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

# Model name
model_name = "OpenGVLab/InternVideo2_5_Chat_8B"



# Load tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)

# Detect device
device = "cuda" if torch.cuda.is_available() else "cpu"

# Load model
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    trust_remote_code=True,
    torch_dtype=torch.float16 if device == "cuda" else torch.float32,
    device_map="auto" if device == "cuda" else None
)

# Move model to device
model.to(device)

# Define inference function
def chat_with_model(prompt):
    inputs = tokenizer(prompt, return_tensors="pt").to(device)
    output = model.generate(**inputs, max_length=200)
    return tokenizer.decode(output[0], skip_special_tokens=True)

# Create Gradio UI
demo = gr.Interface(
    fn=chat_with_model,
    inputs=gr.Textbox(placeholder="Type your prompt here..."),
    outputs="text",
    title="InternVideo2.5 Chatbot",
    description="A chatbot powered by InternVideo2_5_Chat_8B.",
    theme="compact"
)

# Run the Gradio app
if __name__ == "__main__":
    demo.launch()