Create README.md

README.md

@@ -0,0 +1,7 @@
+---
+datasets:
+- imagenet-1k
+metrics:
+- accuracy
+---
+This is a linear classification head for ImageNet-1k for use with timm's vit_base_patch16_224.mae (Masked Auto Encoders). It does not contain any additional normalization or pooling layers. I trained it for 100 epochs on ImageNet-1k but still got a relatively poor performance (52.19).