Jiacheng Liu

Abu Dhabi, UAE

jiacheng.liu [at] mbzuai.ac.ae

My research focuses on efficient language and generative models, and self-improving agents.

Previously MSc at UCL, advised by Prof. Jun Wang; part-time research intern at Huawei Noah’s Ark Lab, London (2023–2024) with Dr. Kun Shao.

Currently, I am a research intern with Ant Group’s LLaDA-Uni team, working on generative model training under the mentorship of Prof. Tao Lin.

publications

^* Equal contribution.

2026

DMLR

Mobile-MMLU: A Mobile Intelligence Language Understanding Benchmark

Sondos Mahmoud Bsharat, Mukul Ranjan, Aidar Myrzakhan, Jiacheng Liu, Bowei Guo, Shengkun Tang, Zhuang Liu, Yuanzhi Li, and Zhiqiang Shen

Journal of Data-centric Machine Learning Research (DMLR), 2026

arXiv Bib HTML Code

@article{bsharat2026mobilemmlu,
  title = {Mobile-{MMLU}: A Mobile Intelligence Language Understanding Benchmark},
  author = {Bsharat, Sondos Mahmoud and Ranjan, Mukul and Myrzakhan, Aidar and Liu, Jiacheng and Guo, Bowei and Tang, Shengkun and Liu, Zhuang and Li, Yuanzhi and Shen, Zhiqiang},
  journal = {Journal of Data-centric Machine Learning Research (DMLR)},
  year = {2026},
}

arXiv

LLaDA-Image: Building Strong Image Generators with Fully Open Training Recipes

Chuyan Chen, Haoxing Chen², Kun Chen², Zhenglin Cheng¹, Long Cui, Ruishan Fang, Zhangxuan Gu, Zhicheng Huang², Zhenzhong Lan, Yuanting Lei, Haoquan Li, Jianguo Li, Rongchuan Li, Sidu Li, Tao Lin, Deyuan Liu, Jiacheng Liu¹, Lin Liu, Yuxuan Lou, Zhisheng Lu, Yuxin Ma, Shuheng Shen, Peng Sun¹, Chaoyang Wang, Hongjun Wang, Xiaomei Wang², Yongxin Wang², Chengzhang Wu, Hongru Wu, and Jun Xie¹

¹ Core contributors to image generators; ² core contributors to the understanding backbone. Authors listed alphabetically by surname.

arXiv preprint arXiv:2609.03796, 2026

arXiv Bib HTML Code

@article{chen2026lladaimage,
  title = {{LLaDA-Image}: Building Strong Image Generators with Fully Open Training Recipes},
  author = {Chen, Chuyan and Chen, Haoxing and Chen, Kun and Cheng, Zhenglin and Cui, Long and Fang, Ruishan and Gu, Zhangxuan and Huang, Zhicheng and Lan, Zhenzhong and Lei, Yuanting and Li, Haoquan and Li, Jianguo and Li, Rongchuan and Li, Sidu and Lin, Tao and Liu, Deyuan and Liu, Jiacheng and Liu, Lin and Lou, Yuxuan and Lu, Zhisheng and Ma, Yuxin and Shen, Shuheng and Sun, Peng and Wang, Chaoyang and Wang, Hongjun and Wang, Xiaomei and Wang, Yongxin and Wu, Chengzhang and Wu, Hongru and Xie, Jun},
  journal = {arXiv preprint arXiv:2609.03796},
  year = {2026},
}

ICML

Hard Labels In! Rethinking the Role of Hard Labels in Mitigating Local Semantic Drift

Jiacheng Cui, Bingkui Tong, Xinyue Bi, Xiaohan Zhao, Jiacheng Liu, and Zhiqiang Shen

In International Conference on Machine Learning (ICML), Jul 2026

arXiv Bib HTML

@inproceedings{cui2026hardlabels,
  title = {Hard Labels In! Rethinking the Role of Hard Labels in Mitigating Local Semantic Drift},
  author = {Cui, Jiacheng and Tong, Bingkui and Bi, Xinyue and Zhao, Xiaohan and Liu, Jiacheng and Shen, Zhiqiang},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2026},
  month = jul,
}

CVPR
BiGain: Unified Token Compression for Joint Generation and Classification

Jiacheng Liu^*, Shengkun Tang^*, Jiacheng Cui, Dongkuan Xu, and Zhiqiang Shen

In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026

Abs arXiv Bib HTML Code

A training-free, plug-and-play framework that preserves generation quality while improving classification accuracy in accelerated diffusion models. Built on frequency separation: Laplacian-gated token merging plus interpolate-extrapolate KV downsampling.
@inproceedings{liu2026bigain, title = {BiGain: Unified Token Compression for Joint Generation and Classification}, author = {Liu, Jiacheng and Tang, Shengkun and Cui, Jiacheng and Xu, Dongkuan and Shen, Zhiqiang}, booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, year = {2026}, }

arXiv

Dive into Claude Code: The Design Space of Today’s and Future AI Agent Systems

Jiacheng Liu, Xiaohan Zhao, Xinyi Shang, and Zhiqiang Shen

arXiv preprint arXiv:2604.14228, 2026

⭐ 2K+ GitHub stars

arXiv Bib HTML Code

@article{liu2026claudecode,
  title = {Dive into Claude Code: The Design Space of Today's and Future {AI} Agent Systems},
  author = {Liu, Jiacheng and Zhao, Xiaohan and Shang, Xinyi and Shen, Zhiqiang},
  journal = {arXiv preprint arXiv:2604.14228},
  year = {2026},
  note = {⭐ 2K+ GitHub stars}
}

ICML

Next-Gen CAPTCHAs: Leveraging the Cognitive Gap for Scalable and Diverse GUI-Agent Defense

Jiacheng Liu^*, Yaxin Luo^*, Jiacheng Cui, Xinyi Shang, Xiaohan Zhao, and Zhiqiang Shen

In International Conference on Machine Learning (ICML), 2026

arXiv Bib HTML Code Website

@inproceedings{liu2026nextgen,
  title = {Next-Gen {CAPTCHAs}: Leveraging the Cognitive Gap for Scalable and Diverse {GUI}-Agent Defense},
  author = {Liu, Jiacheng and Luo, Yaxin and Cui, Jiacheng and Shang, Xinyi and Zhao, Xiaohan and Shen, Zhiqiang},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2026},
}

arXiv

AutoDesign: Meta-Harness Optimization for Long-Horizon Agentic Design

Yaxin Luo, Haobin Jiang, Jialv Zou, Xu Huang, Wenhao Yan, Haodong Li, Zhengrong Yue, Jing Li, Xiaofu Chen, Xiaohan Zhao, Jiacheng Liu, Jiacheng Cui, Zhiqiang Shen, and Xiaotong Li

arXiv preprint arXiv:2608.13560, 2026

arXiv Bib HTML

@article{luo2026autodesign,
  title = {{AutoDesign}: Meta-Harness Optimization for Long-Horizon Agentic Design},
  author = {Luo, Yaxin and Jiang, Haobin and Zou, Jialv and Huang, Xu and Yan, Wenhao and Li, Haodong and Yue, Zhengrong and Li, Jing and Chen, Xiaofu and Zhao, Xiaohan and Liu, Jiacheng and Cui, Jiacheng and Shen, Zhiqiang and Li, Xiaotong},
  journal = {arXiv preprint arXiv:2608.13560},
  year = {2026},
}

ACL

LLMSurgeon: Diagnosing Data Mixture of Large Language Models

Yaxin Luo, Jiacheng Cui, Xiaohan Zhao, Xinyi Shang, Jiacheng Liu, Xinyue Bi, Zhaoyi Li, and Zhiqiang Shen

In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Jul 2026

DOI Bib HTML Code

@inproceedings{luo2026llmsurgeon,
  title = {{LLMS}urgeon: Diagnosing Data Mixture of Large Language Models},
  author = {Luo, Yaxin and Cui, Jiacheng and Zhao, Xiaohan and Shang, Xinyi and Liu, Jiacheng and Bi, Xinyue and Li, Zhaoyi and Shen, Zhiqiang},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  month = jul,
  pages = {42444--42459},
  doi = {10.18653/v1/2026.acl-long.1964},
}

ECCV

From Masks to Pixels and Meaning: A New Taxonomy, Benchmark, and Metrics for VLM Image Tampering

Xinyi Shang, Yi Tang, Jiacheng Cui, Ahmed Elhagry, Salwa K. Al Khatib, Sondos Mahmoud Bsharat, Jiacheng Liu, Xiaohan Zhao, Jing-Hao Xue, Hao Li, Salman Khan, and Zhiqiang Shen

In European Conference on Computer Vision (ECCV), 2026

arXiv Bib HTML Code

@inproceedings{shang2026pixar,
  title = {From Masks to Pixels and Meaning: A New Taxonomy, Benchmark, and Metrics for {VLM} Image Tampering},
  author = {Shang, Xinyi and Tang, Yi and Cui, Jiacheng and Elhagry, Ahmed and Al Khatib, Salwa K. and Bsharat, Sondos Mahmoud and Liu, Jiacheng and Zhao, Xiaohan and Xue, Jing-Hao and Li, Hao and Khan, Salman and Shen, Zhiqiang},
  year = {2026},
  booktitle = {European Conference on Computer Vision (ECCV)}
}

arXiv

FigMirror: Ground It, Code It, Plot It

Xiaohan Zhao^*, Jiacheng Liu^*, Yaxin Luo, and Zhiqiang Shen

arXiv preprint arXiv:2608.28814, 2026

arXiv Bib HTML Code

@article{zhao2026figmirror,
  title = {{FigMirror}: Ground It, Code It, Plot It},
  author = {Zhao, Xiaohan and Liu, Jiacheng and Luo, Yaxin and Shen, Zhiqiang},
  journal = {arXiv preprint arXiv:2608.28814},
  year = {2026},
}

ICML

Exploring 3D Dataset Pruning

Xiaohan Zhao, Xinyi Shang, Jiacheng Liu, and Zhiqiang Shen

In International Conference on Machine Learning (ICML), Jul 2026

arXiv Bib HTML Code

@inproceedings{zhao2026threedpruning,
  title = {Exploring {3D} Dataset Pruning},
  author = {Zhao, Xiaohan and Shang, Xinyi and Liu, Jiacheng and Shen, Zhiqiang},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2026},
  month = jul,
}

2025

NeurIPS

FADRM: Fast and Accurate Data Residual Matching for Dataset Distillation

Jiacheng Cui, Xinyue Bi, Yaxin Luo, Xiaohan Zhao, Jiacheng Liu, and Zhiqiang Shen

In Conference on Neural Information Processing Systems (NeurIPS), 2025

arXiv Bib HTML Code

@inproceedings{cui2025fadrm,
  title = {{FADRM}: Fast and Accurate Data Residual Matching for Dataset Distillation},
  author = {Cui, Jiacheng and Bi, Xinyue and Luo, Yaxin and Zhao, Xiaohan and Liu, Jiacheng and Shen, Zhiqiang},
  booktitle = {Conference on Neural Information Processing Systems (NeurIPS)},
  year = {2025},
}

NeurIPS

Open CaptchaWorld: A Comprehensive Web-based Platform for Testing and Benchmarking Multimodal LLM Agents

Yaxin Luo, Zhaoyi Li, Jiacheng Liu, Jiacheng Cui, Xiaohan Zhao, and Zhiqiang Shen

In NeurIPS Datasets and Benchmarks Track, 2025

arXiv Bib HTML Code

@inproceedings{luo2025captchaworld,
  title = {Open {CaptchaWorld}: A Comprehensive Web-based Platform for Testing and Benchmarking Multimodal {LLM} Agents},
  author = {Luo, Yaxin and Li, Zhaoyi and Liu, Jiacheng and Cui, Jiacheng and Zhao, Xiaohan and Shen, Zhiqiang},
  booktitle = {NeurIPS Datasets and Benchmarks Track},
  year = {2025},
}

2023

arXiv

Pangu-Agent: A Fine-Tunable Generalist Agent with Structured Reasoning

Filippos Christianos, Georgios Papoudakis, Matthieu Zimmer, Thomas Coste, Zhihao Wu, Jingxuan Chen, Khyati Khandelwal, James Doran, Xidong Feng, Jiacheng Liu, Zheng Xiong, Yicheng Luo, Jianye Hao, Kun Shao, Haitham Bou-Ammar, and Jun Wang

arXiv preprint arXiv:2312.14878, 2023

Work conducted as part-time research intern at Huawei Noah’s Ark Lab, London.

arXiv Bib HTML

@article{christianos2023panguagent,
  title = {Pangu-Agent: A Fine-Tunable Generalist Agent with Structured Reasoning},
  author = {Christianos, Filippos and Papoudakis, Georgios and Zimmer, Matthieu and Coste, Thomas and Wu, Zhihao and Chen, Jingxuan and Khandelwal, Khyati and Doran, James and Feng, Xidong and Liu, Jiacheng and Xiong, Zheng and Luo, Yicheng and Hao, Jianye and Shao, Kun and Bou-Ammar, Haitham and Wang, Jun},
  journal = {arXiv preprint arXiv:2312.14878},
  year = {2023},
  note = {Work conducted as part-time research intern at Huawei Noah's Ark Lab, London.}
}

Manuscript Under Review

AutoLMCompress: Autonomous Research for Extreme Compression of Per-Layer Embeddings

Submitted to the NeurIPS 2026 Workshop for Autonomous Machine Learning Research

A verifiable, scalable, and composable framework for autonomous research on language model compression, with cross-layer latent vector quantization for per-layer embeddings.

visitor map