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

MODEL_NAME = "google/muril-base-cased"

# load model (this downloads from huggingface model hub automatically)
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
model = AutoModel.from_pretrained(MODEL_NAME)

def analyze_text(text):
    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)

    with torch.no_grad():
        outputs = model(**inputs)

    # CLS token embedding
    embeddings = outputs.last_hidden_state[:, 0, :]
    return str(embeddings.tolist())

iface = gr.Interface(
    fn=analyze_text,
    inputs=gr.Textbox(lines=4, label="Enter Text (Hindi, English, Tamil etc)"),
    outputs=gr.Textbox(label="MuRIL Semantic Embedding"),
    title="Indian Language Political Text Analyzer (MuRIL)",
    description="This app uses Google's MuRIL model to understand Indian language political text."
)

iface.launch(server_name="0.0.0.0", server_port=7860)