endpoints.py

from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
import torch
import os
import requests
# from pydantic import BaseModel

# from langchain.llms.huggingface_pipeline import HuggingFacePipeline

key = os.environ.get("huggingface_key")
openai_api_key = os.environ.get("openai_key")
app = FastAPI(openapi_url="/api/v1/sparrow-data/openapi.json", docs_url="/api/v1/sparrow-data/docs")

app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_methods=["*"],
    allow_headers=["*"],
    allow_credentials=True,
)
# API_URL = "https://api-inference.huggingface.co/models/mistralai/Mistral-7B-v0.1"
# headers = {"Authorization": f"Bearer {key}"}

# def query(payload):
# 	response = requests.post(API_URL, headers=headers, json=payload)
# 	return response.json()

# from fastapi import FastAPI, HTTPException, Body
# from transformers import GPT2LMHeadModel, GPT2Tokenizer


# model = GPT2LMHeadModel.from_pretrained("EleutherAI/gpt-neo-2.7B")
# tokenizer = GPT2Tokenizer.from_pretrained("EleutherAI/gpt-neo-2.7B")


tokenizer = AutoTokenizer.from_pretrained("WizardLM/WizardCoder-1B-V1.0")
model = AutoModelForCausalLM.from_pretrained("WizardLM/WizardCoder-1B-V1.0")

pipe = pipeline("text-generation",
                model=model,
                tokenizer=tokenizer,
                max_length=4000,
                do_sample=True,
                top_p=0.95,
                repetition_penalty=1.2,
               )
# hf_llm = HuggingFacePipeline(pipeline=pipe)


# class ChatRequest(BaseModel):
#     messages: list
#     temperature: float = 1.0
#     max_tokens: int = 50
#     stream: bool = False

# class ChatResponse(BaseModel):
#     response: str

# @app.post("/v1/chat/completions", response_model=ChatResponse)
# async def chat_completions(request: ChatRequest):
#     try:
#         # Prepare input prompt
#         input_prompt = ""
#         for message in request.messages:
#             role = message.get('role', 'user')
#             content = message.get('content', '')
#             input_prompt += f"{role}: {content}\n"

#         # Tokenize and generate response
#         input_ids = tokenizer.encode(input_prompt, return_tensors='pt')
#         output = model.generate(input_ids, max_length=1024, temperature=request.temperature, max_tokens=request.max_tokens)

#         # Decode and send response
#         response = tokenizer.decode(output[0], skip_special_tokens=True)
#         return {"response": response}

#     except Exception as e:
#         raise HTTPException(status_code=500, detail=str(e))


@app.get("/")
def root():
    return {"message": "R&D LLM API"}
# @app.get("/get")
# def get():
#     result = pipe("name 5 programming languages",do_sample=False)
#     print(result)
#     return {"message": result}


async def askLLM(prompt):
    output = pipe(prompt,do_sample=False)
    return output

@app.post("/ask_llm")
async def ask_llm_endpoint(prompt: str):
    # result = await askLLM(prompt)
    result = pipe(prompt,do_sample=False)
    return {"result": result}


# @app.post("/ask_HFAPI")
# def ask_HFAPI_endpoint(prompt: str):
#     result = query(prompt)
#     return {"result": result}
    
# from langchain.llms import OpenAI

# llm = OpenAI(model_name="text-davinci-003", temperature=0.5, openai_api_key=openai_api_key)

# @app.post("/ask_GPT")
# def ask_GPT_endpoint(prompt: str):
#     result = llm(prompt)
#     return {"result": result}