Question
Do you have text encoders and vae, etc?
Heya @Nielly . I have added the VAE to that but regarding the text encoder part, I was having difficulty loading it in my current VRAM so it's difficult for me to quantize that. I will look into what I can do for that (maybe ask a friend to run my script and pass me the quantized model) or maybe rent a GPU from the cloud. Will need some time for that though : )
Great! Thanks for answering :)
Oh i forgot to mention that VAE is supposed to be safetensors file. Not PTH since thats the save for the current checkpoint. Simple mistake.
Heya @Nielly . The VAE is actually straight from the WAN repo. I did not change anything in that as there were no Linear layers to quantize. Also for the text encoder part, sorry for that. I thought I'd just borrow someone's GPU (probs a friend) and then run on it and then push the quantized model out there but no luck with that. I will just post the script I used to quantize and then maybe someone from the community with enough resources can do that for us. Will be more than happy to merge it. Again, sorry : (
No problem. You did well done. Thank you for your hard work :)