app.py

import asyncio
from chroma import ChromaClient, TransformersEmbeddingFunction

async def main():

  # Chroma client
  client = ChromaClient()

  # Connect to backend
  
  # BERT embedding 
  embedder = TransformersEmbeddingFunction('bert-base-uncased')

  # Create collection
  collection = await client.createCollection({'name':'my_collection', 
                                              'embeddingFunction': embedder})

  # Load document
  with open('Caribbean-Artificial-Intelligence-Policy-Roadmap.txt') as f:
    doc = f.read()

  # Tokenize 
  sentences = sent_tokenize(doc)

  # Generate embeddings
  embeddings = []
  for sentence in sentences:
    embeddings.append(embedder.encode(sentence))

  # Add to Chroma
  await collection.add({'ids': list(range(len(sentences))),
                        'embeddings': embeddings,
                        'documents': sentences})
  
  print("Documents indexed successfully!")

asyncio.run(main())