Siqi Kou（寇思麒）

👋 Hi there! I’m a 2nd-year Ph.D. student at Shanghai Jiao Tong University (SJTU), where I am fortunate to be advised by Prof. Zhijie Deng. My research focuses on developing machine learning methodologies to build more reliable and efficient generative models.

Previously, I received my B.S. in Mathematics from SJTU, where I was fortunate to work closely with Teng Hu and advised by Prof. Ran Yi. In Fall 2023, I also had the opportunity to study as an exchange student at the Chinese University of Hong Kong (CUHK).

Currently, my research interests lie in:

Multimodal Content Generation: Developing native multimodal foundation models for seamless text, image, and video generation; focusing on pre-training architectures and RL post-training to maximize cross-modal synergies.

Efficient LLM/MLLM Inference: Designing algorithms that improve the inference efficiency of generative models; including optimizing both training and inference mechanisms.

Selected Publications [More]

* denotes equal contribution.

Think-Then-Generate: Reasoning-Aware Text-to-Image Diffusion with LLM Encoders
Siqi Kou*, Jiachun Jin*, Zetong Zhou*, Ye Ma, Yugang Wang, Quan Chen, Peng Jiang, Xiao Yang, Jun Zhu, Kai Yu, Zhijie Deng
International Conference on Machine Learning (ICML), 2026. [Paper] | [Code]

Fast and Accurate Causal Parallel Decoding using Jacobi Forcing
Lanxiang Hu*, Siqi Kou*, Yichao Fu, Samyam Rajbhandari, Tajana Rosing, Yuxiong He, Zhijie Deng, Hao Zhang
International Conference on Machine Learning (ICML), 2026. [Paper] | [Code]

Which Data Attributes Stimulate Math and Code Reasoning? An Investigation via Influence Functions
Siqi Kou, Qingyuan Tian, Hanwen Xu, Zihao Zeng, Zhijie Deng
Conference on Neural Information Processing Systems (NeurIPS), 2025. [Paper]

Orthus: Autoregressive Interleaved Image-Text Generation with Modality-Specific Heads
Siqi Kou, Jiachun Jin, Zhihong Liu, Chang Liu, Ye Ma, Jian Jia, Quan Chen, Peng Jiang, Zhijie Deng
International Conference on Machine Learning (ICML), 2025. [Paper] | [Code]

CLLMs: Consistency Large Language Models
Siqi Kou*, Lanxiang Hu*, Zhezhi He, Zhijie Deng, Hao Zhang
International Conference on Machine Learning (ICML), 2024. [Paper] | [Code]

BayesDiff: Estimating Pixel-wise Uncertainty in Diffusion via Bayesian Inference
Siqi Kou, Lei Gan, Dequan Wang, Chongxuan Li, Zhijie Deng
International Conference on Learning Representations (ICLR), 2024. [Paper] | [Code]

Academic Services

Invited talks: Multimodal Learning and Alignment, AITIME Forum, Fall 2025.

TA for: CS3308-02 Machine Learning, instructed by Prof. Zhijie Deng, Shanghai Jiao Tong University.

Reviewers for: ICML 2024-; ICLR 2025-; NeurIPS 2025; CVPR 2025-.