Apply suggestions from code review

Co-authored-by: hlky <[email protected]>

Author: a-r-r-o-w

src/diffusers/models/autoencoders/autoencoder_kl_hunyuan_video.py

@@ -160,7 +160,7 @@ def __init__(
         self.conv1 = HunyuanVideoCausalConv3d(in_channels, out_channels, 3, 1, 0)
 
         self.norm2 = nn.GroupNorm(groups, out_channels, eps=eps, affine=True)
-        self.dropout = torch.nn.Dropout(dropout)
+        self.dropout = nn.Dropout(dropout)
         self.conv2 = HunyuanVideoCausalConv3d(out_channels, out_channels, 3, 1, 0)
 
         self.conv_shortcut = None
@@ -604,7 +604,6 @@ def __init__(
         self.layers_per_block = layers_per_block
 
         self.conv_in = HunyuanVideoCausalConv3d(in_channels, block_out_channels[-1], kernel_size=3, stride=1)
-        self.mid_block = None
         self.up_blocks = nn.ModuleList([])
 
         # mid
@@ -1145,7 +1144,6 @@ def _temporal_tiled_decode(self, z: torch.Tensor, return_dict: bool = True) -> U
             else:
                 result_row.append(tile[:, :, : self.tile_sample_stride_num_frames + 1, :, :])
 
-        print("this:", torch.cat(result_row, dim=2).shape)
         dec = torch.cat(result_row, dim=2)[:, :, :num_sample_frames]
 
         if not return_dict:

src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video.py

@@ -21,7 +21,7 @@
 
 from ...callbacks import MultiPipelineCallbacks, PipelineCallback
 from ...models import AutoencoderKLHunyuanVideo, HunyuanVideoTransformer3DModel
-from ...schedulers import KarrasDiffusionSchedulers
+from ...schedulers import FlowMatchEulerDiscreteScheduler
 from ...utils import logging, replace_example_docstring
 from ...utils.torch_utils import randn_tensor
 from ...video_processor import VideoProcessor
@@ -149,9 +149,6 @@ class HunyuanVideoPipeline(DiffusionPipeline):
             A scheduler to be used in combination with `transformer` to denoise the encoded image latents.
         vae ([`AutoencoderKLHunyuanVideo`]):
             Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations.
-        text_encoder_2 ([`T5EncoderModel`]):
-            [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5EncoderModel), specifically
-            the [google/t5-v1_1-xxl](https://huggingface.co/google/t5-v1_1-xxl) variant.
         text_encoder_2 ([`CLIPTextModel`]):
             [CLIP](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel), specifically
             the [clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14) variant.