Align image_encoder.py drop_path_rate computation with upstream

image_encoder.py

@@ -695,7 +695,7 @@ class VisionTransformer(nn.Module):
       dpr = [drop_path_rate] * depth
     else:
       dpr = [
-          drop_path_rate * i / max(depth - 1, 1) for i in range(depth)
+          x.item() for x in torch.linspace(0, drop_path_rate, depth)
       ]  # stochastic depth decay rule
 
     if ffn_layer == "mlp":