search

chengzeyi / stable-fast

Best inference performance optimization framework for HuggingFace Diffusers on NVIDIA GPUs.
MIT License
1.16k stars 71 forks source link

Is this package compatible with Lycoris LoRAs #78

Open JohnnyRacer opened 9 months ago

JohnnyRacer commented 9 months ago

Hello, I see that in the read me that stable-fast supports LoRAs out of the box. Does anyone know if it's possible to use Lycoris LoRAs with this package? The Nvidia TensorRT extension for A1111 seems have questionable support for them.

chengzeyi commented 9 months ago

Hello, I see that in the read me that stable-fast supports LoRAs out of the box. Does anyone know if it's possible to use Lycoris LoRAs with this package? The Nvidia TensorRT extension for A1111 seems have questionable support for them.

Can you share your desired use case and script with me so that I can test if it works?

Luke2642 commented 8 months ago

Reading up, it's a whole family of algorithms similar to lora:

https://github.com/KohakuBlueleaf/LyCORIS/blob/main/docs/Algo-List.md

The diffusers framework mentions partial support for some of them which means they probably already work?

https://huggingface.co/docs/diffusers/v0.25.1/en/using-diffusers/loading_adapters#kohya-and-thelastben

"LyCORIS checkpoints aren’t fully supported. The load_lora_weights() method loads LyCORIS checkpoints with LoRA and LoCon modules, but Hada and LoKR are not supported."

chengzeyi commented 7 months ago

Reading up, it's a whole family of algorithms similar to lora:

https://github.com/KohakuBlueleaf/LyCORIS/blob/main/docs/Algo-List.md

The diffusers framework mentions partial support for some of them which means they probably already work?

https://huggingface.co/docs/diffusers/v0.25.1/en/using-diffusers/loading_adapters#kohya-and-thelastben

"LyCORIS checkpoints aren’t fully supported. The load_lora_weights() method loads LyCORIS checkpoints with LoRA and LoCon modules, but Hada and LoKR are not supported."

It should.