Starstruk Logo MarkStarstruk

v2.1.0

ruvnet/RuVectorv2.1.0Mar 27, 2026by github-actions[bot]

AI Summary

A major release implementing state-of-the-art (SOTA) gap implementations across vector search, attention mechanisms, and graph learning. It adds 13 new modules across 3 crates with 8,577 lines of new code and 859 tests passing, featuring significant performance improvements and memory optimizations.

Key Highlights

  • Hybrid Search with Reciprocal Rank Fusion (RRF) for 20-49% retrieval improvement
  • Graph RAG with knowledge graph and Leiden community detection for 30-60% improvement
  • FlashAttention-3 reducing memory from O(N²) to O(N)
  • Multi-Head Latent Attention (MLA) with ~93% KV-cache compression
  • 859 Rust tests with zero regressions and no unsafe code

New Features

  • Hybrid Search (RRF) with sparse + dense vector fusion
  • Graph RAG with knowledge graph and community detection
  • DiskANN/Vamana for billion-scale ANN with <10ms latency
  • ColBERT Multi-Vector with per-token late interaction retrieval
  • Matryoshka Embeddings with adaptive-dimension search
  • OPQ with optimized product quantization
  • LSM Compaction for write-heavy workloads
  • FlashAttention-3 with configurable block sizes
  • Multi-Head Latent Attention (MLA)
  • KV-Cache Compression with 3-4 bit quantization
  • Selective State Space Models (Mamba)
  • Speculative Decoding for 2-3x speedup
  • GraphMAE for graph masked autoencoding
  • Complete GCloud training infrastructure
  • Release gate automation with 7 ship/no-ship criteria
  • Dataset governance with validation
  • Nightly training with incremental LoRA
  • TurboQuant sidecar for KV-cache configuration
  • Cloud Run Jobs for GPU training
  • Ablation matrix for testing

Full Release Notes

## v2.1.0 — State-of-the-Art Gap Implementations

13 new modules across 3 crates addressing gaps between RuVector and 2024-2026 research from Google, Meta, DeepSeek, and Microsoft. **8,577 lines of new code, 859 tests passing, zero regressions.**

### Highlights

#### Advanced Search & Retrieval (ruvector-core)
- **Hybrid Search (RRF)** — Sparse + dense vector fusion with Reciprocal Rank Fusion, SPLADE-compatible scoring. 20-49% retrieval improvement.
- **Graph RAG** — Knowledge graph + Leiden community detection + local/global/hybrid search. 30-60% improvement on complex multi-hop queries.
- **DiskANN / Vamana** — SSD-backed billion-scale ANN with alpha-RNG pruning and LRU page cache. <10ms latency.
- **ColBERT Multi-Vector** — Per-token late interaction retrieval with MaxSim, AvgSim, SumMax scoring.
- **Matryoshka Embeddings** — Adaptive-dimension search with funnel and cascade modes for speed with minimal recall loss.
- **OPQ** — Optimized Product Quantization with learned rotation matrix. 10-30% error reduction vs standard PQ.
- **LSM Compaction** — Log-Structured Merge-tree for write-heavy workloads with bloom filters.

#### Attention & Inference (ruvector-attention)
- **FlashAttention-3** — IO-aware tiled attention reducing memory from O(N²) to O(N). Configurable block sizes, causal masking, dropout.
- **Multi-Head Latent Attention (MLA)** — DeepSeek-V2/V3 style KV-cache compression (~93% reduction).
- **KV-Cache Compression** — 3-4 bit asymmetric per-channel quantization (TurboQuant-inspired). H2O, Sliding Window, PyramidKV eviction. 6-8x memory reduction.
- **Selective State Space Models (Mamba)** — Linear-time sequence processing with selective scan and discretization.
- **Speculative Decoding** — Draft-verify pipeline with Medusa multi-head and tree attention for 2-3x generation speedup.

#### Graph Learning (ruvector-gnn)
- **GraphMAE** — Graph Masked Autoencoder with GAT encoder, SCE loss, degree-centrality masking, re-masking regularization.

### Quality
- **859 Rust tests** — 423 (core) + 210 (attention) + 226 (gnn), all passing
- **Zero regressions** from v2.0.6
- **No unsafe code** in any new module
- Security fixes: NaN-safe sort comparisons, quantization input validation

### Published Packages

**crates.io:**
| Crate | Version |
|-------|---------|
| ruvector-core | 2.1.0 |
| ruvector-attention | 2.1.0 |
| ruvector-gnn | 2.1.0 |
| ruvector-attention-wasm | 2.1.0 |
| ruvector-gnn-wasm | 2.1.0 |
| ruvllm | 2.1.0 |

**npm:**
| Package | Version |
|---------|---------|
| ruvector | 0.2.19 |
| ruvector-wasm | 2.1.0 |
| ruvector-attention-wasm | 2.1.0 |
| ruvector-gnn-wasm | 2.1.0 |
| ruvector-attention-unified-wasm | 0.1.0 |
| @ruvector/ruvllm | 2.5.4 |

### CLI Fix
- Fixed \`npx ruvector create\` and \`benchmark\` commands — \`dimension\` → \`dimensions\` field name mismatch (#307)

### Documentation
- Updated root README with all new SOTA modules
- Updated npm README with v2.1 features and TurboQuant section
- Updated @ruvector/ruvllm README with TurboQuant KV-cache compression docs
- ADR-128: SOTA gap analysis and implementation documentation

**Full Changelog**: https://github.com/ruvnet/RuVector/compare/v2.0.6...v2.1.0

---

### Training Pipeline (ADR-129)

Added complete GCloud training infrastructure for continuous model improvement:

- **Release gate automation** — 7 ship/no-ship criteria (G1-G7) with automated checker
- **Dataset governance** — Schema validation, dedup, contamination checks, quality scoring
- **Nightly training** — Incremental LoRA from pi.ruv.io brain learnings → validate → push to HF
- **TurboQuant sidecar** — `.turboquant.json` per-layer KV-cache config profiles
- **Cloud Run Jobs** — 4 GPU jobs (calibration, SFT, benchmark, nightly) + 2 schedulers
- **Ablation matrix** — 5-run isolation testing (baseline → imatrix → SFT → DPO → TQ)

Deploy: `./scripts/training/deploy_training.sh`