Hypergraph checks

.github/workflows/test_tutorials.yml

@@ -71,7 +71,7 @@ jobs:
         torch-version: [2.0.1]
         include:
           - torch-version: 2.0.1
-        test-group: [1, 2, 3, 4]
+        test-group: [1, 2, 3]
 
     steps:
       - uses: actions/checkout@v3

.gitignore

@@ -16,11 +16,12 @@ coverage.xml
 **/__pycache__/**
 __pycache__/
 docs/build/
+data/
 venv_*/
 
 TopoNetX/
 topomodelx/nn/cell/attcxn_layer.py
 topomodelx/base/debug.ipynb
 tutorials/hypergraph/citeseer6cls3703.pt.1
 tutorials/hypergraph/cora
-tags
+tags

.test_durations

@@ -148,18 +148,18 @@
     "test/tutorials/test_cell.py::test_tutorial[tutorials/cell/ccxn_train.ipynb]": 10.362705206964165,
     "test/tutorials/test_cell.py::test_tutorial[tutorials/cell/cwn_train.ipynb]": 10.345093041891232,
     "test/tutorials/test_combinatorial.py::test_tutorial[tutorials/combinatorial/hmc_train.ipynb]": 113.72826133295894,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/allset_train.ipynb]": 10.690549792023376,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/allset_transformer_train.ipynb]": 17.616297374945134,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/dhgcn_train.ipynb]": 47.93890062533319,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/hmpnn_train.ipynb]": 7.804051124723628,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/hnhn_train.ipynb]": 5.239491417305544,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/allset_train.ipynb]": 8.943679120042361,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/allset_transformer_train.ipynb]": 5.0502030230127275,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/dhgcn_train.ipynb]": 80.38319573295303,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/hmpnn_train.ipynb]": 5.240434563951567,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/hnhn_train.ipynb]": 5.292717668809928,
     "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/hnhn_train_bis.ipynb]": 16.502087499015033,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/hypergat_train.ipynb]": 16.043949334183708,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/hypersage_train.ipynb]": 44.49209712515585,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/unigcn_train.ipynb]": 5.906059790868312,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/unigcnii_train.ipynb]": 19.70078912517056,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/unigin_train.ipynb]": 8.093902500113472,
-    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/unisage_train.ipynb]": 8.754238750087097,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/hypergat_train.ipynb]": 16.28355930291582,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/hypersage_train.ipynb]": 8.78812964912504,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/unigcn_train.ipynb]": 6.283768973895349,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/unigcnii_train.ipynb]": 6.922894097981043,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/unigin_train.ipynb]": 6.994051485904492,
+    "test/tutorials/test_hypergraph.py::test_tutorial[tutorials/hypergraph/unisage_train.ipynb]": 4.923791292007081,
     "test/tutorials/test_simplicial.py::test_tutorial[tutorials/simplicial/dist2cycle_train.ipynb]": 5.09599137515761,
     "test/tutorials/test_simplicial.py::test_tutorial[tutorials/simplicial/hsn_train.ipynb]": 4.862493500346318,
     "test/tutorials/test_simplicial.py::test_tutorial[tutorials/simplicial/san_train.ipynb]": 5.597147291759029,

test/nn/cell/test_can.py

@@ -21,7 +21,7 @@ def test_forward(self):
             num_classes=1,
             n_layers=2,
             att_lift=False,
-        )
+        ).to(device)
 
         x_0 = torch.rand(2, 2)
         x_1 = torch.rand(2, 2)

test/nn/cell/test_ccxn.py

@@ -18,7 +18,7 @@ def test_forward(self):
             num_classes=1,
             n_layers=2,
             att=False,
-        )
+        ).to(device)
 
         x_0 = torch.rand(2, 2)
         x_1 = torch.rand(2, 2)

test/nn/cell/test_cwn.py

@@ -18,7 +18,7 @@ def test_forward(self):
             hid_channels=16,
             num_classes=1,
             n_layers=2,
-        )
+        ).to(device)
 
         x_0 = torch.rand(2, 2)
         x_1 = torch.rand(2, 2)

test/nn/combinatorial/test_hmc.py

@@ -15,7 +15,7 @@ def test_forward(self):
         intermediate_channels = [2, 2, 2]
         final_channels = [2, 2, 2]
         channels_per_layer = [[in_channels, intermediate_channels, final_channels]]
-        model = HMC(channels_per_layer, negative_slope=0.2, num_classes=2)
+        model = HMC(channels_per_layer, negative_slope=0.2, num_classes=2).to(device)
 
         x_0 = torch.rand(2, 2)
         x_1 = torch.rand(2, 2)

test/nn/hypergraph/test_allset.py

@@ -15,16 +15,15 @@ def test_forward(self):
         model = AllSet(
             in_channels=4,
             hidden_channels=4,
-            out_channels=4,
             n_layers=2,
             mlp_num_layers=1,
-        )
+        ).to(device)
 
         x_0 = torch.rand(4, 4)
         incidence_1 = torch.from_numpy(np.random.rand(4, 4)).to_sparse()
 
-        x_0 = torch.tensor(x_0).float().to(device)
+        x_0 = x_0.float().to(device)
         incidence_1 = incidence_1.float().to(device)
 
-        y = model(x_0, incidence_1)
-        assert y.shape == torch.Size([])
+        x_0, _ = model(x_0, incidence_1)
+        assert x_0.shape == (4, 4)

test/nn/hypergraph/test_allset_layer.py

@@ -30,8 +30,10 @@ def test_forward(self, allset_layer):
         incidence_1 = torch.tensor(
             [[1, 0, 0], [0, 1, 1], [1, 1, 1]], dtype=torch.float32
         ).to_sparse()
-        output = allset_layer.forward(x_0, incidence_1)
-        assert output.shape == (3, 64)
+        x_0, x_1 = allset_layer.forward(x_0, incidence_1)
+
+        assert x_0.shape == (3, 64)
+        assert x_1.shape == (3, 64)
 
     def test_AllSetBlock(self):
         """Test the AllSetBlock class.

test/nn/hypergraph/test_allset_transformer.py

@@ -16,16 +16,15 @@ def test_forward(self):
             in_channels=2,
             hidden_channels=2,
             heads=1,
-            out_channels=1,
             n_layers=2,
             mlp_num_layers=1,
-        )
+        ).to(device)
 
         x_0 = torch.rand(2, 2)
         incidence_1 = torch.from_numpy(np.random.rand(2, 2)).to_sparse()
 
-        x_0 = torch.tensor(x_0).float().to(device)
+        x_0 = x_0.float().to(device)
         incidence_1 = incidence_1.float().to(device)
 
-        y = model(x_0, incidence_1)
-        assert y.shape == torch.Size([])
+        x_0, _ = model(x_0, incidence_1)
+        assert x_0.shape == (2, 2)

test/nn/hypergraph/test_allset_transformer_layer.py

@@ -36,8 +36,9 @@ def test_forward(self, allset_transformer_layer):
         incidence_1 = torch.tensor(
             [[1, 0, 0], [0, 1, 1], [1, 1, 1]], dtype=torch.float32
         ).to_sparse()
-        output = allset_transformer_layer.forward(x_0, incidence_1)
-        assert output.shape == (3, 64)
+        x_0, x_1 = allset_transformer_layer.forward(x_0, incidence_1)
+        assert x_0.shape == (3, 64)
+        assert x_1.shape == (3, 64)
 
     def test_forward_with_invalid_input(self, allset_transformer_layer):
         """Test the forward pass of the allsettransformer layer with invalid input."""

test/nn/hypergraph/test_dhgcn.py

@@ -11,9 +11,13 @@ class TestDHGCNL:
     def test_forward(self):
         """Test forward method."""
         # device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-        model = DHGCN(channels_node=8, n_layers=2)
+        in_channels, hidden_channels = 8, 4
+        model = DHGCN(
+            in_channels=in_channels, hidden_channels=hidden_channels, n_layers=2
+        )
 
-        x_0 = torch.rand(8, 8)
+        n_nodes = 8
+        x_0 = torch.rand(n_nodes, in_channels)
 
-        y = model(x_0)
-        assert y.shape == torch.Size([])
+        x_0, _ = model(x_0)
+        assert x_0.shape == torch.Size([n_nodes, hidden_channels])

test/nn/hypergraph/test_dhgcn_layer.py

@@ -28,7 +28,7 @@ def test_kmeans(self, dhgcn_layer):
         output = dhgcn_layer.kmeans(x_0, k=dhgcn_layer.k_centroids)
         assert output.shape == (2, 4)
         if torch.cuda.is_available():
-            x_0 = x_0.to("gpu")
+            x_0 = x_0
             output = dhgcn_layer.kmeans(x_0, k=dhgcn_layer.k_centroids)
             assert output.device.type == x_0.device.type
 
@@ -41,8 +41,8 @@ def test_kmeans_with_invalid_input(self, dhgcn_layer):
     def test_forward(self, dhgcn_layer):
         """Test the forward pass of the DHGCN layer."""
         x = torch.randn(4, 10)
-        output = dhgcn_layer.forward(x)
-        assert output.shape == (4, 10)
+        x_0, _ = dhgcn_layer.forward(x)
+        assert x_0.shape == (4, 10)
 
     def test_forward_invalid_input(self, dhgcn_layer):
         """Test the forward pass of the DHGCN layer."""

test/nn/hypergraph/test_hmpnn.py

@@ -12,18 +12,20 @@ class TestHMPNN:
     def test_forward(self):
         """Test forward method."""
         device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-        model = HMPNN(8, (8, 8), 1, 1).to(device)
-
-        x_0 = torch.rand(8, 8)
-        x_1 = torch.rand(8, 8)
-
-        adjacency_1 = torch.from_numpy(np.random.rand(8, 8)).to_sparse()
-
-        x_0, x_1 = (
-            torch.tensor(x_0).float().to(device),
-            torch.tensor(x_1).float().to(device),
-        )
+        in_channels = 8
+        hidden_channels = 32
+        model = HMPNN(
+            in_channels=in_channels,
+            hidden_channels=hidden_channels,
+        ).to(device)
+
+        n_nodes, n_edges = 8, 10
+        x_0 = torch.rand(n_nodes, in_channels).float().to(device)
+        x_1 = torch.rand(n_edges, in_channels).float().to(device)
+
+        adjacency_1 = torch.from_numpy(np.random.rand(n_nodes, n_edges)).to_sparse()
         adjacency_1 = adjacency_1.float().to(device)
 
-        y = model(x_0, x_1, adjacency_1)
-        assert y.shape == torch.Size([8, 1])
+        x_0, x_1 = model(x_0, x_1, adjacency_1)
+        assert x_0.shape == torch.Size([n_nodes, hidden_channels])
+        assert x_1.shape == torch.Size([n_edges, hidden_channels])

test/nn/hypergraph/test_hnhn.py

@@ -3,7 +3,7 @@
 import numpy as np
 import torch
 
-from topomodelx.nn.hypergraph.hnhn import HNHN, HNHNNetwork
+from topomodelx.nn.hypergraph.hnhn import HNHN
 
 
 class TestHNHN:
@@ -14,55 +14,18 @@ def test_forward(self):
         device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
         adjacency_1 = torch.from_numpy(np.random.rand(2, 2)).to_sparse()
-        adjacency_1 = adjacency_1.float().to(device)
+        adjacency_1 = adjacency_1.float()
+        hidden_channels = 5
 
         model = HNHN(
-            channels_node=2,
-            channels_edge=2,
+            in_channels=2,
+            hidden_channels=hidden_channels,
             incidence_1=adjacency_1,
-            n_classes=1,
             n_layers=2,
         ).to(device)
 
-        x_0 = torch.rand(2, 2)
-        x_1 = torch.rand(2, 2)
+        x_0 = torch.rand(2, 2).float().to(device)
 
-        x_0, x_1 = (
-            torch.tensor(x_0).float().to(device),
-            torch.tensor(x_1).float().to(device),
-        )
-
-        y1, y2 = model(x_0, x_1)
-        assert y1.shape == torch.Size([2, 1])
-        assert y2.shape == torch.Size([2])
-
-
-class TestHNHNNetwork:
-    """Test the HNHNNetwork."""
-
-    def test_forward(self):
-        """Test forward method."""
-        device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-
-        adjacency_1 = torch.from_numpy(np.random.rand(2, 2)).to_sparse()
-        adjacency_1 = adjacency_1.float().to(device)
-
-        model = HNHNNetwork(
-            channels_node=2,
-            channels_edge=2,
-            incidence_1=adjacency_1,
-            n_classes=1,
-            n_layers=2,
-        ).to(device)
-
-        x_0 = torch.rand(2, 2)
-        x_1 = torch.rand(2, 2)
-
-        x_0, x_1 = (
-            torch.tensor(x_0).float().to(device),
-            torch.tensor(x_1).float().to(device),
-        )
-
-        y1, y2 = model(x_0, x_1)
-        assert y1.shape == torch.Size([2, 1])
-        assert y2.shape == torch.Size([2])
+        x_0, x_1 = model(x_0)
+        assert x_0.shape == torch.Size([2, hidden_channels])
+        assert x_1.shape == torch.Size([2, hidden_channels])

test/nn/hypergraph/test_hnhn_layer.py

@@ -11,29 +11,28 @@ class TestHNHNLayer:
     @pytest.fixture
     def template_layer(self):
         """Initialize and return an HNHN layer."""
-        channels_node = 5
-        channels_edge = 8
+        self.in_channels = 5
+        self.hidden_channels = 8
         n_nodes = 10
         n_edges = 20
         incidence_1 = torch.randint(0, 2, (n_nodes, n_edges)).float()
 
         return HNHNLayer(
-            channels_node=channels_node,
-            channels_edge=channels_edge,
+            in_channels=self.in_channels,
+            hidden_channels=self.hidden_channels,
             incidence_1=incidence_1,
         )
 
     def test_forward(self, template_layer):
         """Test the forward pass of the HNHN layer."""
         n_nodes, n_edges = template_layer.incidence_1.shape
-        channels_node = template_layer.channels_node
-        channels_edge = template_layer.channels_edge
-        x_0 = torch.randn(n_nodes, channels_node)
-        x_1 = torch.randn(n_edges, channels_edge)
-        x_0_out, x_1_out = template_layer.forward(x_0, x_1)
 
-        assert x_0_out.shape == x_0.shape
-        assert x_1_out.shape == x_1.shape
+        x_0 = torch.randn(n_nodes, self.in_channels)
+        x_0_out, x_1_out = template_layer.forward(x_0)
+
+        assert x_0_out.shape == (n_nodes, self.hidden_channels)
+        assert x_1_out.shape == (n_edges, self.hidden_channels)
+
         return
 
     def test_compute_normalization_matrices(self, template_layer):