Analysis of Classifier-Free Guidance Weight Schedulers
This work provides incremental insights for researchers and practitioners in generative AI by optimizing guidance mechanisms in diffusion models.
The paper tackled the problem of improving Classifier-Free Guidance (CFG) in text-to-image diffusion models by analyzing weight schedulers, finding that simple monotonically increasing schedulers consistently enhance performance with minimal code changes.
Classifier-Free Guidance (CFG) enhances the quality and condition adherence of text-to-image diffusion models. It operates by combining the conditional and unconditional predictions using a fixed weight. However, recent works vary the weights throughout the diffusion process, reporting superior results but without providing any rationale or analysis. By conducting comprehensive experiments, this paper provides insights into CFG weight schedulers. Our findings suggest that simple, monotonically increasing weight schedulers consistently lead to improved performances, requiring merely a single line of code. In addition, more complex parametrized schedulers can be optimized for further improvement, but do not generalize across different models and tasks.