Angshul/SparseGeometricRAG
SparseGeometricRAG CPU-first sparse geometric retrieval for practical top-10 RAG No transformer inference at retrieval time. No retrieval GPU requirement. No dense document-vector dot products. No external API. SparseGeometricRAG is a retrieval system built around one systems objective: make the retrieval layer cheap enough to run on ordinary multicore CPU hardware without turning the corpus into a dense embedding database. It uses sparse TF-IDF geometry, fuzzy… See the full description on the dataset page: https://huggingface.co/datasets/Angshul/SparseGeometricRAG.
0331
1[build-system]2requires = ["setuptools>=69", "wheel"]3build-backend = "setuptools.build_meta"4 5[project]6name = "geomretrieval"7version = "0.2.0"8description = "CPU-first sparse geometric retrieval for top-10 RAG with fuzzy routing, signed residuals, and branch-aware chunk selection."9readme = "README.md"10requires-python = ">=3.10"11authors = [{name = "Research package"}]12dependencies = [13 "numpy>=1.26",14 "scipy>=1.11",15 "scikit-learn>=1.4",16 "joblib>=1.3"17]18 19[project.optional-dependencies]20ann = ["faiss-cpu>=1.8", "hnswlib>=0.8"]21dev = ["pytest>=8"]22 23[project.scripts]24geomretrieval = "geomretrieval.cli:main"25 26[tool.setuptools.packages.find]27where = ["."]28include = ["geomretrieval*"]29