fix: enable unet_3d_condition to support time_cond_proj_dim (huggingface#7364)

Co-authored-by: Sayak Paul <[email protected]>

Author: yhZhai

src/diffusers/models/unets/unet_3d_condition.py

@@ -91,6 +91,8 @@ class UNet3DConditionModel(ModelMixin, ConfigMixin, UNet2DConditionLoadersMixin)
         cross_attention_dim (`int`, *optional*, defaults to 1024): The dimension of the cross attention features.
         attention_head_dim (`int`, *optional*, defaults to 64): The dimension of the attention heads.
         num_attention_heads (`int`, *optional*): The number of attention heads.
+        time_cond_proj_dim (`int`, *optional*, defaults to `None`):
+            The dimension of `cond_proj` layer in the timestep embedding.
     """
 
     _supports_gradient_checkpointing = False
@@ -123,6 +125,7 @@ def __init__(
         cross_attention_dim: int = 1024,
         attention_head_dim: Union[int, Tuple[int]] = 64,
         num_attention_heads: Optional[Union[int, Tuple[int]]] = None,
+        time_cond_proj_dim: Optional[int] = None,
     ):
         super().__init__()
 
@@ -174,6 +177,7 @@ def __init__(
             timestep_input_dim,
             time_embed_dim,
             act_fn=act_fn,
+            cond_proj_dim=time_cond_proj_dim,
         )
 
         self.transformer_in = TransformerTemporalModel(