TQ-DiT: Efficient Time-Aware Quantization for Diffusion Transformers
This work addresses the problem of real-time application limitations for AI researchers and practitioners, offering an incremental improvement in model efficiency.
The paper tackles the computational inefficiency of diffusion transformers (DiTs) by proposing a quantization method with multi-region and time-grouping techniques, achieving performance close to full-precision with only a 0.29 FID increase at 8-bit and outperforming baselines at 6-bit.
Diffusion transformers (DiTs) combine transformer architectures with diffusion models. However, their computational complexity imposes significant limitations on real-time applications and sustainability of AI systems. In this study, we aim to enhance the computational efficiency through model quantization, which represents the weights and activation values with lower precision. Multi-region quantization (MRQ) is introduced to address the asymmetric distribution of network values in DiT blocks by allocating two scaling parameters to sub-regions. Additionally, time-grouping quantization (TGQ) is proposed to reduce quantization error caused by temporal variation in activations. The experimental results show that the proposed algorithm achieves performance comparable to the original full-precision model with only a 0.29 increase in FID at W8A8. Furthermore, it outperforms other baselines at W6A6, thereby confirming its suitability for low-bit quantization. These results highlight the potential of our method to enable efficient real-time generative models.