Uvx privacy-steward for PII removal in texts
Native PyTorch port beats the official CLI on speed and batch processing.

772 MB model runs entirely in-browser with no backend, API calls, or telemetry whatsoever.
Privacy-conscious developers, security teams evaluating AI deployment
Transformers.js · WebLLM · Off Grid
Native PyTorch port beats the official CLI on speed and batch processing.
Local WebGPU AI is cool, but JSON formatters and PDF mergers exist everywhere.
Blocks PII at push time when gitleaks alone would miss it.
Swin Transformer runs in-browser at 200ms; niche but technically impressive inference.
Explicit kernel control over TVM-style black boxes, but benchmarks show mixed wins vs Transformers.js.
Offline voice typing for Linux—finally fills a gap Windows and macOS solved years ago.