import os import torch import logging from flask import Flask, request, jsonify from flask import send_file from vllm import LLM from vllm import SamplingParams from huggingface_hub import login import uuid from llama3 import Llama3 from stable_diffusion import StableDiffusion # Initialize Flask app app = Flask(__name__) # llama_instance = Llama3() sd_instance = StableDiffusion() @app.route('/chat', methods=["POST"]) def generate_text_llama(): data = request.get_json() messages = data.get('messages', []) llm = data.get('llm', 'llama3') if llm == 'llama3': response = llama_instance.chat(messages) else: return jsonify({"error": "Unsupported LLM"}), 400 return jsonify({"response": response}) @app.route('/generate_image', methods=['POST']) def generate_image(): data = request.get_json() prompt = data.get('prompt', '') num_inference_steps = data.get('num_inference_steps', 28) guidance_scale = data.get('guidance_scale', 3.5) image = sd_instance.generate_image(prompt, num_inference_steps, guidance_scale) filename = "{}.png".format(str(uuid.uuid4())) print(f"Generated image filename: {filename}") sd_instance.save_image(image, filename) return send_file(filename, mimetype='image/png') if __name__ == "__main__": print("Starting the Flask app.") app.run(host="0.0.0.0", port=5000)