{"repo":"schwabauerbriantomas-gif/m2m-vector-search","free":true,"listed":false,"github":"https://github.com/schwabauerbriantomas-gif/m2m-vector-search","clone":"git clone https://github.com/schwabauerbriantomas-gif/m2m-vector-search.git","description":"Edge Vector search engine with Vulkan GPU acceleration, hierarchical indexing (HRM2), and native LangChain integration. Gaussian splat-based architecture for similarity search on resource-constrained devices.","language":"Python","stars":24,"topics":["edge-computing","embeddings","gaussian-splatting","gpu-computing","langchain","llamaindex","local-first","python","rag","semantic-search"],"license":"AGPL-3.0","category":"ai-agents","readme_excerpt":"M2M Vector Search Machine-to-Memory — A vector search engine with probabilistic Gaussian Splats, online learning via feedback, energy-based uncertainty quantification, and multi-backend GPU acceleration. Quick Start • Features • Architecture • Semantic Memory • Benchmarks • 📊 Presentation • Changelog --- Interactive Benchmark Presentation 📊 Open the interactive report → (opens in any browser, no server needed) A NotebookLM-style visual dashboard with animated charts, QPS-vs-Recall scatter plots, latency comparisons, and the complete results table — all data-driven, zero fabricated numbers. 📷 Click to view presentation screenshots Hero & Summary Stats ANN-Benchmarks Datasets QPS vs Recall — All 3 datasets Latency Comparison (log scale) Complete Results Table Features & M2M vs FAISS When to Use What --- Features - Gaussian Splats — Each vector is represented as a learnable Gaussian: score(x, i) = αᵢ · exp(-κᵢ · ‖x − μᵢ‖²) . Three parameters (μ, κ, α) encode position, concentration, and importance independently. - Online Learning — Hebbian update rules adapt splat parameters from user feedback after each query. No retraining, no re-indexing. - Energy-Based Model — Native uncertainty quantification via an energy landscape. Every search result carries a confidence score derived from the local energy topology. - HRM2 Engine — Hierarchical Routing with Mixture Models and adaptive probing for sub-linear search at scale. - SOC Consolidation — Sel","default_branch":null,"files":null,"tree":[],"storefront":"/r/schwabauerbriantomas-gif","claimed":false,"request_supported":{"post":"https://gitbuyer.com/r/schwabauerbriantomas-gif/m2m-vector-search/request-supported","requests":0},"note":"indexed from public GitHub; nothing is for sale on this page. Clone it from GitHub. Paid listings live at /search."}