Timestep-Aware Optimization Dynamics Analysis for Diffusion Training
Open Access
- 11 July 2026
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 276-282
Abstract
Multimodal diffusion models have become a central approach for high-quality content generation. Some recent studies explore cross-timestep gradient relations to guide improvements in diffusion training. However, commonly used full-gradient cosine similarity can be distorted by parameter heterogeneity and layer-wise scale differences, leading to misleading estimates of the coupling among timestep-specific objectives. This paper revisits cross-timestep gradient measurement and proposes a corrected protocol for multimodal diffusion training. The corrected measurement shows that gradient similarity decays rapidly with timestep distance, and gradients from distant timesteps become nearly orthogonal as training progresses. These results suggest that diffusion training consists of weakly coupled timestep-wise objectives. They further imply that directly applying history-accumulating mechanisms across timesteps may introduce interference among weakly related timestep objectives. Our study refines the empirical basis for timestep-wise optimization dynamics analysis and provides guidance for optimization design in multimodal content generation.Keywords
Funding Information
- National Natural Science Foundation of China (U2541212)
- National Natural Science Foundation of China (62302023)
This publication has 11 references indexed in Scilit:
- Decouple-Then-Merge: Finetune Diffusion Models as Multi-Task Learning2025
- Adaptive Non-uniform Timestep Sampling for Accelerating Diffusion Model Training2025
- A Closer Look at Time Steps is Worthy of Triple Speed-Up for Diffusion Model Training2025
- Efficient Diffusion Training via Min-SNR Weighting Strategy2023
- Q-Diffusion: Quantizing Diffusion Models2023
- Align Your Latents: High-Resolution Video Synthesis with Latent Diffusion Models2023
- High-Resolution Image Synthesis with Latent Diffusion Models2022
- Perception Prioritized Training of Diffusion Models2022
- Video Diffusion Models2022
- Group Normalization2018