Support ResAdapter with SDXL-Lightning demo in huggingface

README.md

@@ -22,7 +22,7 @@ AutoML, ByteDance Inc.
 ![GitHub Org's stars](https://img.shields.io/github/stars/bytedance%2Fres-adapter)
 
 [![Replicate](https://img.shields.io/badge/Demo-Replicate.com-purple)](https://replicate.com/bytedance/res-adapter) 
-<!-- [![Hugging Face](https://img.shields.io/badge/Demo-%F0%9F%A4%97%20Hugging%20Face-66cdaa)](https://huggingface.co/spaces/ChenyangSi/FreeU) -->
+[![Hugging Face](https://img.shields.io/badge/Demo-%F0%9F%A4%97%20Hugging%20Face-66cdaa)](https://huggingface.co/spaces/ameerazam08/Res-Adapter-GPU-Demo)
 
 
 
@@ -37,7 +37,8 @@ The above images are generated by ResAdapter with [dreamlike-diffusion-1.0](http
 
 
 ## 🚀 Release
-- `[2024/03/12]` We support [gradio demo](https://replicate.com/bytedance/res-adapter) in `replicate.com` (by [@chenxiwh](https://github.com/chenxwh))
+- `[2024/03/12]` We support [gradio demo with SDXL-Lightning](https://replicate.com/bytedance/res-adapter) in `huggingface.com` (by [@Ameer Azam](https://github.com/AMEERAZAM08))
+- `[2024/03/12]` We support [gradio demo with SDXL-Lightning](https://replicate.com/bytedance/res-adapter) in `replicate.com` (by [@chenxiwh](https://github.com/chenxwh))
 - `[2024/03/12]` We release the [resadapter-sdv15](https://huggingface.co/jiaxiangc/res-adapter/tree/main/sd1.5) that supports `128~1024` resolution.
 - `[2024/03/05]` We release the paper about [resadapter](https://arxiv.org/abs/2403.02084) to arxiv.
 - `[2024/03/04]` We release the inference codes and [resadapter-sdv15-i](https://huggingface.co/jiaxiangc/res-adapter/tree/main/sd1.5-i) that supports `128~512`, and [resadapter-xl-i](https://huggingface.co/jiaxiangc/res-adapter/tree/main/sdxl-i) that supports `256~1024`.
@@ -110,21 +111,27 @@ images = pipe(
 save_image(images, f"resadapter_{width}.png", normalize=True, padding=0)
 ```
 
-## 🎉 Use ResAdapter in Gradio
+## 🎉 Support ResAdapter in Gradio
 
-<!-- 🔥 We are happy to provide two demos in `replicate.com` and `huggingface.com`. -->
-🔥 [Update at 2024/03/12] We are happy to provide demos in `replicate.com`.
+🔥 [Update at 2024/03/12] We are happy to provide two demos in `replicate.com` and `huggingface.com`.
 
 Relicate demo: [bytedance/res-adapter](https://replicate.com/bytedance/res-adapter), by ([@chenxiwh](https://github.com/chenxwh))
 
+Huggingface demo: [ameerazam08/Res-Adapter-GPU-Demo](https://huggingface.co/spaces/ameerazam08/Res-Adapter-GPU-Demo), (by [@Ameer Azam](https://github.com/AMEERAZAM08))
 
-<!-- 2. Huggingface demo: [jiaxiangc/ReAdapter-SDXL-Lightning]() in huggingface, which is developed by  -->
+<div align=center>
+<img src="assets/misc/replicate_demo.jpeg" width="56.4%"><img src="assets/misc/huggingface_demo.jpeg" width="43.6%">
 
-## 🎉 Use ResAdapter in ComfyUI
+Left: ResAdapter with SDXL-Lightning in `Replicate.com`, Right: ResAdapter with SDXL-Lightning in `Huggingface.com`
 
-We will support ComfyUI within a week.
+</div>
 
 
+<!-- 2. Huggingface demo: [jiaxiangc/ReAdapter-SDXL-Lightning]() in huggingface, which is developed by  -->
+
+## 🎉 Support ResAdapter in ComfyUI
+
+We will support ComfyUI within a week.
 
 
 ## ⏬ Installation

models/res_adapter/gradio-demo/app.py → app.py

@@ -7,7 +7,7 @@
 import numpy as np
 import PIL.Image
 
-import spaces
+# import spaces
 import torch
 from diffusers import (
     StableDiffusionXLPipeline,
@@ -35,7 +35,8 @@
 
 
 
-base = "stabilityai/stable-diffusion-xl-base-1.0"
+# base = "stabilityai/stable-diffusion-xl-base-1.0"
+base = "/mnt/bn/automl-aigc/chengjiaxiang/models/diffusers/dreamshaper-xl-1-0"
 repo = "ByteDance/SDXL-Lightning"
 ckpt = "sdxl_lightning_4step_unet.safetensors"  # Use the correct ckpt for your step setting!
 
@@ -66,7 +67,7 @@ def randomize_seed_fn(seed: int, randomize_seed: bool) -> int:
     return seed
 
 
-@spaces.GPU(enable_queue=True)
+# @spaces.GPU(enable_queue=True)
 def generate(
     prompt: str,
     negative_prompt: str = "",
@@ -128,6 +129,7 @@ def generate(
 
 examples = [
     "A girl smiling",
+    "A boy smiling",
     "A realistic photograph of an astronaut in a jungle, cold color palette, detailed, 8k",
 
 ]
@@ -195,14 +197,14 @@ def generate(
                 minimum=256,
                 maximum=MAX_IMAGE_SIZE,
                 step=32,
-                value=1024,
+                value=512,
             )
             height = gr.Slider(
                 label="Height",
                 minimum=256,
                 maximum=MAX_IMAGE_SIZE,
                 step=32,
-                value=1024,
+                value=512,
             )
         with gr.Row():
             guidance_scale_base = gr.Slider(