TheSauceSuite

TheSauceSuite / BM25-Turbo-Rust-Python-WASM-CLI

Public

The fastest BM25 scoring engine: 2,300x faster than BM25S. 28K QPS on 8.8M docs. 5 BM25 variants (Robertson, Lucene, ATIRE, BM25L, BM25+). Memory-mapped persistence, BMW pruning, streaming indexing. Built-in HTTP server, MCP tool, HuggingFace Hub integration. Drop-in bm25s replacement for RAG & ML pipelines.

14
1
100% credibility
Found Mar 21, 2026 at 14 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

BM25 Turbo is a blazing-fast search engine that precomputes relevance scores for documents, available as a command-line tool, Python library, WebAssembly module, and Rust crate for building high-performance retrieval applications.

How It Works

1
🔍 Discover fast search magic

You hear about BM25 Turbo, a tool that makes finding info in huge piles of documents lightning-quick, perfect for your notes or website.

2
📦 Get the toolbox

Download the simple program that works on your computer, ready to handle documents in seconds.

3
📚 Gather your writings

Collect all your articles, pages, or texts into one place so the tool can learn from them.

4
⚡ Build your super index

Feed your writings to the tool and watch it create a smart guide that remembers everything instantly.

5
🔎 Ask any question

Type what you're looking for and get the best matches ranked perfectly, super fast.

6
Share your finder
🖥️
Run a web search page

Anyone visits your site and searches your collection instantly.

🐍
Add to your programs

Embed it in scripts or apps for private super-searching.

🚀 Blazing searches forever

Now you find anything in your documents in a blink, saving hours every day.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 14 to 14 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is BM25-Turbo-Rust-Python-WASM-CLI?

BM25 Turbo delivers the fastest BM25 scoring engine in Rust, hitting 28K QPS on 8.8M docs—2,300x faster than bm25s as a drop-in replacement for RAG and ML pipelines. It supports five variants like BM25, BM25+, BM25L, and ATIRE, with BMW pruning for large corpora, memory-mapped persistence, and streaming indexing. Users get a CLI for indexing/searching/serving, Python bindings, WASM for browsers, built-in HTTP/MCP servers, and HuggingFace Hub integration.

Why is it gaining traction?

It crushes query latency (sub-ms on <100K docs) via precomputed sparse scores, no math at serve time, making it the fastest GitHub option for BM25 without full search engines like Elasticsearch. Multi-platform bindings and gRPC distributed mode handle real workloads, while CLI commands like `bm25-turbo serve --mcp` plug straight into AI agents. Benchmarks on MS MARCO prove it scales where bm25s chokes.

Who should use this?

RAG builders in Python needing lexical retrieval before LLMs. ML engineers evaluating on BEIR datasets for hybrid search. Browser devs embedding fast search via WASM, or CLI users indexing JSONL/CSV corpora for quick prototyping.

Verdict

Grab it if BM25 speed is your bottleneck—CLI and docs make onboarding fast despite 14 stars and 1.0% credibility. Maturity shows in benchmarks, but production? Run your own on 8.8M docs first.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.