v0.1.8
feyninc/mandexv0.1.8Mar 23, 2026by chonknick
AI Summary
Overhauls reranking performance by switching to a single pass rerank, utilizing optimized ONNX Runtime (ORT) format, and enabling multi-threaded inference.
Key Highlights
- Reranking performance overhaul resulting in 3.3x faster speeds
- Single rerank pass across packages eliminates redundant ONNX session loads
- Adoption of ORT format for faster model deserialization
- Multi-threaded inference support using half available cores (capped at 4)
New Features
- Single rerank pass
- ORT format support
- Multi-threaded inference
Full Release Notes
## What's Changed ### Reranking Performance Overhaul (3.3x faster) - **Single rerank pass** — collect all FTS5 results across packages, rerank once instead of per-package (eliminates redundant ONNX session loads) - **ORT format** — pre-optimized flatbuffer model (`reranker.ort`) for faster deserialization, with automatic fallback to `.onnx` - **mmap model loading** — zero-copy memory-mapped file access via `commit_from_memory_directly` - **Multi-threaded inference** — uses half available cores (capped at 4) for ONNX intra-op parallelism ### Benchmarks | Scenario | v0.1.7 | v0.1.8 | Speedup | |----------|--------|--------|---------| | All packages (11) | 1.01s | **0.31s** | 3.3x | | Single package | 0.09s | **0.04s** | 2.3x |