See Further When Clear: Adaptive Generative Modeling with Curriculum Consistency Model

Yunpeng Liu, Boxiao Liu, Yi Zhang, Xingzhong Hou, Guanglu Song, Yu Liu, Haihang You

Keywords: adaptive curriculum learning, noise schedule, flow matching, consistency models

Abstract:

Significant advances have been made in the sampling efficiency of diffusion models, driven by Consistency Distillation (CD), which trains a student model to mimic the output of a teacher model at an earlier timestep. However, we found that the learning complexity of the student model varies significantly across different timesteps, leading to suboptimal performance in consistency models.
To address this issue, we propose the Curriculum Consistency Model (CCM), which stabilizes and balances the learning complexity across timesteps. We define the distillation process as a curriculum and introduce Peak Signal-to-Noise Ratio (PSNR) as a metric to quantify the difficulty of each step in this curriculum.
By incorporating adversarial losses, our method achieves competitive single-step sampling Fréchet Inception Distance (FID) scores of 1.64 on CIFAR-10 and 2.18 on ImageNet 64x64.
Moreover, our approach generalizes well to both Flow Matching models and diffusion models. We have extended our method to large-scale text-to-image models, including Stable Diffusion XL and Stable Diffusion 3.

Primary Area: generative models

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 2623