Add class conditioning (#140)

* Add class conditioning ans tests

Signed-off-by: Walter Hugo Lopez Pinaya <[email protected]>

* Add misssing test (#140)

Signed-off-by: Walter Hugo Lopez Pinaya <[email protected]>

Signed-off-by: Walter Hugo Lopez Pinaya <[email protected]>

generative/networks/nets/diffusion_model_unet.py

@@ -1439,6 +1439,8 @@ class DiffusionModelUNet(nn.Module):
         with_conditioning: if True add spatial transformers to perform conditioning.
         transformer_num_layers: number of layers of Transformer blocks to use.
         cross_attention_dim: number of context dimensions to use.
+        num_class_embeds: if specified (as an int), then this model will be class-conditional with `num_class_embeds`
+        classes.
     """
 
     def __init__(
@@ -1455,6 +1457,7 @@ def __init__(
         with_conditioning: bool = False,
         transformer_num_layers: int = 1,
         cross_attention_dim: Optional[int] = None,
+        num_class_embeds: Optional[int] = None,
     ) -> None:
         super().__init__()
         if with_conditioning is True and cross_attention_dim is None:
@@ -1499,6 +1502,11 @@ def __init__(
             nn.Linear(time_embed_dim, time_embed_dim),
         )
 
+        # class embedding
+        self.num_class_embeds = num_class_embeds
+        if num_class_embeds is not None:
+            self.class_embedding = nn.Embedding(num_class_embeds, time_embed_dim)
+
         # down
         self.down_blocks = nn.ModuleList([])
         output_channel = num_channels[0]
@@ -1591,37 +1599,46 @@ def forward(
         x: torch.Tensor,
         timesteps: torch.Tensor,
         context: Optional[torch.Tensor] = None,
+        class_labels: Optional[torch.Tensor] = None,
     ) -> torch.Tensor:
         """
         Args:
             x: input tensor (N, C, SpatialDims).
             timesteps: timestep tensor (N,).
             context: context tensor (N, 1, ContextDim).
+            class_labels: context tensor (N, ).
         """
         # 1. time
         t_emb = get_timestep_embedding(timesteps, self.block_out_channels[0])
         emb = self.time_embed(t_emb)
 
-        # 2. initial convolution
+        # 2. class
+        if self.num_class_embeds is not None:
+            if class_labels is None:
+                raise ValueError("class_labels should be provided when num_class_embeds > 0")
+            class_emb = self.class_embedding(class_labels)
+            emb = emb + class_emb
+
+        # 3. initial convolution
         h = self.conv_in(x)
 
-        # 3. down
+        # 4. down
         down_block_res_samples: List[torch.Tensor] = [h]
         for downsample_block in self.down_blocks:
             h, res_samples = downsample_block(hidden_states=h, temb=emb, context=context)
             for residual in res_samples:
                 down_block_res_samples.append(residual)
 
-        # 4. mid
+        # 5. mid
         h = self.middle_block(hidden_states=h, temb=emb, context=context)
 
-        # 5. up
+        # 6. up
         for upsample_block in self.up_blocks:
             res_samples = down_block_res_samples[-len(upsample_block.resnets) :]
             down_block_res_samples = down_block_res_samples[: -len(upsample_block.resnets)]
             h = upsample_block(hidden_states=h, res_hidden_states_list=res_samples, temb=emb, context=context)
 
-        # 6. output block
+        # 7. output block
         h = self.out(h)
 
         return h

tests/test_diffusion_model_unet.py

@@ -168,6 +168,44 @@ def test_with_conditioning_cross_attention_dim_none(self):
                 norm_num_groups=8,
             )
 
+    def test_shape_conditioned_models_class_conditioning(self):
+        net = DiffusionModelUNet(
+            spatial_dims=2,
+            in_channels=1,
+            out_channels=1,
+            num_res_blocks=1,
+            num_channels=(8, 8, 8),
+            attention_levels=(False, False, True),
+            norm_num_groups=8,
+            num_head_channels=8,
+            num_class_embeds=2,
+        )
+        with eval_mode(net):
+            result = net.forward(
+                x=torch.rand((1, 1, 16, 32)),
+                timesteps=torch.randint(0, 1000, (1,)).long(),
+                class_labels=torch.randint(0, 2, (1,)).long(),
+            )
+            self.assertEqual(result.shape, (1, 1, 16, 32))
+
+    def test_conditioned_models_no_class_labels(self):
+        with self.assertRaises(ValueError):
+            net = DiffusionModelUNet(
+                spatial_dims=2,
+                in_channels=1,
+                out_channels=1,
+                num_res_blocks=1,
+                num_channels=(8, 8, 8),
+                attention_levels=(False, False, True),
+                norm_num_groups=8,
+                num_head_channels=8,
+                num_class_embeds=2,
+            )
+            net.forward(
+                x=torch.rand((1, 1, 16, 32)),
+                timesteps=torch.randint(0, 1000, (1,)).long(),
+            )
+
     def test_script_unconditioned_2d_models(self):
         net = DiffusionModelUNet(
             spatial_dims=2,