3 projects
s4fs
fsspec adapter for S4 gateway-format objects (S4F2 framed + compressed) directly on the backend — no gateway required. Reads out of the box; gateway-compatible writes opt-in via write_enabled=True.
s4-codec
Python bindings for s4-codec — in-process CPU compression (zstd / gzip) for ML and ETL pipelines. GPU codecs are not exposed in Python in v1.0; route GPU workloads through the s4 server gateway.
local-llm-embed
Sentence embeddings from any local causal LLM, no fine-tuning.
