Update dataset card

README.md

@@ -1,63 +1,37 @@
 ---
-dataset_info:
-  features:
-  - name: chunk_uid
-    dtype: string
-  - name: chunk_type
-    dtype: string
-  - name: chunk_index
-    dtype: int64
-  - name: total_chunks
-    dtype: int64
-  - name: section_title
-    dtype: string
-  - name: embed_text
-    dtype: string
-  - name: spans
-    struct:
-    - name: cite_spans
-      list:
-      - name: arxiv_id
-        dtype: string
-      - name: doi
-        dtype: string
-      - name: end
-        dtype: int64
-      - name: openalex_id
-        dtype: string
-      - name: raw
-        dtype: string
-      - name: source_ref_id
-        dtype: string
-      - name: start
-        dtype: int64
-  - name: paper_doi
-    dtype: string
-  - name: paper_id_arxiv
-    dtype: string
-  - name: title
-    dtype: string
-  - name: authors
-    list: string
-  - name: categories
-    list: string
-  - name: year
-    dtype: int64
-  - name: language
-    dtype: string
-  - name: discipline
-    dtype: string
-  - name: dense_vector
-    list: float64
-  splits:
-  - name: train
-    num_bytes: 1351952458
-    num_examples: 127353
-  download_size: 1196779511
-  dataset_size: 1351952458
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
+pretty_name: EviGraph-R Dense Index
+task_categories:
+- feature-extraction
+- text-retrieval
+tags:
+- evigraph-r
+- scientific-literature
+- qdrant
+- dense-embeddings
 ---
+
+# EviGraph-R Dense Index
+
+This dataset contains the dense 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: `dense_vector`.
+- Source collection: `unarxive_chunks`.
+- Embedding model key: `bge-m3`.
+- Runtime profile: `hpc`.
+
+## Build summary
+
+- Repository: `lostelf/unarxive_dense`
+- 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.