Update README.md

README.md

@@ -114,7 +114,7 @@ dataset_info:
     num_bytes: 276768171.75
     num_examples: 2754
   - name: val
-    num_bytes: 31942022.0
+    num_bytes: 31942022
     num_examples: 306
   - name: trainval
     num_bytes: 308710193.5
@@ -123,13 +123,13 @@ dataset_info:
     num_bytes: 670168354.5
     num_examples: 6084
   - name: train800
-    num_bytes: 80944215.0
+    num_bytes: 80944215
     num_examples: 800
   - name: val200
-    num_bytes: 20954456.0
+    num_bytes: 20954456
     num_examples: 200
   - name: train800val200
-    num_bytes: 101898671.0
+    num_bytes: 101898671
     num_examples: 1000
   download_size: 1491513351
   dataset_size: 1491386083.75
@@ -150,4 +150,18 @@ configs:
     path: data/val200-*
   - split: train800val200
     path: data/train800val200-*
+license: cc-by-4.0
 ---
+
+
+# VTAB Caltech101
+
+This dataset has been used for the paper [Fantastic Features and Where to Find Them: A Probing Method to combine Features from Multiple Foundation Models](https://bramtoula.github.io/combo/) (NeurIPS 2025).
+
+It reproduces the settings (splits, labels) used for the Visual Task Adaptation Benchmark (VTAB).
+- **VTAB Paper:** [A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark](https://arxiv.org/abs/1910.04867)
+- **VTAB Repository:** [google-research/task_adaptation](https://github.com/google-research/task_adaptation)
+
+Details of the original dataset:
+- **Original Citation:** Fei-Fei, Li, Robert Fergus, and Pietro Perona. "One-shot learning of object categories." IEEE Transactions on Pattern Analysis and Machine Intelligence.
+- **Caltech 101 Homepage:** [Caltech101 on CaltechDATA](https://data.caltech.edu/records/mzrjq-6wc02)