app.py

import gradio as gr
from diffusers import DiffusionPipeline

# Initialize the pipeline variable globally
pipeline = None

# Load the pipeline and LoRA weights
def load_cust(modelsyu, progress=gr.Progress()):
    progress(10.4, desc="loading your lora models...")
    global pipeline
    pipeline = DiffusionPipeline.from_pretrained(modelsyu)
    # Ensure the pipeline uses the CPU
    pipeline.to("cpu")
    return "Models loaded successfully"

def generate_image(prompt, negative_prompt):
    global pipeline
    # Generate the image with the provided prompts
    if pipeline is None:
        return "Pipeline not loaded. Please load the models first."
    # Make sure the pipeline is set to use the CPU
    pipeline.to("cpu")
    image = pipeline(prompt, negative_prompt=negative_prompt).images[0]
    return image

# Define the Gradio interface
with gr.Blocks() as demo:
    gr.Markdown("# Text to Image Generation Custom models Demo")
    prompt = gr.Textbox(label="Prompt", placeholder="Enter your text prompt here")
    negative_prompt = gr.Textbox(label="Negative Prompt", placeholder="Enter your negative prompt here")
    submit_button = gr.Button("Generate Image")
    with gr.Accordion('Load your custom models first'):
        basem = gr.Textbox(label="Your Lora model")  
        exports = gr.Button("Load your models")
        outputid = gr.Textbox(label="Output", interactive=False)
        exports.click(load_cust, inputs=[basem], outputs=[outputid])
    output_image = gr.Image(label="Generated Image")
    submit_button.click(generate_image, inputs=[prompt, negative_prompt], outputs=output_image)

# Launch the demo
demo.launch()