handler.py

from transformers import AutoModel, AutoTokenizer
import torch

class EndpointHandler():
    def __init__(self, path=""):
        # Initialize the tokenizer and model with pre-trained weights
        self.tokenizer = AutoTokenizer.from_pretrained(path)
        self.model = AutoModel.from_pretrained(path)

    def __call__(self, data):
        # Extract text input from the request data
        inputs = data['inputs']

        # Define a prompt to provide context
        prompt = "Contextual understanding of the following text, from the perspective of Chassidic philosophy: "

        # Combine prompt with the actual input
        combined_input = prompt + inputs

        # Prepare the text for the model
        encoded_input = self.tokenizer(combined_input, return_tensors='pt', padding=True, truncation=True, max_length=512)

        # Generate embeddings without updating gradients
        with torch.no_grad():
            outputs = self.model(**encoded_input)

        # Extract embeddings from the last hidden layer
        embeddings = outputs.last_hidden_state.squeeze().tolist()

        # Return the embeddings as a list (serialized format)
        return {'embeddings': embeddings}