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# Repository manifest2 3- `README.md` — GitHub front page: protocol, method, speed-first tables, results, reproduction.4- `geomretrieval/` — reusable sparse geometric index plus the cleaned current top-10 RAG ranker.5- `experiments/beir/` — generic RAG runners and exact SciFact/TREC-COVID experiment-history scripts.6- `experiments/msmarco_scale/` — complete full-scale MS MARCO experimental code from the supplied scale codebase.7- `results/beir/` — exact small-dataset result JSONs, sweeps, branch-diversity and collision diagnostics.8- `results/msmarco_scale/` — full-scale MS MARCO result JSONs.9- `results/reproduced_reference/` — post-assembly checks of the cleaned runner.10- `baselines/` — local lexical reference code, full baseline policy, and MS MARCO baseline material.11- `configs/` — historical full-scale lock and current RAG-top-10 configuration.12- `docs/` — method, RAG protocol, speed/literature, baseline suite and reproducibility notes.13- `docs/images/` — result and latency figures displayed directly from the root README.14- `scripts/` — reproduction helpers.15- `manifests/` — MS MARCO corpus-shard manifest.16- `tests/` — core and RAG smoke tests.17 18No `.gitignore` and no `.github/` directory are included.19 20## Final reconstruction additions21 22- `RESULTS_FINAL_6DATASETS.md`23- `RECONSTRUCTION.md`24- `results/final_100_to_10/`25- `experiments/postbenchmark_optimizations/`26