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

MODEL_NAME = "mistralai/Mistral-7B-v0.3"
token = os.getenv("HF_API_TOKEN")  # This must be set in the Space's secrets

print("Loading tokenizer...")
tokenizer = AutoTokenizer.from_pretrained(
    MODEL_NAME,
    trust_remote_code=True,
    token=token,
)

print("Loading model in 4-bit...")
model = AutoModelForCausalLM.from_pretrained(
    MODEL_NAME,
    torch_dtype=torch.float16,
    device_map="auto",  # Free Space => CPU
    load_in_4bit=True,
    trust_remote_code=True,
    token=token,
)

model.eval()

def generate_text(prompt):
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_new_tokens=128,
            temperature=0.7,
            repetition_penalty=1.2,
        )
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

demo = gr.Interface(
    fn=generate_text,
    inputs=gr.Textbox(lines=3, label="Your Prompt"),
    outputs=gr.Textbox(label="Mistral 7B Response"),
    title="Mistral 7B (4-bit) Chat",
    description=(
        "A minimal Mistral 7B example on free CPU. "
        "It'll be slow and can OOM with big prompts."
    ),
)

if __name__ == "__main__":
    demo.launch()