Inconsistency Between VAE Types in Pretrained Model and single_inference.py

[jimmysue]

I downloaded a pretrained model from Hugging Face with the intention of using single_inference.py for sampling. However, I noticed a discrepancy between the VAE types: the code in single_inference.py uses AllegroAutoencoderKL3D, while the config.json of the model specifies AutoencoderKLAllegro. How should I resolve this inconsistency?