lora-inference

lora-inference copied to clipboard

Support lora model from civitai

9

Steps: - Get link from https://civitai.com/models/4361/olivia-diffusion - Inference with https://replicate.com/cloneofsimo/lora and a style trained by https://replicate.com/cloneofsimo/lora-training Actual: Error 'NoneType' object is not iterable Expect: can use civitai

seanphan

No module named 't2i_adapters'

3

I just want to launch Kohya-ss LoRA inference on a clean GPU server. Any way i can do this?

kopyl

Update diffusers library to support more lora formats

2

- Just a small dependency update to enable use of different Lora formats in Replicate using the [0.17.0 release of diffusers library](https://github.com/huggingface/diffusers/releases/tag/v0.17.0). - Fixes #3 Thank you 🙏

pskl

Error: Rank should be the same per model

8

I use safetensors on civitai and I get an error：Rank should be the same per model. Am I using it wrong?

daibin0809

gradual degrading of imagery?

4

This issue is investigating the gradual change in generated imagery as loras are loaded/unloaded through normal use: First, the assumption: Using the same inputs (seed / prompt / lora URL...

anotherjesse

In the mean time can you at least provide a convert script to transform to your LoRA file format from regular civitai safetensors format?

dpyy

Enable Restore Face by gfpgan or codeformer

2

Output from Lora sometimes not easy to control. I believe a Face restore option via API would help much.

seanphan

should this be cog-lora similar to cog-stable-diffusion?

anotherjesse

I think 768 usually messes up with doubles 

anotherjesse