lostelf/unarxive_sparse
收藏Hugging Face2026-04-12 更新2026-04-26 收录
下载链接：
https://hf-mirror.com/datasets/lostelf/unarxive_sparse
下载链接
链接失效反馈官方服务：
资源简介：
---
pretty_name: EviGraph-R Sparse Index
task_categories:
- feature-extraction
- text-retrieval
tags:
- evigraph-r
- scientific-literature
- qdrant
- sparse-vectors
---
# EviGraph-R Sparse Index
This dataset contains the sparse retrieval index generated by the EviGraph-R indexing pipeline.
It is exported from the finalized shard records after the collection has been written to Qdrant, so the Hub copy matches the indexed corpus that was prepared for retrieval.
## What is inside
- One row per indexed chunk.
- Original chunk payload metadata used by retrieval and analysis.
- Vector columns: `sparse_indices / sparse_values`.
- Source collection: `unarxive_chunks`.
- Embedding model key: `bge-m3`.
- Runtime profile: `hpc`.
## Build summary
- Repository: `lostelf/unarxive_sparse`
- Split: `train`
- Shards exported: `15`
- Rows exported: `127353`
- Generated at: `2026-04-12T19:26:45.835499+00:00`
## Suggested use
Use this dataset as a portable snapshot of the EviGraph-R retrieval index for reproducible experiments, offline analysis, or mirroring the vector store outside Qdrant.
提供机构：
lostelf
