Embeddings

These are the currently available Basilica embeddings, as well as the embeddings that are scheduled for development. You can sign up to be notified when a new embedding is released, or suggest an embedding by emailing us.

Available

Images

You can embed images using the endpoint api.basilica.ai/embed/images/generic, or using Connection.embed_images in the Python client. This embedding will produce usable results for most images, and works best for photographic images of everyday objects.

See Documentation

Generic Text

You can embed short snippets of English language text using the endpoint api.basilica.ai/embed/text/english, or using Connection.embed_sentences in the Python client. This embedding will produce usable results for most snippets of English language text, and works best for sentence-length snippets of published English text.

See Documentation

Product Reviews

An text embedding specialized for product reviews. You can embed short snippets of text using the endpoint api.basilica.ai/embed/text/product-reviews, or using Connection.embed_sentences with the optarg model='product-reviews' in the Python client. This embedding works best for product reviews written in English, but generalizes well to other semi-formal writing domains.

See Documentation

Email

An text embedding specialized for emails. You can embed short snippets of text using the endpoint api.basilica.ai/embed/text/email, or using Connection.embed_sentences with the optarg model='email' in the Python client. This embedding works best for sentences and paragraphs taken from English-language emails.

See Documentation

Reddit

An text embedding specialized for Reddit comments. You can embed short snippets of text using the endpoint api.basilica.ai/embed/text/reddit, or using Connection.embed_sentences with the optarg model='reddit' in the Python client. This embedding works best for Reddit posts, but generalizes well to other message board text.

See Documentation

Twitter

An text embedding specialized for tweets. You can embed short snippets of text using the endpoint api.basilica.ai/embed/text/twitter, or using Connection.embed_sentences with the optarg model='twitter' in the Python client. This embedding works best for tweets, but generalizes well to other short-form informal data like text messages.

See Documentation

In Development

Websites

We plan to offer embeddings of websites that capture a mix of textual content, visual content, and code structure.

Specialized Images

We plan to offer embeddings for specialized image types, such as human faces or line art. If you have a particular type of specialized image you would be interested in better embeddings for, please email us.

Specialized Text

We plan to offer embeddings for specialized types of text, such as twitter messages or longform text. If you have a particular type of specialized text you would be interested in better embeddings for, please email us.

PDFs

We plan to offer embeddings of PDFs that capture a mix of textual content, visual content, and metadata.

Audio

We plan to offer embeddings of Audio files.

Video

We plan to offer embeddings of Video files.

Get Notified About New Embeddings

Want to suggest an embedding not listed here? Email us.