fix sorting

Author: neph1

finetrainers/models/hunyuan_video/control_specification.py

@@ -4,33 +4,31 @@
 
 import safetensors
 import torch
-from torch.nn import Module
 from accelerate import init_empty_weights
 from diffusers import (
+    AutoencoderKLHunyuanVideo,
     FlowMatchEulerDiscreteScheduler,
     HunyuanVideoPipeline,
     HunyuanVideoTransformer3DModel,
-    AutoencoderKLHunyuanVideo,
 )
 from diffusers.models.autoencoders.vae import DiagonalGaussianDistribution
+from torch.nn import Module
 from transformers import AutoTokenizer, CLIPTextModel, CLIPTokenizer, LlamaModel
+
 from finetrainers.data._artifact import VideoArtifact
 from finetrainers.models.utils import _expand_conv3d_with_zeroed_weights
 from finetrainers.trainer.control_trainer.config import FrameConditioningType
 from finetrainers.utils.serialization import safetensors_torch_save_function
 
-from ... import data
 from ... import functional as FF
 from ...logging import get_logger
 from ...patches.dependencies.diffusers.control import control_channel_concat
-from ...processors import ProcessorMixin
+from ...processors import CLIPPooledProcessor, LlamaProcessor, ProcessorMixin
 from ...typing import ArtifactType, SchedulerType
-from ...utils import get_non_null_items
+from ...utils import _enable_vae_memory_optimizations, get_non_null_items
 from ..modeling_utils import ControlModelSpecification
 from .base_specification import HunyuanLatentEncodeProcessor
-from ...processors import CLIPPooledProcessor, LlamaProcessor, ProcessorMixin
 
-from ...utils import _enable_vae_memory_optimizations, get_non_null_items
 
 logger = get_logger()