in split by rank function, cache the sizes so on backwards there is n…

…ot an extra call

setup.py

@@ -3,7 +3,7 @@
 setup(
   name = 'st-moe-pytorch',
   packages = find_packages(exclude=[]),
-  version = '0.0.23',
+  version = '0.0.24',
   license='MIT',
   description = 'ST - Mixture of Experts - Pytorch',
   author = 'Phil Wang',

st_moe_pytorch/distributed.py

@@ -85,16 +85,24 @@ def __init__(self, *, dim = 0):
     def forward(self, x, sizes = None):
         return AllGatherFunction.apply(x, self.dim, sizes)
 
-class SplitByRank(Function):
+class SplitByRankFunction(Function):
     @staticmethod
     def forward(ctx, x):
         rank = dist.get_rank()
-        return x[rank]
+        out = x[rank]
+
+        if isinstance(x, tuple):
+            sizes = tuple(map(lambda t: t.shape[0], x))
+        else:
+            sizes = (x.shape[1],) * x.shape[0]
+
+        ctx.sizes = torch.tensor(sizes, device = out.device, dtype = torch.long)
+        return out
 
     @staticmethod
     def backward(ctx, grads):
         grads = rearrange(grads, '... -> 1 ...')
-        grads = all_gather_variable_dim(grads)
+        grads = all_gather_variable_dim(grads, sizes = ctx.sizes)
         return grads
 
-split_by_rank = SplitByRank.apply
+split_by_rank = SplitByRankFunction.apply