fasttextembed 1.0.0

Fast, dependency-free text embeddings (BAAI/bge-small-en-v1.5) in pure C

fasttextembed (Python)

Fast, dependency-free text embeddings for BAAI/bge-small-en-v1.5, powered by a small pure-C engine (no PyTorch, no ONNX Runtime). The model (~64 MB) is downloaded and cached on first use.

pip install fasttextembed

from fasttextembed import TextEmbedding

model = TextEmbedding()                          # downloads + caches the model on first call
vectors = model.embed(["hello world", "fast"])   # list of 384-float vectors
one = model.embed_one("hello world")

• embed(texts, threads=0) — batch; threads<=0 uses all cores.
• Returns plain list[list[float]] (wrap with numpy.asarray(...) if you want an array).

Environment overrides: FTE_MODEL_DIR (use local model.fte/vocab.tsv), FTE_MODEL_URL (download base URL), FTE_CACHE (cache dir), FTE_LIB (path to the C shared library).