import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("microsoft/Phi-3.5-mini-instruct", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("microsoft/Phi-3.5-mini-instruct",
    device_map="auto",
    trust_remote_code=True)

# Function to generate text based on the prompt
def generate_text(prompt, max_length=100):
    inputs = tokenizer(prompt, return_tensors="pt")
    inputs = inputs.to(model.device)
    outputs = model.generate(**inputs, max_length=max_length)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

# Create Gradio interface
iface = gr.Interface(
    fn=generate_text,
    inputs="text",
    outputs="text",
    title="Microsoft Phi 3.5B Instruct - Text Generation"
)
iface.launch()