Bump pyyaml from 5.3.1 to 5.4

.gitignore

@@ -0,0 +1,36 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# VS Code
+.vscode/
+
+# MAC
+.DS_Store
+
+
+data/*

dataset.py

@@ -0,0 +1,11 @@
+import json
+from torch.utils.data import IterableDataset
+
+class ConceptFlowDataset(IterableDataset):
+    def __init__(self, txt_file, config):
+        self.root_dir = config.data_dir        
+        self.txt_file = txt_file
+
+    def __iter__(self):
+        f = open(f'{self.root_dir}/{self.txt_file}')
+        return map(json.loads, f)

model/central.py

@@ -134,18 +134,14 @@ class LeftMMFixed(torch.autograd.Function):
             Implementation of matrix multiplication of a Sparse Variable with a Dense Variable, returning a Dense one.
             This is added because there's no autograd for sparse yet. No gradient computed on the sparse weights.
             """
-
-            def __init__(self):
-                super(LeftMMFixed, self).__init__()
-                self.sparse_weights = None
-
-            def forward(self, sparse_weights, x):
-                if self.sparse_weights is None:
-                    self.sparse_weights = sparse_weights
-                return torch.mm(self.sparse_weights, x)
-
-            def backward(self, grad_output):
-                sparse_weights = self.sparse_weights
+            @staticmethod
+            def forward(ctx, sparse_weights, x):
+                ctx.sparse_weights = sparse_weights
+                return torch.mm(ctx.sparse_weights, x)
+
+            @staticmethod
+            def backward(ctx, grad_output):
+                sparse_weights = ctx.sparse_weights
                 return None, torch.mm(sparse_weights.t(), grad_output)
 
         I = X._indices()
@@ -156,6 +152,5 @@ def backward(self, grad_output):
         lookup = Y[I[0, :], I[2, :], :]
         X_I = torch.stack((I[0, :] * M + I[1, :], use_cuda(torch.arange(Z).type(torch.LongTensor))), 0)
         S = use_cuda(Variable(torch.sparse.FloatTensor(X_I, V, torch.Size([B * M, Z])), requires_grad=False))
-        prod_op = LeftMMFixed()
-        prod = prod_op(S, lookup)
+        prod = LeftMMFixed.apply(S, lookup)
         return prod.view(B, M, K)