Continuous Semi-Implicit Models

Longlin Yu, Jiajun Zha, Tong Yang, Tianyu Xie, Xiangyu Zhang, S.-H. Chan, Cheng Zhang
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:73375-73400, 2025.

Abstract

Semi-implicit distributions have shown great promise in variational inference and generative modeling. Hierarchical semi-implicit models, which stack multiple semi-implicit layers, enhance the expressiveness of semi-implicit distributions and can be used to accelerate diffusion models given pretrained score networks. However, their sequential training often suffers from slow convergence. In this paper, we introduce CoSIM, a continuous semi-implicit model that extends hierarchical semi-implicit models into a continuous framework. By incorporating a continuous transition kernel, CoSIM enables efficient, simulation-free training. Furthermore, we show that CoSIM achieves consistency with a carefully designed transition kernel, offering a novel approach for multistep distillation of generative models at the distributional level. Extensive experiments on image generation demonstrate that CoSIM performs on par or better than existing diffusion model acceleration methods, achieving superior performance on FD-DINOv2.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-yu25v, title = {Continuous Semi-Implicit Models}, author = {Yu, Longlin and Zha, Jiajun and Yang, Tong and Xie, Tianyu and Zhang, Xiangyu and Chan, S.-H. and Zhang, Cheng}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {73375--73400}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/yu25v/yu25v.pdf}, url = {https://proceedings.mlr.press/v267/yu25v.html}, abstract = {Semi-implicit distributions have shown great promise in variational inference and generative modeling. Hierarchical semi-implicit models, which stack multiple semi-implicit layers, enhance the expressiveness of semi-implicit distributions and can be used to accelerate diffusion models given pretrained score networks. However, their sequential training often suffers from slow convergence. In this paper, we introduce CoSIM, a continuous semi-implicit model that extends hierarchical semi-implicit models into a continuous framework. By incorporating a continuous transition kernel, CoSIM enables efficient, simulation-free training. Furthermore, we show that CoSIM achieves consistency with a carefully designed transition kernel, offering a novel approach for multistep distillation of generative models at the distributional level. Extensive experiments on image generation demonstrate that CoSIM performs on par or better than existing diffusion model acceleration methods, achieving superior performance on FD-DINOv2.} }
Endnote
%0 Conference Paper %T Continuous Semi-Implicit Models %A Longlin Yu %A Jiajun Zha %A Tong Yang %A Tianyu Xie %A Xiangyu Zhang %A S.-H. Chan %A Cheng Zhang %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-yu25v %I PMLR %P 73375--73400 %U https://proceedings.mlr.press/v267/yu25v.html %V 267 %X Semi-implicit distributions have shown great promise in variational inference and generative modeling. Hierarchical semi-implicit models, which stack multiple semi-implicit layers, enhance the expressiveness of semi-implicit distributions and can be used to accelerate diffusion models given pretrained score networks. However, their sequential training often suffers from slow convergence. In this paper, we introduce CoSIM, a continuous semi-implicit model that extends hierarchical semi-implicit models into a continuous framework. By incorporating a continuous transition kernel, CoSIM enables efficient, simulation-free training. Furthermore, we show that CoSIM achieves consistency with a carefully designed transition kernel, offering a novel approach for multistep distillation of generative models at the distributional level. Extensive experiments on image generation demonstrate that CoSIM performs on par or better than existing diffusion model acceleration methods, achieving superior performance on FD-DINOv2.
APA
Yu, L., Zha, J., Yang, T., Xie, T., Zhang, X., Chan, S. & Zhang, C.. (2025). Continuous Semi-Implicit Models. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:73375-73400 Available from https://proceedings.mlr.press/v267/yu25v.html.

Related Material