Improving Flow Matching by Aligning Flow Divergence
Authors: Yuhao Huang, Taos Transue, Shih-Hsin Wang, William M Feldman, Hong Zhang, Bao Wang
ICML 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We validate the performance of FDM across several benchmark tasks, including synthetic density estimation, trajectory sampling for dynamical systems, video generation, and DNA sequence generation. Our numerical results, presented in Section 5, show that our proposed FDM can improve likelihood estimation and enhance sample generation by a remarkable margin over CFM. |
| Researcher Affiliation | Academia | 1Department of Mathematics, University of Utah, Salt Lake City, UT, USA 2Scientific Computing and Imaging (SCI) Institute, Salt Lake City, UT, USA 3Mathematics and Computer Science Division, 240 Argonne National Laboratory, Lemont, IL, USA. |
| Pseudocode | No | The paper describes methods and theoretical foundations, but does not include any explicitly labeled pseudocode or algorithm blocks. It refers to 'algorithms' in a general sense, but no structured steps are provided. |
| Open Source Code | Yes | Code is available at Utah-Math-Data-Science. |
| Open Datasets | Yes | Our experiments utilize a numerical simulation-based dataset for density estimation and trajectory sampling, a dataset extracted from a database of human promoters (Hon et al., 2017) for DNA design, and the KTH human motion dataset (Schuldt et al., 2004) and the BAIR Robot Pushing dataset (Ebert et al., 2017) for video prediction. |
| Dataset Splits | Yes | We train the models on a training set of 32,000 trajectories computed by the ODE solver using Adam for 2,000 epochs with a batch size of 500. |
| Hardware Specification | Yes | Experiments are conducted on multiple NVIDIA RTX 3090 GPUs. |
| Software Dependencies | No | Our implementation utilizes Py Torch Lightning (Falcon, 2019) for synthetic density estimation, DNA sequence generation, and video generation, while JAX (Bradbury et al., 2018) and Tensor Flow (Abadi et al., 2016) are employed for dynamical systems-related experiments. |
| Experiment Setup | Yes | We train models using FM and FDM for 2 104 iterations using a batch size of 512. |