This release includes breaking changes for platform teams planning a safe upgrade.
✓ No known CVEs patched in this version
Topics
+14 more
Summary
AI summaryGPU‑accelerated ONNX embeddings can be enabled for up to 10× faster indexing.
Full changelog
What's New
GPU-Accelerated Embeddings (Optional)
ONNX embeddings can now run on NVIDIA GPUs for 5-10x faster indexing. Opt-in — CPU remains the default.
pip install knowledge-rag[gpu]
models:
embedding:
gpu: true
Falls back to CPU automatically if CUDA is unavailable.
README Audit
Full rewrite — removed 5 obsolete "What's New" sections, fixed stale data (format counts, cache paths, Ollama references), consolidated changelog, added supported formats table. 1462 → 1089 lines (-25%).
Supported Formats Table
12 formats documented with parser type, default status, and notes:
MD, TXT, PDF, PY, JSON, CSV, DOCX, XLSX, PPTX, IPYNB, MQH, MQ4
Install/Upgrade:
pip install --upgrade knowledge-rag
# With GPU:
pip install --upgrade knowledge-rag[gpu]
Full Changelog: https://github.com/lyonzin/knowledge-rag/compare/v3.4.3...v3.5.0
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About lyonzin/knowledge-rag
Local RAG system for Claude Code with hybrid search (BM25 + semantic), cross-encoder reranking, markdown-aware chunking, query expansion, and 12 MCP tools. Runs entirely offline with zero external servers.
Related context
Beta — feedback welcome: [email protected]