from huggingface_hub import InferenceClient import gradio as gr import requests client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1") manual_url = "https://script.google.com/macros/s/AKfycbzlFFMNw0Zc7IKt4Sn7KB5qggPMlh_2mdMn5uOjw93MF2L-58SPb4ZPUQKHJppLSLBi/exec" manual_content = None def get_manual_content(): global manual_content response = requests.get(manual_url) if response.status_code == 200: manual_content = response.text def format_prompt(message, history): global manual_content if manual_content is None or not history or not any(user_prompt for user_prompt, _ in history): get_manual_content() prompt = "" if manual_content: prompt += f"[INST]Leggi questo MANUALE dopo ti farò delle domande: [/INST] {manual_content}" for user_prompt, bot_response in history: prompt += f"[INST] {user_prompt} [/INST]" prompt += f" {bot_response} " prompt += f"[INST]Tu sei Zucchetti AI, il mio assistente personale di AdHoc e... io ho un problema in AdHoc, rispondi proponendo SOLO con soluzione relativa al problema a partire dal manuale: {message} [/INST]" return prompt def generate( prompt, history, temperature=0.2, max_new_tokens=300, top_p=0.95, repetition_penalty=1.0, ): temperature = float(temperature) if temperature < 1e-2: temperature = 1e-2 top_p = float(top_p) generate_kwargs = dict( temperature=temperature, max_new_tokens=max_new_tokens, top_p=top_p, repetition_penalty=repetition_penalty, do_sample=True, seed=42, ) formatted_prompt = format_prompt(prompt, history) stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False) output = "" for response in stream: output += response.token.text yield output return output mychatbot = gr.Chatbot( avatar_images=["./user.png", "./LogoAHR.jpg"], bubble_full_width=False, show_label=False, show_copy_button=True, likeable=True,value=[[None, "Ciao sono Zucchetti AI il tuo assistente personale per AdHoc Revolution…"]], ) demo = gr.ChatInterface(fn=generate, chatbot=mychatbot, title="Zucchetti AI", textbox=gr.Textbox(placeholder="Descrivi il tuo problema..."), theme="gradio/base", submit_btn="Invia", retry_btn=None, undo_btn=None, clear_btn="Cancella" ) demo.queue().launch(show_api=True, share=True)