mit-han-lab
Adding a training pipeline for the DC-AE (Deep Compression Autoencoder train pipeline)
Hi there!
I wanted to take a moment to express my appreciation for the incredible work you've done by open-sourcing this code! Huge thanks for that!
I would like to kindly request if you could consider adding a training pipeline for the DCAE (Deep Compression Autoencoder). This addition could greatly benefit the research community and streamline the training process for many users.
Thank you once again for your hard work and dedication! 🙏
A DCAE training pipeline would be incredibly helpful for the community. Thanks for bringing this up! Looking forward to seeing it.
I think a training pipeline for the DC-AE would be super beneficial. It would really help researchers and other users get more out of this awesome project. Thanks a bunch for your hard work and for open-sourcing it!
This code supports training from scratch or fine-tuning the model, and it supports loading existing weight files without requiring conversion.
https://github.com/lavinal712/AutoencoderKL/tree/dc-ae
