Skip to main content

​
What are Embeddings?

Embeddings are numerical vector representations of text that capture semantic meaning. When you create an embedding on a column:
  • Text is converted into high-dimensional vectors using an embedding model
  • Vectors are stored alongside your original data
  • Similar content gets similar vector representations
  • Semantic search becomes possible using natural language queries

​
Creating Embeddings

from timbal.platform.embeddings import create_embedding

# Create an embedding on a text column
await create_embedding(
    org_id="your-org-id",
    kb_id="your-kb-id",
    name="product_descriptions",  # Name for this embedding
    table_name="Products",        # Table containing the text
    column_name="description",    # Column to create embeddings for
    model="text-embedding-3-small",  # Embedding model to use
    with_gin_index=True          # Create index for better performance
)
You can create multiple embeddings on different columns or even the same column with different models.
  • Embeddings can only be created on text columns - they convert text into vector representations
  • Embedding names must be unique across your entire knowledge base - two different tables cannot have embeddings with the same name
  • Only create embeddings on columns you want to search semantically - unnecessary embeddings waste storage and processing time

​
Available Embedding Models

First, check what models are available for your organization: 
from timbal.platform.embeddings import list_embedding_models

# Get available embedding models
models = await list_embedding_models(org_id="your-org-id")

print("Available embedding models:")
for model in models:
    print(f"  - {model}")
Common embedding models include:
  • text-embedding-3-small - OpenAI’s efficient model, good for most use cases
  • text-embedding-3-large - OpenAI’s most powerful model, higher accuracy
  • text-embedding-ada-002 - OpenAI’s previous generation model
  • Additional models may be available depending on your organization setup
Check out semantic search to learn how to use embeddings for search.