Add model card and metadata for MixFlow

README.md

@@ -1,3 +1,40 @@
----
-license: mit
----
+---
+license: mit
+pipeline_tag: unconditional-image-generation
+---
+
+# MixFlow: Mixed Source Distributions Improve Rectified Flows
+
+This repository contains the weights for MixFlow, as presented in the paper [MixFlow: Mixed Source Distributions Improve Rectified Flows](https://huggingface.co/papers/2604.09181).
+
+MixFlow is a training strategy for rectified flows that reduces generative path curvature and improves sampling efficiency. It introduces **$\kappa$-FC**, a general formulation that conditions the source distribution on an arbitrary signal to align it better with the data distribution.
+
+- **GitHub Repository:** [https://github.com/NazirNayal8/MixFlow](https://github.com/NazirNayal8/MixFlow)
+- **Paper:** [https://arxiv.org/abs/2604.09181](https://arxiv.org/abs/2604.09181)
+
+## Description
+
+Diffusion models and their variations, such as rectified flows, generate diverse and high-quality images, but they are still hindered by slow iterative sampling caused by the highly curved generative paths they learn. MixFlow trains a flow model on linear mixtures of a fixed unconditional distribution and a $\kappa$-FC-based distribution. This simple mixture improves the alignment between the source and data, provides better generation quality with less required sampling steps, and accelerates the training convergence considerably.
+
+## Usage
+
+Inference is performed using the scripts provided in the official repository. For example, to run inference for one of the experiments:
+
+```bash
+bash scripts/run_inference.sh <experiment_config> /path/to/model.ckpt <output_directory>
+```
+
+Refer to the [GitHub README](https://github.com/NazirNayal8/MixFlow) for detailed instructions on environment setup and evaluation.
+
+## Citation
+
+```bibtex
+@inproceedings{
+nayal2026mixflow,
+title={MixFlow: Mixed Source Distributions Improve Rectified Flows},
+author={Nazir Nayal and Christopher Wewer and Jan Eric Lenssen},
+booktitle={ICLR 2026 2nd Workshop on Deep Generative Model in Machine Learning: Theory, Principle and Efficacy},
+year={2026},
+url={https://openreview.net/forum?id=uWktyU3OIJ}
+}
+```