Embedding out dimension

CHANGELOG.md

@@ -8,6 +8,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added 
 - `SimilarityModuleBase`, `DistanceSimilarityModule`, similarity module to `TransformerTorchBackbone`, parameters to transformer-based models `similarity_module_type`, `similarity_module_kwargs` ([#272](https://github.com/MobileTeleSystems/RecTools/pull/272))
+- `out_dim` property to `IdEmbeddingsItemNet`, `CatFeaturesItemNet` and `SumOfEmbeddingsConstructor` ([#276](https://github.com/MobileTeleSystems/RecTools/pull/276))
 
 ## [0.12.0] - 24.02.2025
 

rectools/models/nn/item_net.py

@@ -46,6 +46,11 @@ def get_all_embeddings(self) -> torch.Tensor:
         """Return item embeddings."""
         raise NotImplementedError()
 
+    @property
+    def out_dim(self) -> int:
+        """Return item embedding output dimension."""
+        raise NotImplementedError()
+
     @property
     def device(self) -> torch.device:
         """Return ItemNet device."""
@@ -222,6 +227,11 @@ def from_dataset_schema(
             )
         return None
 
+    @property
+    def out_dim(self) -> int:
+        """Return categorical item embedding output dimension."""
+        return self.embedding_bag.embedding_dim
+
 
 class IdEmbeddingsItemNet(ItemNetBase):
     """
@@ -317,6 +327,11 @@ def from_dataset_schema(
         n_items = dataset_schema.items.n_hot
         return cls(n_factors, n_items, dropout_rate)
 
+    @property
+    def out_dim(self) -> int:
+        """Return item embedding output dimension."""
+        return self.ids_emb.embedding_dim
+
 
 class ItemNetConstructorBase(ItemNetBase):
     """
@@ -467,3 +482,8 @@ def forward(self, items: torch.Tensor) -> torch.Tensor:
             item_emb = self.item_net_blocks[idx_block](items)
             item_embs.append(item_emb)
         return torch.sum(torch.stack(item_embs, dim=0), dim=0)
+
+    @property
+    def out_dim(self) -> int:
+        """Return item net constructor output dimension."""
+        return self.item_net_blocks[0].out_dim

tests/models/nn/test_item_net.py

@@ -60,6 +60,12 @@ def test_embedding_shape_after_model_pass(self, n_items: int, n_factors: int) ->
         expected_item_ids = item_id_embeddings(items)
         assert expected_item_ids.shape == (n_items, n_factors)
 
+    @pytest.mark.parametrize("n_factors", ((2), (10)))
+    def test_out_dim(self, n_factors: int) -> None:
+        item_id_embeddings = IdEmbeddingsItemNet.from_dataset(DATASET, n_factors=n_factors, dropout_rate=0.5)
+        out_dim = item_id_embeddings.out_dim
+        assert out_dim == n_factors
+
 
 @pytest.mark.filterwarnings("ignore::DeprecationWarning")
 class TestCatFeaturesItemNet:
@@ -295,6 +301,15 @@ def test_warns_when_dataset_schema_categorical_features_are_none(self) -> None:
             """
         )
 
+    @pytest.mark.parametrize("n_factors", ((2), (10)))
+    def test_out_dim(self, dataset_item_features: Dataset, n_factors: int) -> None:
+        cat_item_embeddings = CatFeaturesItemNet.from_dataset(
+            dataset_item_features, n_factors=n_factors, dropout_rate=0.5
+        )
+        assert isinstance(cat_item_embeddings, CatFeaturesItemNet)
+        out_dim = cat_item_embeddings.out_dim
+        assert out_dim == n_factors
+
 
 @pytest.mark.filterwarnings("ignore::DeprecationWarning")
 class TestSumOfEmbeddingsConstructor:
@@ -485,3 +500,20 @@ def test_raise_when_no_item_net_blocks(
             SumOfEmbeddingsConstructor.from_dataset(
                 ds, n_factors=10, dropout_rate=0.5, item_net_block_types=item_net_block_types
             )
+
+    @pytest.mark.parametrize(
+        "item_net_block_types,n_factors",
+        (
+            ((IdEmbeddingsItemNet,), 8),
+            ((IdEmbeddingsItemNet, CatFeaturesItemNet), 16),
+            ((CatFeaturesItemNet,), 16),
+        ),
+    )
+    def test_out_dim(
+        self, dataset_item_features: Dataset, item_net_block_types: tp.Sequence[tp.Type[ItemNetBase]], n_factors: int
+    ) -> None:
+        item_net = SumOfEmbeddingsConstructor.from_dataset(
+            dataset_item_features, n_factors=n_factors, dropout_rate=0.5, item_net_block_types=item_net_block_types
+        )
+        out_dim = item_net.out_dim
+        assert out_dim == n_factors