titans-pytorch 0.3.0__py3-none-any.whl → 0.3.2__py3-none-any.whl

titans_pytorch/memory_models.py

@@ -36,10 +36,14 @@ class MemoryMLP(Module):
     def __init__(
         self,
         dim,
-        depth
+        depth,
+        expansion_factor = 4.
     ):
         super().__init__()
-        self.weights = ParameterList([Parameter(torch.randn(dim, dim)) for _ in range(depth)])
+        dim_hidden = int(dim * expansion_factor)
+        dims = (dim, *((dim_hidden,) * (depth - 1)), dim)
+
+        self.weights = ParameterList([Parameter(torch.randn(dim_in, dim_out)) for dim_in, dim_out in zip(dims[:-1], dims[1:])])
 
         self.ln = LayerNorm(dim)
 

titans_pytorch/neural_memory.py

@@ -299,7 +299,8 @@ class NeuralMemory(Module):
         accept_weight_residual = False,
         gated_transition = False,
         default_model_kwargs: dict = dict(
-            depth = 2
+            depth = 2,
+            expansion_factor = 4.
         )
     ):
         super().__init__()
@@ -652,13 +653,14 @@ class NeuralMemory(Module):
         next_last_update = TensorDict()
         next_last_momentum = TensorDict()
 
-        for (param_name, surprise), (_, last_update), (_, last_momentum) in zip(surprises.items(), past_last_update.items(), past_last_momentum.items()):
+        for (param_name, surprise), (_, last_update) in zip(surprises.items(), past_last_update.items()):
 
             update = surprise
 
             # derive momentum with associative scan - eq (10)
 
             if has_momentum:
+                last_momentum = past_last_momentum[param_name]
                 update = self.assoc_scan(adaptive_momentum, surprise, prev = last_momentum) # momentum is S / surprise in the paper
                 momentum = update
                 next_last_momentum[param_name] = momentum[:, -1]

{titans_pytorch-0.3.0.dist-info → titans_pytorch-0.3.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: titans-pytorch
-Version: 0.3.0
+Version: 0.3.2
 Summary: Titans
 Project-URL: Homepage, https://pypi.org/project/titans-pytorch/
 Project-URL: Repository, https://github.com/lucidrains/titans-pytorch

titans_pytorch-0.3.2.dist-info/RECORD

@@ -0,0 +1,9 @@
+titans_pytorch/__init__.py,sha256=Y3m_ZlpEqYwp-Md1ARhNGJxq8bQp8ty1o039nZOOJo0,276
+titans_pytorch/associative_scan.py,sha256=Y-iYqmFuG-NoCKu6kgql1mhowXTeJfyawi3eUIXamp0,2650
+titans_pytorch/mac_transformer.py,sha256=5rO4GQxSyFWWEc3pc3xNyG0sK5EXE7MmxKI-_kEMl2M,24941
+titans_pytorch/memory_models.py,sha256=TJl7b9Rd5BP8aQXK8itap5YN3DyomUVxCRJDgPuRGBk,4843
+titans_pytorch/neural_memory.py,sha256=QiEnHnZfQ8ptuXNVy4NZf9-XMbMOl2_1PT_YIG1GQBc,27739
+titans_pytorch-0.3.2.dist-info/METADATA,sha256=Ar1OdcY09w-q3RlVKlxcgrtcVzZE6cRKqnjwQ4F-9Z8,6815
+titans_pytorch-0.3.2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+titans_pytorch-0.3.2.dist-info/licenses/LICENSE,sha256=1yCiA9b5nhslTavxPjsQAO-wpOnwJR9-l8LTVi7GJuk,1066
+titans_pytorch-0.3.2.dist-info/RECORD,,

titans_pytorch-0.3.0.dist-info/RECORD

@@ -1,9 +0,0 @@
-titans_pytorch/__init__.py,sha256=Y3m_ZlpEqYwp-Md1ARhNGJxq8bQp8ty1o039nZOOJo0,276
-titans_pytorch/associative_scan.py,sha256=Y-iYqmFuG-NoCKu6kgql1mhowXTeJfyawi3eUIXamp0,2650
-titans_pytorch/mac_transformer.py,sha256=5rO4GQxSyFWWEc3pc3xNyG0sK5EXE7MmxKI-_kEMl2M,24941
-titans_pytorch/memory_models.py,sha256=Q9SAIyAbStF5Tz0EhvRbn3yAdE3nk3xKc1ndieIe714,4671
-titans_pytorch/neural_memory.py,sha256=NICWfFySGeduH4-Wgn6UavWl18pddIL1JrtpT_dhXDw,27689
-titans_pytorch-0.3.0.dist-info/METADATA,sha256=i-tiJAahxkvzYxHOWLiwP2z8DyEPq6mFgSQ_ThvpN9A,6815
-titans_pytorch-0.3.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-titans_pytorch-0.3.0.dist-info/licenses/LICENSE,sha256=1yCiA9b5nhslTavxPjsQAO-wpOnwJR9-l8LTVi7GJuk,1066
-titans_pytorch-0.3.0.dist-info/RECORD,,