minor: EVA-L

baaivision · Dec 18, 2022 · e5e77f1 · e5e77f1
1 parent 9533b37
commit e5e77f1
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/eva/README.md b/eva/README.md
@@ -126,7 +126,7 @@ encoder-decoder architecture ([`modeling_mae_pretrain.py`](./modeling_mae_pretra
 
 </div>
 
-**EVA-L** can reach up to **89.2** top-1 accuracy on ImageNet-1K. To our knowledge, EVA-L is the best open-sourced large-sized vision encoder to date.
+**EVA-L** can reach up to **89.2** top-1 accuracy on ImageNet-1K, which is very similar to the fine-tuned EVA-CLIP teacher (89.4 top-1 accuracy). To our knowledge, EVA-L is the best open-sourced large-sized vision encoder to date.
 
 
 ## Setup