ValueError: .to is not supported for 4-bit or 8-bit bitsandbytes models. Please use the model as it is, since the model has already been set to the correct devices and casted to the correct dtype.

and this is my code please fix

from fastapi import FastAPI
from pydantic import BaseModel
import torch
from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
import json
import re

app = FastAPI()

─── Configuration ─────────────────────────────────────────────────────────

MODEL_NAME = "HuggingFaceH4/zephyr-7b-alpha"
SYSTEM_PROMPT = """You are an expert Agile product coach. When asked to plan a project:

• Provide 2-3 line summary
• List Epics with descriptions
• Include 3-5 User Stories per Epic
• Recommend tech stack
• Add estimates and prioritization

Format your reply with markdown sections followed by JSON:

{"project": {"summary": "...", "epics": [...]}}```"""

# ─── Pydantic Models ───────────────────────────────────────────────────────
class ChatRequest(BaseModel):
    message: str
    history: list = []

class ChatResponse(BaseModel):
    reply: str
    history: list
    structured_data: dict = None

# ─── Model Initialization ──────────────────────────────────────────────────
quant_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_use_double_quant=True,
    bnb_4bit_compute_dtype=torch.bfloat16
)

tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
model = AutoModelForCausalLM.from_pretrained(
    MODEL_NAME,
    quantization_config=quant_config
)

pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    device=0 if torch.cuda.is_available() else -1
)

# ─── Helper Functions ──────────────────────────────────────────────────────
def extract_json(text: str) -> dict:
    match = re.search(r'```json\n(.*?)\n```', text, re.DOTALL)
    if match:
        try:
            return json.loads(match.group(1))
        except json.JSONDecodeError:
            return None
    return None

def format_messages(history: list, new_message: str) -> list:
    messages = [{"role": "system", "content": SYSTEM_PROMPT}]
    
    for i, content in enumerate(history):
        messages.append({
            "role": "user" if i % 2 == 0 else "assistant",
            "content": content
        })
    
    messages.append({"role": "user", "content": new_message})
    return messages

# ─── API Endpoint ──────────────────────────────────────────────────────────


@app
	.post("/chat", response_model=ChatResponse)
async def chat_endpoint(request: ChatRequest):
    try:
        # Format conversation history
        messages = format_messages(request.history, request.message)
        prompt = tokenizer.apply_chat_template(
            messages,
            tokenize=False,
            add_generation_prompt=True
        )
        
        # Generate response
        outputs = pipe(
            prompt,
            max_new_tokens=512,
            do_sample=True,
            temperature=0.7,
            top_p=0.9,
            pad_token_id=tokenizer.eos_token_id
        )
        
        full_response = outputs[0]["generated_text"][len(prompt):]
        structured_data = extract_json(full_response)
        
        # Clean response text
        clean_response = re.sub(r'<\|.*?\|>', '', full_response).strip()
        
        return ChatResponse(
            reply=clean_response,
            history=request.history + [request.message, clean_response],
            structured_data=structured_data
        )
    
    except Exception as e:
        return {"error": str(e)}

if __name__ == "__main__":
    import uvicorn
    uvicorn.run(app, host="0.0.0.0", port=8000)