Update README for dataset archives

README.md

@@ -2,7 +2,7 @@
 pretty_name: SmartRoute_data
 license: unknown
 task_categories:
-- retrieval
+- text-retrieval
 language:
 - en
 tags:
@@ -15,31 +15,33 @@ tags:
 
 # SmartRoute_data
 
-Anonymous dataset bundle containing selected SmartRoute-related FilterVector benchmark files.
+Compressed per-dataset tar.zst archives for SmartRoute-related FilterVector benchmark data.
 
 ## Repository Layout
 
-All dataset files are stored under `SmartRoute_data/<dataset>/` in this repository.
+Each dataset is stored at the repository root as a separate `.tar.zst` archive.
 
-Included datasets:
-- `SmartRoute_data/Amazon/`
-- `SmartRoute_data/BookReviews/`
-- `SmartRoute_data/Genome/`
-- `SmartRoute_data/Laion/`
-- `SmartRoute_data/Music/`
-- `SmartRoute_data/Reviews/`
-- `SmartRoute_data/Tiktok/`
-- `SmartRoute_data/VariousImg/`
+Included archives:
+- `Amazon.tar.zst`
+- `BookReviews.tar.zst`
+- `Genome.tar.zst`
+- `Laion.tar.zst`
+- `Music.tar.zst`
+- `Reviews.tar.zst`
+- `Tiktok.tar.zst`
+- `VariousImg.tar.zst`
 
-For each dataset directory, this repository may include:
+## Archive Contents
 
-- `*_base_labels.txt`
-- `*_base.bin`
-- `*_base.fvecs`
-- query subdirectories whose names end with `random_300`
+Each archive may include:
+
+- directories whose names end with `random_300`
+- `{dataset}_base_labels.txt`
+- `*_base.bin` and/or `*_base.fvecs`
+- pre-compressed `*.zst` variants when available
 
 ## Notes
 
-- Repository visibility at upload time: `private`
-- Files were uploaded via the Hugging Face Hub API.
-- Review the contents before making the repository public.
+- Repository visibility at upload time: `public`
+- Archives are uploaded one dataset at a time with `hf upload`.
+- Decompress with `tar --zstd -xf <dataset>.tar.zst`.