Add Tables 3-4, stop asserting UZLF's release status without evidence

README.md

@@ -49,7 +49,7 @@ restricted:
 
 | Source | Status |
 |---|---|
-| UZLF | KU Leuven hospital-internal fundus dataset, not publicly released |
+| UZLF | KU Leuven-associated fundus dataset; we haven't found a public release page or license terms for it |
 | GRAPE | CC0 1.0 |
 | MAGRABIA | CC BY-NC 4.0 |
 | MESSIDOR | redistribution and commercial use explicitly prohibited |
@@ -80,6 +80,12 @@ run this demo on.
 
 ![Example predictions across UZLF, LES-AV, INSPIRE, FIVES, AV-WIDE, DRIVE, and TREND](https://huggingface.co/JonathanFhima/RLAD-SwinV2-Large/resolve/main/benchmark_examples.png)
 
+## Ablations
+
+![Table 3: Impact of increasing the number of training datasets — adding newly introduced or annotated datasets to the SwinV2-tiny training pipeline](https://huggingface.co/JonathanFhima/RLAD-SwinV2-Large/resolve/main/table3_dataset_ablation.png)
+
+![Table 4: Pretraining Objective and Generation Method — impact of pretraining objectives and AV-conditioned generation on Local, External, and OOD performance](https://huggingface.co/JonathanFhima/RLAD-SwinV2-Large/resolve/main/table4_pretraining.png)
+
 ## How it works
 
 A SwinV2-Large backbone (`microsoft/swinv2-large-patch4-window12-192-22k`