Jingzhe Shi

Hi! I'm Jingzhe Shi (史景喆), previously an undergraduate student majoring Computer Science and Technology at IIIS, Tsinghua University (a.k.a Yao Class, directed by the Turing Award Laureate Andrew Chi-Chih Yao).

I am privileged to have the opportunity to collaborate with and learn from great PhD researchers and professors. Recently, I am honored to be advised and to be collaborating with Jiahao Qiu (PhD student advised by Prof. Mengdi Wang at Princeton) and Wanjia Zhao (PhD student advised by Prof. James Zou at Stanford). Previously, I am privileged to be mentored by Prof. Xiaolong Wang at UCSD and Prof. Hang Zhao at Tsinghua.

I am also privileged to meet with a group of talented friends when attending Physics Olympiads, and we founded CPHOS to provide Physics Olympiad simulations for high school contestants for free through an online platform. With some talented friends I met at CPHOS, we also conducted some interesting researches related to it, including the CHOPS project.

I am lucky to have collaborated with Qinwei Ma, my high school & university classmate; and with Doctor Lei Li, currently a Post-doc at UW (we met at CPHOS).

I enjoyed learning Physics and contributing to Physics Olympiads. I won a gold medal at IPhO 2021 (ranking 10th globally), and I was an invited online marker for IPhO 2022.

📬 Email | 📑 CV | 🎓 Google Scholar | 💻 Github

News

2025-10:✨Our paper PRISM-Physics: Causal DAG-Based Process Evaluation for Physics Reasoning is now publicly available.

2025-09:✨Our paper Physics Supernova: AI Agent Matches Elite Gold Medalists at IPhO 2025 was accepted by NeurIPS 2025 LLM Eval workshop (oral) and NeurIPS 2025 LAW workshop.

2025-02: Our paper Explaining Context Length Scaling and Bounds for Language Models is now available on arXiv.

2024-09: Our paper Scaling Law for Time Series Forecasting was accepted by NeurIPS 2024.

2024-07: Our paper CHOPS was accepted by COLM 2024.

Personal Interest

I am interested in improving Physics Olympiads and doing DL/ML researches.

After attending CPhO in high school and winning a gold medal at IPhO, I have been committed to popularizing educational resources for Physics Olympiads, thus I have worked hard to found CPHOS.
Amazed by the current progress of artifitial intelligence I learnt in college, I have been involved in many researches related to DL/ML.

Research Interest

My research interests lie broadly in Deep Learning and Machine Learning.

For Applications of DL/ML, I am interested in derivatives of LLMs in all aspects, including Multi-agent researches and Multimodality LMs.

For Physics of DL/ML and DL/ML for physics, I am interested in Physics of Large Neural Networks, e.g. Scaling Law and its explanation in various areas and scalable models. Also, I believe physics-related problems (e.g. physics olympiads theory & experiment tests) are great benchmarks for evaluating reasoning, physics-understanding and real-world-modeling abilities of AI systems.

Selected Publications & Preprints (in time order)

(* for equal contribution, ^ for equal correspondence)

	PRISM-Physics: Causal DAG-Based Process Evaluation for Physics Reasoning Wanjia Zhao, Qinwei Ma, Jingzhe Shi, Shirley Wu, Jiaqi Han, Yijia Xiao, Si-Yuan Chen, Xiao Luo, Ludwig Schmidt James Zou Preprint* Project Page / arXiv We proposed Prism-Physics, a Physics Olympiad Problem dataset with DAG-based scoring policy and rule-based equation equivalence comparison, to evaluate LLMs' physics reasoning abilities in a fine-grained process-score-based manner. Our scoring framework and method are designed to be adaptable to other datasets that feature math reasoning process.
	Physics Supernova: AI Agent Matches Elite Gold Medalists at IPhO 2025 Jiahao Qiu, Jingzhe Shi, Xinzhe Juan, Zelin Zhao, Jiayi Geng, Shilong Liu, Hongru Wang, Sanfeng Wu, Mengdi Wang NeurIPS 2025 LLM Eval Workshop (Oral), NeurIPS 2025 LAW Workshop Code / arXiv We proposed an agent-based framework for solving International Physics Olympiad Problems. We conducted experiments on IPhO 2025 Theory Problems, validating the effectiveness of agent-based approaches.
	Explaining Context Length Scaling and Bounds for Language Models Jingzhe Shi, Qinwei Ma, Hongyi Liu, Hang Zhao^, Jeng-Neng Hwang, Serge Belongie, Lei Li^ arXiv 2025 Code / arXiv We explain context length scaling from Intrinsic Space perspective, with theoretical assumptions and deductions validated by experiments on Natural Language and Synthetic Dataset. We find that Intrinsic Entropy, a metric measured from middle-states of LLMs, shows linear relationship to next token prediction loss, which could potentially be an interesting phenomenon to explore, and potentially implies the successes of sparse representations.
	Scaling Law for Time Series Forecasting Jingzhe Shi, Qinwei Ma, Huan Ma, Lei Li NeurIPS 2024 Code / arXiv / OpenReview We proposed a theoretical framework for Scaling Law for Time Series Forecasting, taking into account look back horizon as well as dataset size and model size. We conducted experiments to validate our theory proposed and assumptions made. Our key theoretical and experimental findings were that optimal look back horizon does exist and it increases with dataset size, calling for a more fair comparison when proposing new time series forecasting models.
	CHOPS: CHat with custOmer Profile Systems for Customer Service with LLMs Jingzhe Shi, Jialuo Li, Qinwei Ma, Zaiwen Yang, Huan Ma, Lei Li COLM 2024 Code / arXiv / OpenReview We proposed CHOPS, an LLM agent designed to efficiently access user information, interact with existing systems, and provided accurate, safe responses by leveraging a combination of small and large LLMs. Validated using the CPHOS-dataset we proposed in the same work, CHOPS demonstrated its potential to enhance or replace human customer service.
	Large Trajectory Models are Scalable Motion Predictors and Planners Qiao Sun, Shiduo Zhang, Danjiao Ma, Jingzhe Shi, Derun Li, Simian Luo, Yu Wang, Ningyi Xu, Guangzhi Cao, Hang Zhao arXiv 2023 Code / arXiv We leveraged successful backbones in NLP for trajectory prediction, demonstrating scalability on diverse datasets and achieving state-of-the-art performance on Nuplan dataset. I was responsible for the decoder part. I ustilized DDPM to generate trajectory in Key Point Space to capture multi-modal distribution of future trajectories.

Social Work Experience

CPHOS
2020.12 - Present
Co-founder, Former Techgroup Leader, Council Member

CPHOS is an academical non-profit organization dedicated to providing Physics Olympiad simulations for high school contestants for free through an online platform.

CPHOS was founded in the late 2020 by a group of 10 (including myself), now it has 100+ members. 1000+ students from 200+ high schools participate in most Olympiads held by CPHOS.

I led the tech group to develope tools supporting online Olympiads, as well as conducting interesting researches including the CHOPS project.

Optiver
2024.07-2024.08
Quantitative Trading Intern, at Optiver Shanghai Office

Optiver is a leading global market making firm.

Education Experience

	UCSD 2024.02 - 2024.06 Visiting researcher Research Advisor: Prof. Xiaolong Wang.
	Tsinghua University 2021.09 - Present Undergraduate Student Research Advisor: Prof. Hang Zhao.

Representative Honors and Awards

(complete list can be found in my CV)

2023&2024: Technological Innovation Scholarship of Tsinghua University.

2021: First-Class Freshmen Scholarship of Tsinghua University.

2021: Gold Medalist 🏅 in the 51st International Physics Olympiad (IPhO 2021), ranking tenth globally.

2021: One of the top 5 students selected as national team member in a series of Domestic Physics Olympiads in China (i.e., CPhO), from an estimated 660k contestants (of the first round of CPhO). During this I have won a series of National level and Provincial level prizes in CPhO.

Language and Skills

Language: Chinese (Native), English (TOEFL: 113; R30,L28,S26,W29; test taken in Nov. 2024.), Japanese (daily dialogue).

Programming languages: Python, C/C++, etc.

Tools: Git, LaTeX, SQL, etc.

Service

ICLR 2025 & 2026, COLM 2025, NeurIPS 2025 Reviewer.

IPhO 2022 (the 52nd International Physics Olympiad) Marker. In that year IPhO was held in Switzerland, but due to pandemic IPhO had to invite extra markers. I was invited and fulfilled my job as an online marker to mark, discuss with my marker partner and to do rebuttal with team leaders from countries and regions all around the world through online meetings. I feel super lucky to have contributed to this top-tier Physics Olympiad as marker just one year after I attended it as contestant and won a gold medal.

This homepage is designed based on Jon Barron's homepage and deployed on GitHub Pages. Last updated: Apr, 2025.
© 2025 Jingzhe Shi