datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
cvefixes-security-ir-graphrag
CVEfixes Security IR GraphRAG
This is a content-addressed, remotely routable Security IR release of the exact pinned CVEfixes snapshot. It packages the byte-identical original Parquet data together with a searchable corpus, BM25 postings, CUDA-generated vectors, typed graph nodes and edges, bounded adjacency indexes, and a source-CID-to-original-row lookup.
All entries are inert, non-authoritative evidence. Candidate and formal-logic rows cannot grant execution authority; exact… See the full description on the dataset page: https://huggingface.co/datasets/Publicus/cvefixes-security-ir-graphrag.patent-legal-ir-graphrag
Patent Legal IR GraphRAG
Single-dataset retrieval release at
justicedao/patent-legal-ir-graphrag.
Layout matches Publicus IR GraphRAG releases
(Publicus/cvefixes-security-ir-graphrag, Publicus/skillcenter-ir):
Family
Path
Notes
Corpus
data/corpus/*.parquet
dense document_index, CID keys
BM25 documents
data/bm25/documents/*.parquet
lengths + entry CID
BM25 postings
data/bm25/postings/*.parquet
sorted terms, FTS5 IDF, sparse lists
Vectors
data/vectors/*.parquet… See the full description on the dataset page: https://huggingface.co/datasets/justicedao/patent-legal-ir-graphrag.federal-register-full-graphrag-v20260810
Federal Register sparse GraphRAG (research)
Cutoff-bound corpus 1994-01-01 through 2026-08-10 from FederalRegister.gov
inventory + GovInfo HTML/XML. Thin-client layout matches
Publicus/skillcenter-ir / justicedao/patent-legal-ir-graphrag:
primary key entry_cid (CIDv1) with dense document_index 0..N-1
nested BM25 posting cells (≤4096 pointers/row, FTS5 IDF)
compact Parquet locators with CID + SHA-256
thenlper/gte-small 384-d centroid shards (≤4096 rows, ≤2 shards/centroid)… See the full description on the dataset page: https://huggingface.co/datasets/justicedao/federal-register-full-graphrag-v20260810.personamem-graphrag-memory