allow for qk norm to be turned off for na vit nested tensor

Author: lucidrains

setup.py

@@ -6,7 +6,7 @@
 setup(
   name = 'vit-pytorch',
   packages = find_packages(exclude=['examples']),
-  version = '1.8.7',
+  version = '1.8.8',
   license='MIT',
   description = 'Vision Transformer (ViT) - Pytorch',
   long_description=long_description,

vit_pytorch/na_vit_nested_tensor.py

@@ -41,7 +41,7 @@ def FeedForward(dim, hidden_dim, dropout = 0.):
     )
 
 class Attention(Module):
-    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0.):
+    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0., qk_norm = True):
         super().__init__()
         self.norm = nn.LayerNorm(dim, bias = False)
 
@@ -56,8 +56,8 @@ def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0.):
         # in the paper, they employ qk rmsnorm, a way to stabilize attention
         # will use layernorm in place of rmsnorm, which has been shown to work in certain papers. requires l2norm on non-ragged dimension to be supported in nested tensors
 
-        self.query_norm = nn.LayerNorm(dim_head, bias = False)
-        self.key_norm = nn.LayerNorm(dim_head, bias = False)
+        self.query_norm = nn.LayerNorm(dim_head, bias = False) if qk_norm else nn.Identity()
+        self.key_norm = nn.LayerNorm(dim_head, bias = False) if qk_norm else nn.Identity()
 
         self.dropout = dropout
 
@@ -111,13 +111,13 @@ def transpose_head_seq(t):
         return self.to_out(out)
 
 class Transformer(Module):
-    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0.):
+    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0., qk_norm = True):
         super().__init__()
         self.layers = ModuleList([])
 
         for _ in range(depth):
             self.layers.append(ModuleList([
-                Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout),
+                Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout, qk_norm = qk_norm),
                 FeedForward(dim, mlp_dim, dropout = dropout)
             ]))
 
@@ -146,6 +146,7 @@ def __init__(
         dim_head = 64,
         dropout = 0.,
         emb_dropout = 0.,
+        qk_rmsnorm = True,
         token_dropout_prob: float | None = None
     ):
         super().__init__()
@@ -184,7 +185,7 @@ def __init__(
 
         self.dropout = nn.Dropout(emb_dropout)
 
-        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout)
+        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout, qk_rmsnorm)
 
         # final attention pooling queries
 

vit_pytorch/na_vit_nested_tensor_3d.py

@@ -41,7 +41,7 @@ def FeedForward(dim, hidden_dim, dropout = 0.):
     )
 
 class Attention(Module):
-    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0.):
+    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0., qk_norm = True):
         super().__init__()
         self.norm = nn.LayerNorm(dim, bias = False)
 
@@ -56,8 +56,8 @@ def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0.):
         # in the paper, they employ qk rmsnorm, a way to stabilize attention
         # will use layernorm in place of rmsnorm, which has been shown to work in certain papers. requires l2norm on non-ragged dimension to be supported in nested tensors
 
-        self.query_norm = nn.LayerNorm(dim_head, bias = False)
-        self.key_norm = nn.LayerNorm(dim_head, bias = False)
+        self.query_norm = nn.LayerNorm(dim_head, bias = False) if qk_norm else nn.Identity()
+        self.key_norm = nn.LayerNorm(dim_head, bias = False) if qk_norm else nn.Identity()
 
         self.dropout = dropout
 
@@ -123,13 +123,13 @@ def transpose_head_seq(t):
         return self.to_out(out)
 
 class Transformer(Module):
-    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0.):
+    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0., qk_norm = True):
         super().__init__()
         self.layers = ModuleList([])
 
         for _ in range(depth):
             self.layers.append(ModuleList([
-                Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout),
+                Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout, qk_norm = qk_norm),
                 FeedForward(dim, mlp_dim, dropout = dropout)
             ]))
 
@@ -161,6 +161,7 @@ def __init__(
         dropout = 0.,
         emb_dropout = 0.,
         num_registers = 4,
+        qk_rmsnorm = True,
         token_dropout_prob: float | None = None
     ):
         super().__init__()
@@ -209,7 +210,7 @@ def __init__(
 
         self.dropout = nn.Dropout(emb_dropout)
 
-        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout)
+        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout, qk_rmsnorm)
 
         # final attention pooling queries