app.py

import spaces
import json
import subprocess
from llama_cpp import Llama
from llama_cpp_agent import LlamaCppAgent, MessagesFormatterType
from llama_cpp_agent.providers import LlamaCppPythonProvider
from llama_cpp_agent.chat_history import BasicChatHistory
from llama_cpp_agent.chat_history.messages import Roles
import gradio as gr
from huggingface_hub import hf_hub_download
import torch
from diffusers import FluxPipeline
import numpy as np
import random

# Download models
hf_hub_download(
    repo_id="bartowski/gemma-2-9b-it-GGUF",
    filename="gemma-2-9b-it-Q5_K_M.gguf",
    local_dir="./models"
)

# Initialize global variables
llm = None
llm_model = None

# Set up image generation
dtype = torch.bfloat16
device = "cuda" if torch.cuda.is_available() else "cpu"
pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=dtype)
pipe.enable_model_cpu_offload()  # Remove this line if you have enough GPU power
MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 2048

@spaces.GPU(duration=120)
def respond_and_generate_image(
    message,
    history,
    model,
    system_message,
    max_tokens,
    temperature,
    top_p,
    top_k,
    repeat_penalty,
    seed=42,
    randomize_seed=False,
    width=1024,
    height=1024,
    num_inference_steps=4,
):
    global llm, llm_model
    
    # Chatbot response
    chat_template = MessagesFormatterType.GEMMA_2
    
    if llm is None or llm_model != model:
        llm = Llama(
            model_path=f"models/{model}",
            flash_attn=True,
            n_gpu_layers=81,
            n_batch=1024,
            n_ctx=8192,
        )
        llm_model = model

    provider = LlamaCppPythonProvider(llm)

    agent = LlamaCppAgent(
        provider,
        system_prompt=f"{system_message}\nYou can also describe images for generation.",
        predefined_messages_formatter_type=chat_template,
        debug_output=True
    )
    
    settings = provider.get_provider_default_settings()
    settings.temperature = temperature
    settings.top_k = top_k
    settings.top_p = top_p
    settings.max_tokens = max_tokens
    settings.repeat_penalty = repeat_penalty
    settings.stream = True

    messages = BasicChatHistory()

    for msn in history:
        user = {
            'role': Roles.user,
            'content': msn[0]
        }
        assistant = {
            'role': Roles.assistant,
            'content': msn[1]
        }
        messages.add_message(user)
        messages.add_message(assistant)
    
    stream = agent.get_chat_response(
        message,
        llm_sampling_settings=settings,
        chat_history=messages,
        returns_streaming_generator=True,
        print_output=False
    )
    
    response = ""
    for output in stream:
        response += output
    
    # Image generation
    if randomize_seed:
        seed = random.randint(0, MAX_SEED)
    generator = torch.Generator("cpu").manual_seed(seed)
    
    # Truncate the prompt if it's too long
    max_length = 77  # CLIP's maximum token length
    truncated_response = ' '.join(response.split()[:max_length])
    
    try:
        image = pipe(
            prompt=truncated_response,
            guidance_scale=0.0,
            output_type="pil",
            num_inference_steps=num_inference_steps,
            max_sequence_length=256,
            generator=generator,
            width=width,
            height=height
        ).images[0]
    except Exception as e:
        print(f"Error generating image: {e}")
        image = None
    
    history.append((message, response))
    
    return history, image, seed, history  # Return history twice to match the expected output count

description = """
<p><center>
<a href="https://huggingface.co/google/gemma-2-9b-it" target="_blank">[9B it Model]</a>
<a href="https://huggingface.co/bartowski/gemma-2-9b-it-GGUF" target="_blank">[9B it Model GGUF]</a>
<a href="https://huggingface.co/black-forest-labs/FLUX.1-schnell" target="_blank">[FLUX.1-shnell Model]</a>
</center></p>
"""

demo = gr.Blocks()

with demo:
    gr.Markdown("# Chat with Gemma 2 and Generate Images with FLUX.1")
    gr.Markdown(description)

    with gr.Row():
        with gr.Column(scale=1):
            chatbot = gr.Chatbot()
            msg = gr.Textbox(label="Message")
            clear = gr.Button("Clear")

        with gr.Column(scale=1):
            generated_image = gr.Image(label="Generated Image")
            used_seed = gr.Number(label="Seed Used")

    with gr.Accordion("Advanced Settings", open=False):
        model = gr.Dropdown([
            'gemma-2-9b-it-Q5_K_M.gguf',
        ],
        value="gemma-2-9b-it-Q5_K_M.gguf",
        label="Model"
        )
        system_message = gr.Textbox(value="You are a helpful assistant.", label="System message")
        max_tokens = gr.Slider(minimum=1, maximum=4096, value=2048, step=1, label="Max tokens")
        temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
        top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p")
        top_k = gr.Slider(minimum=0, maximum=100, value=40, step=1, label="Top-k")
        repeat_penalty = gr.Slider(minimum=0.0, maximum=2.0, value=1.1, step=0.1, label="Repetition penalty")
        seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=42)
        randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
        width = gr.Slider(label="Width", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=1024)
        height = gr.Slider(label="Height", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=1024)
        num_inference_steps = gr.Slider(label="Number of inference steps", minimum=1, maximum=50, step=1, value=4)

    state = gr.State([])

    msg.submit(respond_and_generate_image, 
               inputs=[msg, state, model, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty, seed, randomize_seed, width, height, num_inference_steps], 
               outputs=[chatbot, generated_image, used_seed, state])
    clear.click(lambda: ([], None, None, []), outputs=[chatbot, generated_image, used_seed, state])

if __name__ == "__main__":
    demo.launch()