I am currently a second-year MSML student at Carnegie Mellon University advised by Prof. William W. Cohen. Previously, I completed my undergraduate studies in Computer Science at University of Rochester, where I had the privilege of being advised by Prof. Jiebo Luo. I have also collaborated with Prof. Haohan Wang at UIUC DREAM Lab.

In the summer of 2024, I was a visiting researcher at Washington University in St. Louis with Prof. Jiaxin Huang. Currently, for summer 2025, I am a Student Researcher at Google Research advised by Dr. Si Si.

My research interests lie in improving training and inference efficiency as well as model alignment of both LLMs and VLMs.

Feel free to email me if you are interested in collaborating or discussing research ideas.

🔥 News

2025.09: 🎉🎉 Our Survey Paper on “Reliable and Responsible Foundation Models” is out and accepted by TMLR.
2025.06: 🎉🎉 One paper “CrossWordBench: Evaluating the Reasoning Capabilities of LLMs and LVLMs with Controllable Puzzle Generation” is accepted by COLM 2025.
2025.06: 🎉🎉 New preprints “Semi-structured LLM Reasoners Can Be Rigorously Audited” and “PosS:Position Specialist Generates Better Draft for Speculative Decoding” is out on arxiv. Code is also released.
2025.05: 🎓🎓 I will join Google Research as a Student Researcher for summer 2025.
2025.02: 🎉🎉 One paper “Taming Overconfidence in LLMs: Reward Calibration in RLHF” is accepted by ICLR 2025 (poster).
2024.09: 🎉🎉 One paper “S²FT: Efficient, Scalable and Generalizable LLM Fine-tuning by Structured Sparsity” is accepted by NIPS 2024 (poster).
2024.08: 📚📚 I will start my Master in Machine Learning at Carnegie Mellon University.

📖 Educations

2024.08 - Present, M.S. in Machine Learning, Carnegie Mellon University, USA
2020.08 - 2024.05, B.S. in Computer Science, University of Rochester, USA

📝 Publications (* denotes equal contribution)

2025

Reliable and Responsible Foundation Models

Xinyu Yang , Junlin Han , Rishi Bommasani , Jinqi Luo , Wenjie Qu , Wangchunshu Zhou , Adel Bibi , Xiyao Wang

+ 44 more
Jaehong Yoon , Elias Stengel-Eskin , Shengbang Tong , Lingfeng Shen , Rafael Rafailov , Runjia Li , Zhaoyang Wang , Yiyang Zhou , Chenhang Cui , Yu Wang , Wenhao Zheng , Huichi Zhou , Jindong Gu , Zhaorun Chen , Peng Xia , Tony Lee , Thomas P Zollo , Vikash Sehwag , Jixuan Leng , Jiuhai Chen , Yuxin Wen , Huan Zhang , Zhun Deng , Linjun Zhang , Pavel Izmailov , Pang Wei Koh , Yulia Tsvetkov , Andrew Gordon Wilson , Jiaheng Zhang , James Zou , Cihang Xie , Hao Wang , Philip Torr , Julian McAuley , David Alvarez-Melis , Florian Tramèr , Kaidi Xu , Suman Jana , Chris Callison-Burch , Rene Vidal , Filippos Kokkinos , Mohit Bansal , Beidi Chen , Huaxiu Yao
(full list hidden)

Transactions on Machine Learning Research (TMLR) 2025 | [ OpenReview Preview ]
POSS: Position Specialist Generates Better Draft for Speculative Decoding

Langlin Huang , Chengsong Huang , Jixuan Leng , Di Huang , Jiaxin Huang

Preprint 2025 | [ arXiv Code Preview ]
Semi-structured LLM Reasoners Can Be Rigorously Audited

Jixuan Leng , Cassandra A. Cohen , Zhixian Zhang , Chenyan Xiong , William W. Cohen

Preprint 2025 | [ arXiv Code Preview ]
CrossWordBench: Evaluating the Reasoning Capabilities of LLMs and LVLMs with Controllable Puzzle Generation

Jixuan Leng , Chengsong Huang , Langlin Huang , Bill Yuchen Lin , William W. Cohen , Haohan Wang , Jiaxin Huang

Second Conference on Language Modeling (COLM) 2025 | [ arXiv Code Dataset Preview ]
Efficient Test-Time Scaling via Self-Calibration

Chengsong Huang , Langlin Huang , Jixuan Leng , Jiacheng Liu , Jiaxin Huang

Preprint 2025 | [ arXiv Code Preview ]
Taming Overconfidence in LLMs: Reward Calibration in RLHF

Jixuan Leng , Chengsong Huang , Banghua Zhu , Jiaxin Huang

International Conference on Learning Representations (ICLR) 2025 | [ arXiv Code OpenReview Preview ]

2024

S²FT: Efficient, Scalable and Generalizable LLM Fine-tuning by Structured Sparsity

Xinyu Yang , Jixuan Leng , Geyang Guo , Jiawei Zhao , Ryumei Nakada , Linjun Zhang , Huaxiu Yao , Beidi Chen

Conference on Neural Information Processing Systems (NeurIPS) 2024 | [ arXiv Code OpenReview Preview ]
Development of UroSAM: A Machine Learning Model to Automatically Identify Kidney Stone Composition from Endoscopic Video

Jixuan Leng , Junfei Liu , Galen Cheng , Haohan Wang , Scott Quarrier , Jiebo Luo , Rajat Jain

Journal of Endourology 2024 | [ Website ]

Introduction: Chemical composition analysis is important in prevention counseling for kidney stone disease. Advances in laser technology have made dusting techniques more prevalent, but this offers no consistent way to collect enough material to send for chemical analysis, leading many to forgo this test. We developed a novel machine learning (ML) model to effectively assess stone composition based on intraoperative endoscopic video data. Methods: Two endourologists performed ureteroscopy for kidney stones ≥ 10 mm. Representative videos were recorded intraoperatively. Individual frames were extracted from the videos, and the stone was outlined by human tracing. An ML model, UroSAM, was built and trained to automatically identify kidney stones in the images and predict the majority stone composition as follows: calcium oxalate monohydrate (COM), dihydrate (COD), calcium phosphate (CAP), or uric acid (UA). UroSAM was built on top of the publicly available Segment Anything Model (SAM) and incorporated a U-Net convolutional neural network (CNN). Discussion: A total of 78 ureteroscopy videos were collected; 50 were used for the model after exclusions (32 COM, 8 COD, 8 CAP, 2 UA). The ML model segmented the images with 94.77% precision. Dice coefficient (0.9135) and Intersection over Union (0.8496) confirmed good segmentation performance of the ML model. A video-wise evaluation demonstrated 60% correct classification of stone composition. Subgroup analysis showed correct classification in 84.4% of COM videos. A post hoc adaptive threshold technique was used to mitigate biasing of the model toward COM because of data imbalance; this improved the overall correct classification to 62% while improving the classification of COD, CAP, and UA videos. Conclusions: This study demonstrates the effective development of UroSAM, an ML model that precisely identifies kidney stones from natural endoscopic video data. More high-quality video data will improve the performance of the model in classifying the majority stone composition.
Choosing Wisely and Learning Deeply: Selective Cross-Modality Distillation via CLIP for Domain Generalization

Jixuan Leng , Yijiang Li , Haohan Wang

Transactions on Machine Learning Research (TMLR) 2024 | [ arXiv Code OpenReview ]

🔬 Research Experience

2025.06 - Present, Student Researcher, Google Research, advised by Dr. Si Si.
2025.01 - 2025.05, Independent Study, Carnegie Mellon University, advised by Prof. William W. Cohen.
2024.06 - Present, Visiting Researcher (Onsite during summer), Washington University in St. Louis, advised by Prof. Jiaxin Huang.
2023.12 - 2024.05, Honor Independent Study, VIStA (Visual Intelligence & Social Multimedia Analytics), advised by Prof. Jiebo Luo and Dr. Rajat K. Jain.
2022.12 - 2024.05, Research Internship (Remote), Dream Lab iSchool UIUC, advised by Prof. Haohan Wang

📝 Service

2025.09 - Present, ICLR reviewer.
2024.10 - Present, TMLR reviewer.
2022.01 - 2023.05, Teaching Assistant for CSC261/461: Database System, University of Rochester.

Jixuan Leng

🔥 News

📖 Educations

📝 Publications (* denotes equal contribution)

🔬 Research Experience

📝 Service

Contact Information

Send Message