benjamin-paine commited on
Commit
57f7c80
1 Parent(s): a467d97

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +30 -1
README.md CHANGED
@@ -1,10 +1,39 @@
1
  ---
2
  license: openrail++
3
  ---
4
- This repository contains offset versions of https://huggingface.co/mhdang/dpo-sdxl-text2image-v1 and https://huggingface.co/mhdang/dpo-sd1.5-text2image-v1.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5
 
6
  These can be added directly to any initialized UNet to inject DPO training into it. See the code below for usage (diffusers only.)
7
 
 
 
 
 
 
 
 
 
8
  ```py
9
  from __future__ import annotations
10
 
 
1
  ---
2
  license: openrail++
3
  ---
4
+
5
+ # Contents
6
+
7
+ This repository contains:
8
+ 1. Half-Precision LoRA versions of https://huggingface.co/mhdang/dpo-sdxl-text2image-v1 and https://huggingface.co/mhdang/dpo-sd1.5-text2image-v1.
9
+ 2. Full-Precision offset versions of https://huggingface.co/mhdang/dpo-sdxl-text2image-v1 and https://huggingface.co/mhdang/dpo-sd1.5-text2image-v1.
10
+
11
+ # Creation
12
+
13
+ ## LoRA
14
+
15
+ The LoRA were created using Kohya SS.
16
+
17
+ 1.5: https://civitai.com/models/240850/sd15-direct-preference-optimization-dpo extracted from https://huggingface.co/fp16-guy/Stable-Diffusion-v1-5_fp16_cleaned/blob/main/sd_1.5.safetensors.
18
+ XL: https://civitai.com/models/238319/sd-xl-dpo-finetune-direct-preference-optimization extracted from https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/sd_xl_base_1.0_0.9vae.safetensors
19
+
20
+ ## Offsets
21
+
22
+ The offsets were calculated in Pytorch using the following formula:
23
+
24
+ 1.5: https://huggingface.co/mhdang/dpo-sd1.5-text2image-v1/blob/main/unet/diffusion_pytorch_model.safetensors - https://huggingface.co/runwayml/stable-diffusion-v1-5/blob/main/unet/diffusion_pytorch_model.bin
25
+ XL: https://huggingface.co/mhdang/dpo-sdxl-text2image-v1/blob/main/unet/diffusion_pytorch_model.safetensors - https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/unet/diffusion_pytorch_model.safetensors
26
 
27
  These can be added directly to any initialized UNet to inject DPO training into it. See the code below for usage (diffusers only.)
28
 
29
+ # License
30
+
31
+ These models are derivces from all OpenRail++ models, and are licensed under OpenRail++ themselves.
32
+
33
+ # Usage
34
+
35
+ ## Offsets
36
+
37
  ```py
38
  from __future__ import annotations
39