Why no save?

[俞航]

HI, Thanks for sharing this great quantization technique

But I am not sure I understand why saving is not supported at this moment, in the main.py

    if args.save or args.save_safetensors:
        raise NotImplementedError()

What are the obstacles that prevent compressed models from being saved/reload?

Do you know any techniques that allow dumping the model VRAM to disk and reload from disk directly?

[qwopqwop200]

#1

[俞航]

Sure, gonna stay tuned!