Update Gradio app with multiple files

app.py

@@ -7,7 +7,7 @@ with gr.Blocks(title="AI Text-to-Image Generator", theme=gr.themes.Soft()) as de
     # AI Text-to-Image Generator
     [Built with anycoder](https://huggingface.co/spaces/akhaliq/anycoder)
 
-    Generate images from text prompts using Stable Diffusion.
+    Generate images from text prompts using FLUX (fast generation).
     """)
     
     with gr.Row():

config.py

@@ -1,7 +1,7 @@
 # Configuration constants
 DEFAULT_PROMPT = "A beautiful landscape"
-DEFAULT_NEGATIVE_PROMPT = "blurry, low quality"
+DEFAULT_NEGATIVE_PROMPT = ""
 DEFAULT_STEPS = 20
 DEFAULT_GUIDANCE = 7.5
-IMAGE_HEIGHT = 512
-IMAGE_WIDTH = 512
+IMAGE_HEIGHT = 1024
+IMAGE_WIDTH = 1024

models.py

@@ -1,40 +1,43 @@
 import torch
-from diffusers import StableDiffusionPipeline
+from diffusers import DiffusionPipeline
 import spaces
 
 # Configuration
-MODEL_ID = "CompVis/stable-diffusion-v1-4"  # Can be changed to other SD models
+MODEL_ID = 'black-forest-labs/FLUX.1-dev'
 DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
 
-# Load pipeline
-pipe = StableDiffusionPipeline.from_pretrained(MODEL_ID, torch_dtype=torch.float16)
+# Set dtype based on device for compatibility
+dtype = torch.bfloat16 if DEVICE == "cuda" else torch.float32
+
+# Load pipeline with appropriate dtype for device
+pipe = DiffusionPipeline.from_pretrained(MODEL_ID, dtype=dtype)
 pipe.to(DEVICE)
 
-# AoT Compilation for faster inference
+# AoT Compilation for faster inference (requires GPU)
 @spaces.GPU(duration=1500)
 def compile_transformer():
-    with spaces.aoti_capture(pipe.unet):
-        pipe("test prompt", num_inference_steps=1)
+    with spaces.aoti_capture(pipe.transformer) as call:
+        pipe("test prompt")
     
     exported = torch.export.export(
-        pipe.unet,
+        pipe.transformer,
         args=call.args,
         kwargs=call.kwargs,
     )
     return spaces.aoti_compile(exported)
 
 # Apply compiled model
-compiled_unet = compile_transformer()
-spaces.aoti_apply(compiled_unet, pipe.unet)
+compiled_transformer = compile_transformer()
+spaces.aoti_apply(compiled_transformer, pipe.transformer)
 
 @spaces.GPU
 def generate_image(prompt, negative_prompt="", num_inference_steps=20, guidance_scale=7.5):
     """
-    Generate an image from text prompt using Stable Diffusion.
+    Generate an image from text prompt using FLUX.
     
     Args:
         prompt (str): The text prompt for image generation.
-        negative_prompt (str): Negative prompt to avoid certain elements.
+        negative_prompt (str): Negative prompt (not used in FLUX).
         num_inference_steps (int): Number of denoising steps.
         guidance_scale (float): Scale for classifier-free guidance.
     
@@ -44,11 +47,10 @@ def generate_image(prompt, negative_prompt="", num_inference_steps=20, guidance_
     try:
         result = pipe(
             prompt=prompt,
-            negative_prompt=negative_prompt if negative_prompt else None,
             num_inference_steps=int(num_inference_steps),
             guidance_scale=float(guidance_scale),
-            height=512,
-            width=512
+            height=1024,
+            width=1024
         )
         return result.images[0]
     except Exception as e:

requirements.txt

@@ -1,7 +1,14 @@
-gradio>=4.0.0
-torch>=2.0.0
-diffusers>=0.25.0
-transformers>=4.30.0
-accelerate>=0.25.0
-safetensors>=0.4.0
-spaces
+gradio
+spaces
+git+https://github.com/huggingface/diffusers
+git+https://github.com/huggingface/transformers
+torch
+torchvision
+accelerate
+tokenizers
+sentencepiece
+Pillow
+requests
+numpy
+scipy
+matplotlib

utils.py

@@ -13,4 +13,21 @@ def preprocess_inputs(prompt, negative_prompt, steps, guidance):
     negative_prompt = negative_prompt.strip() if negative_prompt else ""
     steps = max(1, min(50, int(steps)))
     guidance = max(1.0, min(20.0, float(guidance)))
-    return prompt, negative_prompt, steps, guidance
+    return prompt, negative_prompt, steps, guidance
+
+=== gradio>=4.0.0
+torch>=2.0.0
+diffusers>=0.27.0
+transformers>=4.36.0
+accelerate>=0.25.0
+safetensors>=0.4.0
+spaces
+torchao>=0.4.0 ===
+gradio>=4.0.0
+torch>=2.0.0
+diffusers>=0.27.0
+transformers>=4.36.0
+accelerate>=0.25.0
+safetensors>=0.4.0
+spaces
+torchao>=0.4.0