Configure image dataset viewer

README.md

@@ -10,10 +10,13 @@ tags:
 - image-classification
 pretty_name: Lurcher 10x Microscopy
 configs:
-- config_name: viewer
+- config_name: images
+  drop_labels: true
   data_files:
-  - split: folds
-    path: viewer/lurcher_folds.parquet
+  - split: train
+    path:
+    - images/**/*.jpg
+    - metadata.csv
 ---
 
 # Lurcher 10x Microscopy Dataset
@@ -50,16 +53,18 @@ Each row has a path relative to the dataset repository root and a normalized cla
 
 ## Dataset Viewer
 
-The Hub dataset viewer uses:
+The Hub dataset viewer is configured as an image dataset using:
 
 ```text
-viewer/lurcher_folds.parquet
+images/**/*.jpg
+metadata.csv
 ```
 
-This table mirrors the released fold manifests and includes both `image_path` and an `image` column so the viewer can render thumbnails directly.
+`metadata.csv` links each image to its relative path, class, and case id so the viewer can render thumbnails directly while preserving the original `image_path` column. The predefined experiment folds remain in the JSONL manifests below `folds/`.
 
 ## Metadata
 
+- `metadata.csv`: one row per image for the Dataset Viewer, with `file_name`, `image_path`, class, and case id.
 - `metadata/images.csv`: one row per image with class and case id.
 - `metadata/folds.csv`: per-fold split counts and manifest file paths.