File size: 1,507 Bytes
b1accca 84cf5d4 71b54be b1accca 71b54be b1accca 8c3bfa5 71b54be b1accca 71b54be b1accca 7e1b115 b1accca c305f1e b1accca c14b372 6325084 b1accca f6eeeb2 c305f1e 71b54be b1accca |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 |
from fastapi import FastAPI, Request
from fastapi.templating import Jinja2Templates
from huggingface_hub import InferenceClient
app = FastAPI()
templates = Jinja2Templates(directory="templates")
client = InferenceClient(
"mistralai/Mistral-7B-Instruct-v0.1"
)
async def format_prompt(message, history):
prompt = "<s>"
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
async def generate(
prompt: str,
temperature: float = 0.9,
max_new_tokens: int = 256,
top_p: float = 0.95,
repetition_penalty: float = 1.0,
):
temperature = float(temperature)
if temperature < 1e-2:
temperature = 1e-2
top_p = float(top_p)
generate_kwargs = {
"temperature": temperature,
"max_new_tokens": max_new_tokens,
"top_p": top_p,
"repetition_penalty": repetition_penalty,
"do_sample": True,
"seed": 42,
}
formatted_prompt = await format_prompt(prompt, [])
response = client.text_generation(formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=True)
return response
@app.get("/")
async def index(request: Request):
return templates.TemplateResponse("index.html", {"request": request})
@app.post("/generate/")
async def chatbot_response(prompt: str):
response = await generate(prompt)
return {"response": response}
|