Embeddings

Start here

• Semantic search with embeddings
• Lodestar.Embeddings
• Performance — Lodestar.Embeddings

Namespaces

• Loading vocabularies — Lodestar.Embeddings.Persistence — A tokenizer is only correct if its vocabulary is the model's own.
• Pooling — Lodestar.Embeddings.Pooling — A transformer returns one vector per token.
• Vector search — Lodestar.Embeddings.Search — Embedding a corpus is the expensive half and it happens once.
• Tokenization — Lodestar.Embeddings.Tokenization — A transformer does not read text; it reads token ids.