complete the conformer, just layers of conformer blocks, to ready for…

… soundstorm

README.md

@@ -49,8 +49,39 @@ block = ConformerBlock(
 )
 
 x = torch.randn(1, 1024, 512)
+
 block(x) # (1, 1024, 512)
 ```
+
+Conformer - just multiple `ConformerBlock` from above
+
+```python
+import torch
+from conformer import Conformer
+
+conformer = Conformer(
+    dim = 512,
+    depth = 12,          # 12 blocks
+    dim_head = 64,
+    heads = 8,
+    ff_mult = 4,
+    conv_expansion_factor = 2,
+    conv_kernel_size = 31,
+    attn_dropout = 0.,
+    ff_dropout = 0.,
+    conv_dropout = 0.
+)
+
+x = torch.randn(1, 1024, 512)
+
+conformer(x) # (1, 1024, 512)
+```
+
+## Todo
+
+- [ ] switch to a better relative positional encoding. shaw's is dated
+- [ ] flash attention with a better RPE
+
 ## Citations
 
 ```bibtex

conformer/__init__.py

@@ -1 +1 @@
-from conformer.conformer import ConformerConvModule, ConformerBlock
+from conformer.conformer import ConformerConvModule, ConformerBlock, Conformer

conformer/conformer.py

@@ -85,7 +85,13 @@ def __init__(
 
         self.dropout = nn.Dropout(dropout)
 
-    def forward(self, x, context = None, mask = None, context_mask = None):
+    def forward(
+        self,
+        x,
+        context = None,
+        mask = None,
+        context_mask = None
+    ):
         n, device, h, max_pos_emb, has_context = x.shape[-2], x.device, self.heads, self.max_pos_emb, exists(context)
         context = default(context, x)
 
@@ -95,6 +101,7 @@ def forward(self, x, context = None, mask = None, context_mask = None):
         dots = einsum('b h i d, b h j d -> b h i j', q, k) * self.scale
 
         # shaw's relative positional embedding
+
         seq = torch.arange(n, device = device)
         dist = rearrange(seq, 'i -> i ()') - rearrange(seq, 'j -> () j')
         dist = dist.clamp(-max_pos_emb, max_pos_emb) + max_pos_emb
@@ -199,3 +206,41 @@ def forward(self, x, mask = None):
         x = self.ff2(x) + x
         x = self.post_norm(x)
         return x
+
+# Conformer
+
+class Conformer(nn.Module):
+    def __init__(
+        self,
+        dim,
+        *,
+        depth,
+        dim_head = 64,
+        heads = 8,
+        ff_mult = 4,
+        conv_expansion_factor = 2,
+        conv_kernel_size = 31,
+        attn_dropout = 0.,
+        ff_dropout = 0.,
+        conv_dropout = 0.
+    ):
+        super().__init__()
+        self.layers = nn.ModuleList([])
+
+        for _ in range(depth):
+            self.layers.append(ConformerBlock(
+                dim = dim,
+                dim_head = dim_head,
+                heads = heads,
+                ff_mult = ff_mult,
+                conv_expansion_factor = conv_expansion_factor,
+                conv_kernel_size = conv_kernel_size,
+
+            ))
+
+    def forward(self, x):
+
+        for block in self.layers:
+            x = block(x)
+
+        return x

setup.py

@@ -3,15 +3,20 @@
 setup(
   name = 'conformer',
   packages = find_packages(),
-  version = '0.2.5',
+  version = '0.3.0',
   license='MIT',
   description = 'The convolutional module from the Conformer paper',
   author = 'Phil Wang',
   author_email = '[email protected]',
   url = 'https://github.com/lucidrains/conformer',
-  keywords = ['transformers', 'artificial intelligence', 'transformer'],
+  keywords = [
+      'artificial intelligence',
+      'deep learning',
+      'transformers',
+      'audio'
+  ],
   install_requires=[
-      'einops',
+      'einops>=0.6.1',
       'torch'
   ],
   classifiers=[