update docs

docs/datasets/index.md

@@ -22,7 +22,8 @@
 | Dataset                            | #Slides  | Slide Size                | Magnification (μm/px)  | Task                       | Cancer Type      |
 |------------------------------------|----------|---------------------------|------------------------|----------------------------|------------------|
 | [Camelyon16](camelyon16.md)        | 400      | ~100-250k x ~100-250k x 3 |  40x (0.25)            | Classification (2 classes) | Breast           |
-| [PANDA](panda.md)                  | 10,616    | ~20k x 20k x 3            |  20x (0.5)             | Classification (6 classes) | Prostate         |
+| [PANDA](panda.md)                  | 9,555    | ~20k x 20k x 3            |  20x (0.5)             | Classification (6 classes) | Prostate         |
+| [PANDASmall](panda_small.md)       | 1,909     | ~20k x 20k x 3           |  20x (0.5)             | Classification (6 classes) | Prostate         |
 
 
 ### Radiology datasets

docs/datasets/panda_small.md

@@ -0,0 +1,11 @@
+# PANDASmall
+
+A small version of the PANDA dataset for quicker benchmarking. 
+
+We generate stratified splits, using only 20% of the original dataset:
+
+| Splits   | Train       | Validation  | Test       |  
+|----------|-------------|-------------|------------|
+| #Samples | 955 (10%)   | 477 (5%)    | 477 (5%) |
+
+See [PANDA](panda.md) for the description of the the full dataset.

docs/leaderboards.md

@@ -23,7 +23,7 @@ The runs use the default setup described in the section below.
 
 *eva* trains the decoder on the "train" split and uses the "validation" split for monitoring, early stopping and checkpoint selection. Evaluation results are reported on the "test" split if available and otherwise on the "validation" split.
 
-For details on the FM-backbones and instructions to replicate the results, check out [Replicate evaluations](user-guide/advanced/replicate_evaluations.md). For information on the tasks, check out [Datasets](datasets/index.md).
+For details on the FM-backbones and instructions to replicate the results, check out [Replicate evaluations](user-guide/advanced/replicate_evaluations.md). For information on the tasks, check out [Datasets](datasets/index.md). For Camelyon16 runtime optimization we use only 1000 foreground patches per slide which impacts the performance on this benchmark accross all models. 
 
 ## Evaluation protocol
 

mkdocs.yml

@@ -93,6 +93,7 @@ nav:
       - Slide-level:
         - Camelyon16: datasets/camelyon16.md
         - PANDA: datasets/panda.md
+        - PANDASmall: datasets/panda_small.md
     - Radiology:
       - TotalSegmentator: datasets/total_segmentator.md
       - LiTS: datasets/lits.md