app.py

from fastapi import FastAPI
from pydantic import BaseModel
from huggingface_hub import InferenceClient

app = FastAPI()

client = InferenceClient("mistralai/Mistral-7B-Instruct-v0.2")

class ChatRequest(BaseModel):
    prompt: str
    history: list

def format_prompt(message, history):
    system_prompt = "You are Mistral, a gentle and a useful AI assistant. My input is "
    prompt = "<s>"
    prompt += f"[INST] {system_prompt} [/INST]"
    for user_prompt, bot_response in history:
        prompt += f"[INST] {user_prompt} [/INST]"
        prompt += f" {bot_response}</s> "
    prompt += f"[INST] {message} [/INST]"
    return prompt

def generate(prompt, history, temperature=0.2, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0):
    temperature = float(temperature)
    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)

    response = client.text_generation(formatted_prompt, **generate_kwargs)
    return response.generated_text

@app.post("/generate/")
async def chat(request: ChatRequest):
    return {"response": generate(request.prompt, request.history)}