🔍 vROM.js

⏳ WASM ⏳ Embeddings ⏳ Storage

Loading WASM module...

🧩 vROM Hub

Pre-computed HNSW index packages — plug-and-play RAG cartridges for vROM.js.
Load a vROM to instantly search documentation without embedding anything locally.

📦 Official vROMs

🌐 Community vROMs

Coming soon — publish your own vROMs to the Hub!
Use the vROM Builder tab or tools/vrom_builder.py

🔨 vROM Builder

Create vROM packages directly in your browser. Paste text, upload markdown, or fetch HF docs — the builder will chunk, embed, and package an HNSW index.

1 Add Source Content

Fetches raw markdown from HF docs. You can add multiple pages.

2 Configure

3 Build vROM

Chunks sources, embeds with the loaded model, and builds an HNSW index — all in the browser. Load the model first (Engine tab).

🐍 Python CLI (for large corpora)

For building vROMs from large documentation sets, use the Python builder in tools/vrom_builder.py:

# Install dependencies pip install sentence-transformers huggingface_hub # Build a vROM from HF docs pages python tools/vrom_builder.py

Same HNSW algorithm and JSON schema — output is directly loadable by VectorDB.load(). See vrom-hf-docs for an example.