SQLite disk cache for embeddings keyed by content hash, model id and dimension, with batched misses, bit-exact float64 storage and LRU eviction
Describe in two or three sentences the concrete problem this utility solves, who hits it, and what they do today without it. Replace this paragraph when the first real feature lands.
pip install trysac-embedding-cache
# or
uv add trysac-embedding-cacheRequires Python 3.10 or newer. No runtime dependencies.
import embedding_cache
# normalize() trims surrounding whitespace and lower-cases the text.
print(embedding_cache.normalize(" Hello World ")) # -> "hello world"
# Blank input is rejected explicitly instead of returning an empty string.
try:
embedding_cache.normalize(" ")
except ValueError as exc:
print(f"rejected: {exc}")- Architecture and decisions
- Testing and security
- Project record and stack
- Changelog and contributing guide
MIT (c) 2026 Miguel Gomez