pushing sdxl turbo inference

app.py

@@ -2,6 +2,15 @@ import gradio as gr
 import torch    
 import os
 from utils import call
+from diffusers import (
+    DDPMScheduler,
+    DDIMScheduler,
+    PNDMScheduler,
+    LMSDiscreteScheduler,
+    EulerAncestralDiscreteScheduler,
+    EulerDiscreteScheduler,
+    DPMSolverMultistepScheduler,
+)
 from diffusers.pipelines import StableDiffusionXLPipeline
 StableDiffusionXLPipeline.__call__ = call
 import os
@@ -57,7 +66,9 @@ class Demo:
         self.generating = False
         self.device = 'cuda'
         self.weight_dtype = torch.bfloat16
-        self.pipe = StableDiffusionXLPipeline.from_pretrained('stabilityai/stable-diffusion-xl-base-1.0', torch_dtype=self.weight_dtype).to(self.device)
+        model_id = "stabilityai/sdxl-turbo"
+        euler_anc = EulerAncestralDiscreteScheduler.from_pretrained(model_id, subfolder="scheduler")
+        self.pipe = StableDiffusionXLPipeline.from_pretrained(model_id, scheduler=euler_anc, torch_dtype=self.weight_dtype).to(self.device)
         self.pipe.enable_xformers_memory_efficient_attention()
         with gr.Blocks() as demo:
             self.layout()
@@ -78,7 +89,7 @@ class Demo:
 
                 with gr.Row():
 
-                    self.explain_infr = gr.Markdown(value='This is a demo of [Concept Sliders: LoRA Adaptors for Precise Control in Diffusion Models](https://sliders.baulab.info/). To try out a model that can control a particular concept, select a model and enter any prompt, choose a seed, and finally choose the SDEdit timestep for structural preservation. Higher SDEdit timesteps results in more structural change. For example, if you select the model "Surprised Look" you can generate images for the prompt "A picture of a person, realistic, 8k" and compare the slider effect to the image generated by original model.  We have also provided several other pre-fine-tuned models like "repair" sliders to repair flaws in SDXL generated images (Check out the "Pretrained Sliders" drop-down). You can also train and run your own custom sliders. Check out the "train" section for custom concept slider training.')
+                    self.explain_infr = gr.Markdown(value='This is a demo of [Concept Sliders: LoRA Adaptors for Precise Control in Diffusion Models](https://sliders.baulab.info/). To try out a model that can control a particular concept, select a model and enter any prompt, choose a seed, and finally choose the SDEdit timestep for structural preservation. Higher SDEdit timesteps results in more structural change. For example, if you select the model "Surprised Look" you can generate images for the prompt "A picture of a person, realistic, 8k" and compare the slider effect to the image generated by original model.  We have also provided several other pre-fine-tuned models like "repair" sliders to repair flaws in SDXL generated images (Check out the "Pretrained Sliders" drop-down). You can also train and run your own custom sliders. Check out the "train" section for custom concept slider training. <b>Current Inference is running on SDXL Turbo!</b>')
 
                 with gr.Row():
 
@@ -316,10 +327,10 @@ class Demo:
 
 
         generator = torch.manual_seed(seed)
-        edited_image = self.pipe(prompt, num_images_per_prompt=1, num_inference_steps=20, generator=generator, network=network, start_noise=int(start_noise), scale=float(scale), unet=unet).images[0]
+        edited_image = self.pipe(prompt, num_images_per_prompt=1, num_inference_steps=3, generator=generator, network=network, start_noise=int(start_noise), scale=float(scale), unet=unet, guidance_scale=1).images[0]
         
         generator = torch.manual_seed(seed)
-        original_image = self.pipe(prompt, num_images_per_prompt=1, num_inference_steps=20, generator=generator, network=network, start_noise=start_noise, scale=0, unet=unet).images[0]
+        original_image = self.pipe(prompt, num_images_per_prompt=1, num_inference_steps=3, generator=generator, network=network, start_noise=start_noise, scale=0, unet=unet, guidance_scale=1).images[0]
         
         del unet, network
         unet = None