Upload README.md with huggingface_hub

README.md

@@ -20,11 +20,11 @@ configs:
 - config_name: code_image
   data_files:
   - split: train
-    path: preview/code_image/train-*
+    path: code_image/train-*
 - config_name: code_only
   data_files:
   - split: train
-    path: preview/code_only/train-*
+    path: code_only/train-*
 dataset_info:
 - config_name: code_image
   features:
@@ -50,7 +50,7 @@ dataset_info:
     dtype: image
   splits:
   - name: train
-    num_examples: 102
+    num_examples: 3600
 - config_name: code_only
   features:
   - name: id
@@ -75,7 +75,7 @@ dataset_info:
     dtype: image
   splits:
   - name: train
-    num_examples: 102
+    num_examples: 3600
 ---
 
 <p align="center">
@@ -226,7 +226,7 @@ data/
     └── GT_w_o_anno_chart/        # 1,200 unannotated charts, same layout
 ```
 
-The [Dataset Viewer](https://huggingface.co/datasets/chartanno/ChartAnno/viewer) above shows a stratified preview (2 charts × 3 levels per chart type) for each config.
+The [Dataset Viewer](https://huggingface.co/datasets/chartanno/ChartAnno/viewer) above serves the **full dataset**: all 3,600 rows per config are loadable via `load_dataset("chartanno/ChartAnno", "<config>")`. Chart images in the viewer parquet are embedded as compressed JPEG thumbnails for fast previewing; the full-resolution images are included in `chartanno.tar.gz`.
 
 ## 5. Data Fields
 
@@ -262,7 +262,7 @@ Details of the data fields are as follows:
 | `GT code` | Ground-truth annotated chart code. |
 | `GT chart` | Path to the annotated ground-truth chart image. |
 
-Image paths are relative to the repository root of the extracted `data/` folder.
+Image paths are relative to the repository root of the extracted `data/` folder. In the viewer parquet, the two image fields are embedded (compressed thumbnail) versions of these files.
 
 ## 6. Evaluation