Ricky Tian Qi Chen

I am a Member of Technical Staff at Prometheus AI.

Previously, I was at Meta and worked on both frontier model development and open-ended research. At TBD Lab, I worked across the stack from data to post-training for Muse Image and Muse Video using multimodal RL. At FAIR, I was researching foundations of generative modeling, investigating scaling laws and new approaches to multimodal generation [OneFlow | Edit Flows]. Among my research outputs, pre-training methods such as Flow Matching became the de facto approach for frontier media generation. Post-training methods such as Adjoint Matching have been applied to large-scale diffusion finetuning for internal generative models and scientific AI [Adjoint Sampling | ASBS].

CV | X | Github | Google Scholar | rtqichen@gmail.com

Talk Slides

  • Unlocking Discontinuities in Flow Models: Jumps, Control Flow, Insertions, Deletions, etc CVPR 2025 Workshop on Visual Generative Modeling: What's After Diffusion? 2025 slides | website
  • Stochastic Control for Large Scale Reward-Driven Generative Modeling ICLR 2025 Workshop on Deep Generative Model in Machine Learning: Theory, Principle and Efficacy 2025 slides | website
  • Flow Matching Tutorial NeurIPS 2024 2024 slides | website
  • The Flow Matching Recipe for Generative Modeling: from continuous to discrete ICML 2024 Workshop on Structured Probabilistic Inference & Generative Modeling 2024 slides | website
  • Flow Matching on General Geometries [15min version] ICLR 2024 Oral Presentation 2024 slides | website
  • Discovering Latent Dynamics of the World: A Simulation-Free Perspective Vanderbilt Machine Learning Seminar Series 2024 slides | website
  • Flow Matching on General Geometries Learning on Graphs Conference New York Meetup 2024 slides | website
  • Generative Flows: Beyond Distribution Matching Flatiron Institute Workshop on Topics of Measure Transport 2023 slides | website
  • Simulation-free generative modeling with neural ODEs ICIAM 2023 Theoretical and computational advances in measure transport 2023 slides | website
  • Simulation-free generative modeling with neural ODEs AFOSR Topics at the Intersection of Deep Learning and Computational Nonlinear Control 2023 slides | website