40 projects
faster-paddle
Fast CPU OCR — PaddleOCR PP-OCRv6 tiny/small/medium in Rust + ONNX Runtime, with hardware-aware scheduling and multi-image recognition.
fast-universal-sentence-encoder
Fast, dependency-light Google USE multilingual v3 sentence embeddings (pure numpy, no C++ deps)
leantext
None
leanvec-scale
None
minivectordb-ooc
An out-of-core vector database with a built-in multilingual sentence encoder. The successor to minivectordb.
cxm25
CXM25 and CXM25-LTR: BM25-inspired lexical retrieval models with no embeddings
tinyzchunk
GPU-free, tokenizer-free chunker for RAG (EN + PT-BR). Robust to PDF/OCR noise, markdown, code and CRLF. Weights fetched from HuggingFace on first use.
nanoranker
None
leangeo
None
faster-table-recognizer
Fast, dependency-light table detection (microsoft/table-transformer-detection) — a from-scratch Rust engine, drop-in replacement for the HuggingFace transformers pipeline.
faster-clip
Fast CPU inference (Rust) for CLIP embeddings, inference-free SPLADE, and BERT fill-mask — drop-in replacement for the transformers / sentence-transformers models in a RAG stack. No PyTorch, no ONNX.
fast-translate
Portable, self-contained EN<->PT-BR translation with a native Rust engine (no external binary) and baked-in models
leanvec
Testing
nanoe5
nanoE5.c - blazing-fast 4-bit CPU text embeddings (multilingual-e5-small), model bundled, OpenAI-compatible server, zero ML dependencies
leanqueue
Embedded, durable, crash-resistant message queue (Rust core + Python)
harness-exchange
Exchange agent/chat sessions between codex, opencode, and claude-code
small-hybrid-reranker
Lightweight hybrid reranker with baked-in model artifact.
lightning-disk-kv
"Disk-based redis" - Built around LMDB and Rust, with sharding for maximum throughput
llm-utility-pack
None
semantic-compressor
None
strive-ranker
None
vector-cache-lmdb
Fast LMDB-backed vector cache with Python bindings and embedding helpers
lfm-onnx-hf
Standalone LFM ONNX inference with first-run Hugging Face download and local cache
dynamic-pattern-mining
Dynamic, low-resource pattern mining with sklearn-compatible API
profile-correspondence-encoder
Scikit-learn compatible Profile Correspondence Encoder for high-cardinality categorical features
tokenlearn-static-model
Rust-accelerated static embedding model with baked-in 10k/64d/int8 weights.
rb-elastic-hash
A Rust-based implementation of Elastic Hashing (arXiv:2501.02305v2)
leankv
Testing
tiny-lfm-builtin
LiquidAI-LFM2-350M rust inference engine embedded in a python package
full-text-sparse-engine
None
distry-py
Distributed task execution framework
simple-fastapi-auth
A *heavily* opinionated library for simplifying the boilerplate of FastAPI authentication.
minivectordb-simple
A python project aimed at extracting embeddings from textual data and performing semantic search. This version supports only the basic onnx model, no bge-m3 or e5 models.
wise-telemetry
None
chatmemorydb-lmdb-onnx
None
chatmemorydb-lmdb
None
minivectordb
A python project aimed at extracting embeddings from textual data and performing semantic search.
chatmemorydb
None
webtextcrawler
Python project for crawling the web and extracting text.
text-util-en-pt
Python project for text cleaning. Some specifics for English and Portuguese languages.
