model_local.py

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_name = "microsoft/Phi-4-mini-instruct"

# Load model and tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, device_map="auto")

# Generate a response
def chat_with_model(prompt, max_new_tokens=100):
    inputs = tokenizer(prompt, return_tensors="pt").to("cpu")  # Move to GPU if available
    output = model.generate(**inputs, max_new_tokens=max_new_tokens)
    return tokenizer.decode(output[0], skip_special_tokens=True)

# Test the model
user_input = "Explain quantum computing in simple terms."
response = chat_with_model(user_input)
print("Chatbot Response:", response)