[UNet2DConditionModel, UNet2DModel] pass norm_num_groups to all the blocks (#442)

* pass norm_num_groups to unet blocs and attention

* fix UNet2DConditionModel

* add norm_num_groups arg in vae

* add tests

* remove comment

* Apply suggestions from code review

Author: patil-suraj

src/diffusers/models/attention.py

@@ -113,14 +113,15 @@ def __init__(
         d_head: int,
         depth: int = 1,
         dropout: float = 0.0,
+        num_groups: int = 32,
         context_dim: Optional[int] = None,
     ):
         super().__init__()
         self.n_heads = n_heads
         self.d_head = d_head
         self.in_channels = in_channels
         inner_dim = n_heads * d_head
-        self.norm = torch.nn.GroupNorm(num_groups=32, num_channels=in_channels, eps=1e-6, affine=True)
+        self.norm = torch.nn.GroupNorm(num_groups=num_groups, num_channels=in_channels, eps=1e-6, affine=True)
 
         self.proj_in = nn.Conv2d(in_channels, inner_dim, kernel_size=1, stride=1, padding=0)
 

src/diffusers/models/unet_2d.py

@@ -114,6 +114,7 @@ def __init__(
                 add_downsample=not is_final_block,
                 resnet_eps=norm_eps,
                 resnet_act_fn=act_fn,
+                resnet_groups=norm_num_groups,
                 attn_num_head_channels=attention_head_dim,
                 downsample_padding=downsample_padding,
             )
@@ -151,6 +152,7 @@ def __init__(
                 add_upsample=not is_final_block,
                 resnet_eps=norm_eps,
                 resnet_act_fn=act_fn,
+                resnet_groups=norm_num_groups,
                 attn_num_head_channels=attention_head_dim,
             )
             self.up_blocks.append(up_block)

src/diffusers/models/unet_2d_condition.py

@@ -114,6 +114,7 @@ def __init__(
                 add_downsample=not is_final_block,
                 resnet_eps=norm_eps,
                 resnet_act_fn=act_fn,
+                resnet_groups=norm_num_groups,
                 cross_attention_dim=cross_attention_dim,
                 attn_num_head_channels=attention_head_dim,
                 downsample_padding=downsample_padding,
@@ -153,6 +154,7 @@ def __init__(
                 add_upsample=not is_final_block,
                 resnet_eps=norm_eps,
                 resnet_act_fn=act_fn,
+                resnet_groups=norm_num_groups,
                 cross_attention_dim=cross_attention_dim,
                 attn_num_head_channels=attention_head_dim,
             )

src/diffusers/models/unet_blocks.py

@@ -31,6 +31,7 @@ def get_down_block(
     resnet_eps,
     resnet_act_fn,
     attn_num_head_channels,
+    resnet_groups=None,
     cross_attention_dim=None,
     downsample_padding=None,
 ):
@@ -44,6 +45,7 @@ def get_down_block(
             add_downsample=add_downsample,
             resnet_eps=resnet_eps,
             resnet_act_fn=resnet_act_fn,
+            resnet_groups=resnet_groups,
             downsample_padding=downsample_padding,
         )
     elif down_block_type == "AttnDownBlock2D":
@@ -55,6 +57,7 @@ def get_down_block(
             add_downsample=add_downsample,
             resnet_eps=resnet_eps,
             resnet_act_fn=resnet_act_fn,
+            resnet_groups=resnet_groups,
             downsample_padding=downsample_padding,
             attn_num_head_channels=attn_num_head_channels,
         )
@@ -69,6 +72,7 @@ def get_down_block(
             add_downsample=add_downsample,
             resnet_eps=resnet_eps,
             resnet_act_fn=resnet_act_fn,
+            resnet_groups=resnet_groups,
             downsample_padding=downsample_padding,
             cross_attention_dim=cross_attention_dim,
             attn_num_head_channels=attn_num_head_channels,
@@ -104,6 +108,7 @@ def get_down_block(
             add_downsample=add_downsample,
             resnet_eps=resnet_eps,
             resnet_act_fn=resnet_act_fn,
+            resnet_groups=resnet_groups,
             downsample_padding=downsample_padding,
         )
 
@@ -119,6 +124,7 @@ def get_up_block(
     resnet_eps,
     resnet_act_fn,
     attn_num_head_channels,
+    resnet_groups=None,
     cross_attention_dim=None,
 ):
     up_block_type = up_block_type[7:] if up_block_type.startswith("UNetRes") else up_block_type
@@ -132,6 +138,7 @@ def get_up_block(
             add_upsample=add_upsample,
             resnet_eps=resnet_eps,
             resnet_act_fn=resnet_act_fn,
+            resnet_groups=resnet_groups,
         )
     elif up_block_type == "CrossAttnUpBlock2D":
         if cross_attention_dim is None:
@@ -145,6 +152,7 @@ def get_up_block(
             add_upsample=add_upsample,
             resnet_eps=resnet_eps,
             resnet_act_fn=resnet_act_fn,
+            resnet_groups=resnet_groups,
             cross_attention_dim=cross_attention_dim,
             attn_num_head_channels=attn_num_head_channels,
         )
@@ -158,6 +166,7 @@ def get_up_block(
             add_upsample=add_upsample,
             resnet_eps=resnet_eps,
             resnet_act_fn=resnet_act_fn,
+            resnet_groups=resnet_groups,
             attn_num_head_channels=attn_num_head_channels,
         )
     elif up_block_type == "SkipUpBlock2D":
@@ -191,6 +200,7 @@ def get_up_block(
             add_upsample=add_upsample,
             resnet_eps=resnet_eps,
             resnet_act_fn=resnet_act_fn,
+            resnet_groups=resnet_groups,
         )
     raise ValueError(f"{up_block_type} does not exist.")
 
@@ -323,6 +333,7 @@ def __init__(
                     in_channels // attn_num_head_channels,
                     depth=1,
                     context_dim=cross_attention_dim,
+                    num_groups=resnet_groups,
                 )
             )
             resnets.append(
@@ -414,6 +425,7 @@ def __init__(
                     num_head_channels=attn_num_head_channels,
                     rescale_output_factor=output_scale_factor,
                     eps=resnet_eps,
+                    num_groups=resnet_groups,
                 )
             )
 
@@ -498,6 +510,7 @@ def __init__(
                     out_channels // attn_num_head_channels,
                     depth=1,
                     context_dim=cross_attention_dim,
+                    num_groups=resnet_groups,
                 )
             )
         self.attentions = nn.ModuleList(attentions)
@@ -966,6 +979,7 @@ def __init__(
                     num_head_channels=attn_num_head_channels,
                     rescale_output_factor=output_scale_factor,
                     eps=resnet_eps,
+                    num_groups=resnet_groups,
                 )
             )
 
@@ -1047,6 +1061,7 @@ def __init__(
                     out_channels // attn_num_head_channels,
                     depth=1,
                     context_dim=cross_attention_dim,
+                    num_groups=resnet_groups,
                 )
             )
         self.attentions = nn.ModuleList(attentions)

src/diffusers/models/vae.py

@@ -59,6 +59,7 @@ def __init__(
         down_block_types=("DownEncoderBlock2D",),
         block_out_channels=(64,),
         layers_per_block=2,
+        norm_num_groups=32,
         act_fn="silu",
         double_z=True,
     ):
@@ -86,6 +87,7 @@ def __init__(
                 resnet_eps=1e-6,
                 downsample_padding=0,
                 resnet_act_fn=act_fn,
+                resnet_groups=norm_num_groups,
                 attn_num_head_channels=None,
                 temb_channels=None,
             )
@@ -99,13 +101,12 @@ def __init__(
             output_scale_factor=1,
             resnet_time_scale_shift="default",
             attn_num_head_channels=None,
-            resnet_groups=32,
+            resnet_groups=norm_num_groups,
             temb_channels=None,
         )
 
         # out
-        num_groups_out = 32
-        self.conv_norm_out = nn.GroupNorm(num_channels=block_out_channels[-1], num_groups=num_groups_out, eps=1e-6)
+        self.conv_norm_out = nn.GroupNorm(num_channels=block_out_channels[-1], num_groups=norm_num_groups, eps=1e-6)
         self.conv_act = nn.SiLU()
 
         conv_out_channels = 2 * out_channels if double_z else out_channels
@@ -138,6 +139,7 @@ def __init__(
         up_block_types=("UpDecoderBlock2D",),
         block_out_channels=(64,),
         layers_per_block=2,
+        norm_num_groups=32,
         act_fn="silu",
     ):
         super().__init__()
@@ -156,7 +158,7 @@ def __init__(
             output_scale_factor=1,
             resnet_time_scale_shift="default",
             attn_num_head_channels=None,
-            resnet_groups=32,
+            resnet_groups=norm_num_groups,
             temb_channels=None,
         )
 
@@ -178,15 +180,15 @@ def __init__(
                 add_upsample=not is_final_block,
                 resnet_eps=1e-6,
                 resnet_act_fn=act_fn,
+                resnet_groups=norm_num_groups,
                 attn_num_head_channels=None,
                 temb_channels=None,
             )
             self.up_blocks.append(up_block)
             prev_output_channel = output_channel
 
         # out
-        num_groups_out = 32
-        self.conv_norm_out = nn.GroupNorm(num_channels=block_out_channels[0], num_groups=num_groups_out, eps=1e-6)
+        self.conv_norm_out = nn.GroupNorm(num_channels=block_out_channels[0], num_groups=norm_num_groups, eps=1e-6)
         self.conv_act = nn.SiLU()
         self.conv_out = nn.Conv2d(block_out_channels[0], out_channels, 3, padding=1)
 
@@ -405,6 +407,7 @@ def __init__(
         latent_channels: int = 3,
         sample_size: int = 32,
         num_vq_embeddings: int = 256,
+        norm_num_groups: int = 32,
     ):
         super().__init__()
 
@@ -416,6 +419,7 @@ def __init__(
             block_out_channels=block_out_channels,
             layers_per_block=layers_per_block,
             act_fn=act_fn,
+            norm_num_groups=norm_num_groups,
             double_z=False,
         )
 
@@ -433,6 +437,7 @@ def __init__(
             block_out_channels=block_out_channels,
             layers_per_block=layers_per_block,
             act_fn=act_fn,
+            norm_num_groups=norm_num_groups,
         )
 
     def encode(self, x: torch.FloatTensor, return_dict: bool = True) -> VQEncoderOutput:
@@ -509,6 +514,7 @@ def __init__(
         layers_per_block: int = 1,
         act_fn: str = "silu",
         latent_channels: int = 4,
+        norm_num_groups: int = 32,
         sample_size: int = 32,
     ):
         super().__init__()
@@ -521,6 +527,7 @@ def __init__(
             block_out_channels=block_out_channels,
             layers_per_block=layers_per_block,
             act_fn=act_fn,
+            norm_num_groups=norm_num_groups,
             double_z=True,
         )
 
@@ -531,6 +538,7 @@ def __init__(
             up_block_types=up_block_types,
             block_out_channels=block_out_channels,
             layers_per_block=layers_per_block,
+            norm_num_groups=norm_num_groups,
             act_fn=act_fn,
         )
 

tests/test_modeling_common.py

@@ -99,6 +99,26 @@ def test_output(self):
         expected_shape = inputs_dict["sample"].shape
         self.assertEqual(output.shape, expected_shape, "Input and output shapes do not match")
 
+    def test_forward_with_norm_groups(self):
+        init_dict, inputs_dict = self.prepare_init_args_and_inputs_for_common()
+
+        init_dict["norm_num_groups"] = 16
+        init_dict["block_out_channels"] = (16, 32)
+
+        model = self.model_class(**init_dict)
+        model.to(torch_device)
+        model.eval()
+
+        with torch.no_grad():
+            output = model(**inputs_dict)
+
+            if isinstance(output, dict):
+                output = output.sample
+
+        self.assertIsNotNone(output)
+        expected_shape = inputs_dict["sample"].shape
+        self.assertEqual(output.shape, expected_shape, "Input and output shapes do not match")
+
     def test_forward_signature(self):
         init_dict, _ = self.prepare_init_args_and_inputs_for_common()
 

tests/test_models_unet.py

@@ -293,3 +293,7 @@ def test_output_pretrained_ve_large(self):
         # fmt: on
 
         self.assertTrue(torch.allclose(output_slice, expected_output_slice, rtol=1e-2))
+
+    def test_forward_with_norm_groups(self):
+        # not required for this model
+        pass