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

# Load a smaller model that fits within 16GB RAM
model_name = "deepseek-ai/deepseek-coder-1.3b-instruct"

# Load tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_name)

# Load model in CPU-friendly format (low precision for efficiency)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.float32,  # Use float32 since CPU-only
    device_map="cpu"  # Ensure it runs only on CPU
)

# Function to generate comments
def generate_code_comments(code_snippet):
    prompt = f"### Code:\n{code_snippet}\n### Add meaningful comments to this code:\n"
    inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True, max_length=512)
    outputs = model.generate(**inputs, max_length=512)
    commented_code = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return commented_code

# Create Gradio interface
iface = gr.Interface(
    fn=generate_code_comments,
    inputs="text",
    outputs="text",
    title="AI Code Comment Generator",
    description="Enter a code snippet, and the AI will add meaningful comments.",
)

iface.launch()