default the depth to length of computed execution order of layers

README.md

@@ -693,7 +693,7 @@ model = TransformerWrapper(
 )
 ```
 
-If you wish to do something more sophisticated, say 3 layers, with each layer recurrent 4 times before onto the next, that is possible as well.
+If you wish to do something more sophisticated, say 3 layers, with each layer recurrent 4 times before onto the next, that is possible as well. Be aware the `layers_execute_order` is 0-indexed
 
 ```python
 import torch
@@ -716,11 +716,6 @@ model = TransformerWrapper(
         )
     )
 )
-
-x = torch.randint(0, 256, (1, 1024))
-
-model(x) # (1, 1024, 20000)
-
 ```
 
 ### Understanding and Improving Transformer From a Multi-Particle Dynamic System Point of View

setup.py

@@ -3,7 +3,7 @@
 setup(
   name = 'x-transformers',
   packages = find_packages(exclude=['examples']),
-  version = '1.29.0',
+  version = '1.29.1',
   license='MIT',
   description = 'X-Transformers - Pytorch',
   author = 'Phil Wang',

x_transformers/x_transformers.py

@@ -1059,7 +1059,6 @@ def __init__(
         dim_head = attn_kwargs.get('dim_head', DEFAULT_DIM_HEAD)
 
         self.dim = dim
-        self.depth = depth
         self.causal = causal
         self.layers = ModuleList([])
 
@@ -1179,6 +1178,13 @@ def __init__(
 
         self.num_attn_layers = len(list(filter(equals('a'), layer_types)))
 
+        # validate and set the depth
+
+        depth = default(depth, len(self.layers_execute_order))
+        assert depth == len(self.layers_execute_order)
+
+        self.depth = depth
+
         # stochastic depth
 
         self.layer_dropouts = cast_tuple(layer_dropout, len(layer_types))