app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "deepseek-ai/DeepSeek-V3"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True) # ADD trust_remote_code=True

def predict(message, history):
    prompt = tokenizer.apply_chat_template(
        [{"role": "user", "content": message}],
        tokenize=False,
        add_generation_prompt=True
    )
    inputs = tokenizer(prompt, return_tensors="pt").to("cuda") # Assuming you have CUDA available in your Space
    outputs = model.generate(**inputs, max_new_tokens=50) # Adjust max_new_tokens as needed
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response

iface = gr.ChatInterface(
    fn=predict,
    inputs=gr.Chatbox(),
    outputs=gr.Chatbot()
)

iface.launch()