Skip to main content

Upload vectors to the blockchain

In this lesson, you will upload vector embeddings along with full movie metadata to your Chromia chain, making the data searchable on-chain through semantic queries.

Run the script​

Navigate to the python/ folder and execute the following command:

python upload_vectors.py

This script will:

  • Load data from data/movie_vectors.jsonl
  • Send batches of movie vectors and metadata to the blockchain using the add_movie operation
  • Display progress updates as it uploads each movie

How it works?​

The script processes movies in batches (default: 50 per batch) to maintain transaction efficiency and prevent overload.

Each movie uploads with a single blockchain operation that includes:

  • The vector embedding generated from the plot
  • Full movie metadata such as title, release date, box office earnings, and more

The script signs and submits each batch to your Chromia chain.

Retry logic​

If the script encounters an error during a batch upload, it will automatically retry up to 5 times, applying exponential backoff (2s, 4s, 8s, etc.). This strategy effectively manages network issues and temporary transaction failures.

Output and confirmation​

As the script runs, you will see live progress in the terminal, indicating how many movies have successfully uploaded.

Once the process concludes, your on-chain database will be fully populated and ready for semantic search.

What’s next?​

In the next module, you will run actual queries and discover the possibilities of on-chain semantic search.