app.py

import gradio as gr

import json
from sentence_transformers import SentenceTransformer, util

def predict(body):
    # Check if body is a string and try to convert it to a dictionary
    if isinstance(body, str):
        try:
            body = json.loads(body)
        except json.JSONDecodeError:
            return {"error": "Invalid input format. Please provide a dictionary or a valid JSON string."}

    # Extract text1 and text2 from the body
    t1 = body.get("text1")
    t2 = body.get("text2")

    # Check if text1 and text2 are not None
    if t1 is None or t2 is None:
        return {"error": "Missing text1 or text2 in the input."}

    # Initialize the model
    model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')

    # Compute embeddings for both texts
    embedding_1 = model.encode(t1, convert_to_tensor=True)
    embedding_2 = model.encode(t2, convert_to_tensor=True)

    # Compute the cosine similarity
    similarity_score = util.pytorch_cos_sim(embedding_1, embedding_2)

    # Return the result in the desired format
    return {"similarity score": similarity_score.item()}
iface = gr.Interface(fn=predict, inputs="text", outputs="text")
iface.launch(share=True)