Collaborative Few-Step Distillation and Low-Bit Quantization for Wan2.2 Dual-Expert Video Diffusion Models
This work addresses the deployment challenge of large video diffusion models by enabling efficient inference with reduced steps and lower bit-widths, which is important for practical applications requiring high-quality video generation.
The paper presents a compression pipeline for Wan2.2-T2V-A14B video diffusion models that combines few-step distillation with low-bit quantization, achieving quantized models that match or surpass the full-precision baseline at 8 and 20 steps, with the 20-step setting offering the best quality-efficiency trade-off.
Large video diffusion models achieve strong visual quality but remain expensive to deploy because each sample requires many denoising steps and a large resident parameter footprint. This paper studies a deployment-oriented compression pipeline for Wan2.2-T2V-A14B by combining few-step distribution-matching distillation with low-bit quantization. The pipeline follows the model's dual-expert denoising route, calibrates the high-noise and low-noise branches separately, protects sensitive entrance layers, and uses HiF4-style low-bit representation to improve dynamic-range coverage. Quantization is calibrated on the distilled few-step student rather than on the original long-step trajectory, reducing activation-distribution mismatch during inference. The proposed co-design keeps the quantized model close to the same-step full-precision model and surpasses the original full-precision baseline at 8 and 20 steps on average. The 20-step setting gives the best quality-efficiency trade-off in the tested configurations.