Skip to content

README.md

Latest commit

 

History

History
52 lines (40 loc) · 2.05 KB

README.md

File metadata and controls

52 lines (40 loc) · 2.05 KB
Raw

rupixel — Rust crates

These crates implement the PixelRAG → Rust port on the ruvector substrate. All shipped code is real — no unimplemented!() stubs.

pixelrag-core      pipeline + tile logic + ANN index adaptor (HNSW / IVF-Flat)
pixelrag-encoder   real all-MiniLM-L6-v2 embedder (WASM/CPU sidecar)
pixelrag-cli       benchmark harness (recall/ndcg/mrr + latency/build/memory)

The pixelrag-cli/sidecar/ folder holds the Node embedding sidecar (embed_sidecar.mjs + package.json); run npm install there once before the --embedder real benchmark. The visual encoder (Qwen3-VL over rendered screenshots) is roadmap, not shipped stub code.

Important: these build inside the ruvector monorepo, not standalone

The crates depend on ruvector crates via path dependencies, e.g.:

# pixelrag-core/Cargo.toml
ruvector-core  = { path = "../ruvector-core" }    # HNSW backend
ruvector-rairs = { path = "../ruvector-rairs" }   # IVF-Flat backend

So they compile when these crates sit alongside the ruvector crates — i.e. in a ruvnet/ruvector checkout (where they are also registered in the workspace members), or via the external/rupixel submodule wired into that workspace.

This directory is the source-of-record snapshot for the standalone rupixel repo / npm package. The canonical, continuously-built copy lives in the ruvector monorepo under crates/pixelrag-*. Treat the monorepo as authoritative if the two ever drift.

Build & bench (from a ruvector checkout)

cargo build -p pixelrag-core -p pixelrag-encoder -p pixelrag-render \
            -p pixelrag-serve -p pixelrag-cli

cargo run -p pixelrag-cli -- benchmark \
  --ground-truth tests/fixtures/pixelrag/ground-truth.json \
  --queries tests/fixtures/pixelrag/queries.json \
  --metrics ndcg,mrr,recall@10 \
  --index-backend hnsw      # or: ivf-flat

Current benchmarks use a synthetic embedder on a subset fixture — plumbing validation, not semantic retrieval quality. See ../docs/ADR-264-* and ../docs/BENCH.md.