Yiqi Liu, M A Umar

MODELSWARD 2024

2023

MARBLE: Music Audio Representation Benchmark for Universal Evaluation

R Yuan, Y Ma, Y Li, G Zhang, X Chen, H Yin, Yiqi Liu, et al.

NeurIPS 2023

Work Experience

May 2026 – Present

Alibaba DAMO Academy

Research Intern

2024 – 2025

University of Manchester

Teaching Assistant

  • Natural Language Processing COMP34711
  • Machine Learning COMP24112
  • Introduction to Programming COMP16321
  • Natural Language Understanding COMP34812

Education

Dec 2023 – Present

University of Manchester

PhD Student

LLM-as-a-Judge, LLM agents, and mechanistic interpretability.

Dec 2022 – Dec 2023

University of Sheffield

PhD Student · Transferred to Manchester

Natural language generation and automatic evaluation.

Sep 2018 – Jun 2022

Chongqing University

BEng in Engineering Mechanics

Research Support

Service

Reviewer: ARR, ICML, ICLR, NLPCC, ICCV, and ISMIR.