main.py

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

app = FastAPI()

# Use your model
client = InferenceClient("ManojINaik/codsw")

class Item(BaseModel):
    prompt: str
    history: list
    system_prompt: str
    temperature: float = 0.0
    max_new_tokens: int = 1048
    top_p: float = 0.15
    repetition_penalty: float = 1.0

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

def generate(item: Item):
    try:
        # Ensure valid temperature
        temperature = max(float(item.temperature), 1e-2)
        top_p = float(item.top_p)

        generate_kwargs = {
            "temperature": temperature,
            "max_new_tokens": item.max_new_tokens,
            "top_p": top_p,
            "repetition_penalty": item.repetition_penalty,
            "do_sample": True,
            "seed": 42,
        }

        # Format the prompt
        formatted_prompt = format_prompt(f"{item.system_prompt}, {item.prompt}", item.history)

        # Call text_generation on your model (correct argument: formatted_prompt)
        stream = client.text_generation(
            formatted_prompt,  # Use the formatted prompt directly
            **generate_kwargs,
            stream=True,
        )
        output = "".join([response.token.text for response in stream])
        return output

    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Internal error: {str(e)}")

@app.post("/generate/")
async def generate_text(item: Item):
    return {"response": generate(item)}