txtai-formats.py

from txtai import Embeddings

# SQLite + NumPy
embeddings = Embeddings(content=True, backend="numpy")
embeddings.index(["test"])
embeddings.save("test")

# SQLite File
# sqlite3 test/documents "SELECT * FROM sections"
# 0|0|test||2026-01-16 16:12:03.410194+00:00

# NumPy Embeddings
# python -c 'import numpy as np; print(np.load("test/embeddings")[:,:5])'
# [[ 0.01157344  0.02513618 -0.03670187  0.05932486 -0.00714904]]

# Also works with DuckDB!
embeddings = Embeddings(content="duckdb", backend="numpy")
embeddings.index(["test"])
embeddings.save("test")

# DuckDB File
# duckdb -c 'SELECT * from sections' test/documents
# ┌─────────┬─────────┬─────────┬─────────┬────────────────────────────┐
# │ indexid │   id    │  text   │  tags   │           entry            │
# │  int32  │ varchar │ varchar │ varchar │         timestamp          │
# ├─────────┼─────────┼─────────┼─────────┼────────────────────────────┤
# │    0    │ 0       │ test    │ NULL    │ 2026-01-16 11:14:29.692397 │
# └─────────┴─────────┴─────────┴─────────┴────────────────────────────┘

# Embeddings database configuration is JSON
# jq "." test/config.json
# {
#   "content": "duckdb",
#   "backend": "numpy",
#   "path": "sentence-transformers/all-MiniLM-L6-v2",
#   ...
# }