[Chore] Bump Python version and apply formatting changes for v0.12 release (#126)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung authored and GitHub committed 2025-12-16 15:19:40 +08:00
1 parent 31fedcfa2b
commit c091a03d82
61 files changed
+534 -553

No files matched your search

@@ -7,7 +7,7 @@ import logging
import math
import os
from collections.abc import Iterable
from typing import Any, Optional, Union
from typing import Any
import numpy as np
import torch
@@ -75,10 +75,10 @@ def calculate_shift(
def retrieve_timesteps(
scheduler,
num_inference_steps: Optional[int] = None,
device: Optional[Union[str, torch.device]] = None,
timesteps: Optional[list[int]] = None,
sigmas: Optional[list[float]] = None,
num_inference_steps: int | None = None,
device: str | torch.device | None = None,
timesteps: list[int] | None = None,
sigmas: list[float] | None = None,
**kwargs,
) -> tuple[torch.Tensor, int]:
r"""
@@ -186,7 +186,7 @@ def get_timestep_embedding(
def apply_rotary_emb_qwen(
x: torch.Tensor,
freqs_cis: Union[torch.Tensor, tuple[torch.Tensor]],
freqs_cis: torch.Tensor | tuple[torch.Tensor],
use_real: bool = True,
use_real_unbind_dim: int = -1,
) -> tuple[torch.Tensor, torch.Tensor]:
@@ -359,8 +359,8 @@ class QwenImagePipeline(
def _get_qwen_prompt_embeds(
self,
prompt: Union[str, list[str]] = None,
dtype: Optional[torch.dtype] = None,
prompt: str | list[str] = None,
dtype: torch.dtype | None = None,
):
dtype = dtype or self.text_encoder.dtype
@@ -400,10 +400,10 @@ class QwenImagePipeline(
def encode_prompt(
self,
prompt: Union[str, list[str]],
prompt: str | list[str],
num_images_per_prompt: int = 1,
prompt_embeds: Optional[torch.Tensor] = None,
prompt_embeds_mask: Optional[torch.Tensor] = None,
prompt_embeds: torch.Tensor | None = None,
prompt_embeds_mask: torch.Tensor | None = None,
max_sequence_length: int = 1024,
):
r"""
@@ -601,23 +601,23 @@ class QwenImagePipeline(
def forward(
self,
req: OmniDiffusionRequest,
prompt: Union[str, list[str]] = "",
negative_prompt: Union[str, list[str]] = "",
prompt: str | list[str] = "",
negative_prompt: str | list[str] = "",
true_cfg_scale: float = 4.0,
height: int | None = None,
width: int | None = None,
num_inference_steps: int = 50,
sigmas: Optional[list[float]] = None,
sigmas: list[float] | None = None,
guidance_scale: float = 1.0,
num_images_per_prompt: int = 1,
generator: Optional[Union[torch.Generator, list[torch.Generator]]] = None,
latents: Optional[torch.Tensor] = None,
prompt_embeds: Optional[torch.Tensor] = None,
prompt_embeds_mask: Optional[torch.Tensor] = None,
negative_prompt_embeds: Optional[torch.Tensor] = None,
negative_prompt_embeds_mask: Optional[torch.Tensor] = None,
output_type: Optional[str] = "pil",
attention_kwargs: Optional[dict[str, Any]] = None,
generator: torch.Generator | list[torch.Generator] | None = None,
latents: torch.Tensor | None = None,
prompt_embeds: torch.Tensor | None = None,
prompt_embeds_mask: torch.Tensor | None = None,
negative_prompt_embeds: torch.Tensor | None = None,
negative_prompt_embeds_mask: torch.Tensor | None = None,
output_type: str | None = "pil",
attention_kwargs: dict[str, Any] | None = None,
callback_on_step_end_tensor_inputs: list[str] = ["latents"],
max_sequence_length: int = 512,
) -> DiffusionOutput:
@@ -7,7 +7,7 @@ import logging
import math
import os
from collections.abc import Iterable
from typing import Any, Optional, Union
from typing import Any
import numpy as np
import PIL.Image
@@ -134,10 +134,10 @@ def calculate_dimensions(target_area: float, ratio: float):
def retrieve_timesteps(
scheduler,
num_inference_steps: Optional[int] = None,
device: Optional[Union[str, torch.device]] = None,
timesteps: Optional[list[int]] = None,
sigmas: Optional[list[float]] = None,
num_inference_steps: int | None = None,
device: str | torch.device | None = None,
timesteps: list[int] | None = None,
sigmas: list[float] | None = None,
**kwargs,
) -> tuple[torch.Tensor, int]:
r"""
@@ -173,7 +173,7 @@ def retrieve_timesteps(
def retrieve_latents(
encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "argmax"
encoder_output: torch.Tensor, generator: torch.Generator | None = None, sample_mode: str = "argmax"
):
"""Retrieve latents from VAE encoder output."""
if hasattr(encoder_output, "latent_dist"):
@@ -305,9 +305,9 @@ class QwenImageEditPipeline(
def _get_qwen_prompt_embeds(
self,
prompt: Union[str, list[str]] = None,
image: Optional[torch.Tensor] = None,
dtype: Optional[torch.dtype] = None,
prompt: str | list[str] = None,
image: torch.Tensor | None = None,
dtype: torch.dtype | None = None,
):
dtype = dtype or self.text_encoder.dtype
@@ -350,9 +350,9 @@ class QwenImageEditPipeline(
def _get_qwen_prompt_embeds(
self,
prompt: Union[str, list[str]] = None,
image: Optional[Union[PIL.Image.Image, torch.Tensor]] = None,
dtype: Optional[torch.dtype] = None,
prompt: str | list[str] = None,
image: PIL.Image.Image | torch.Tensor | None = None,
dtype: torch.dtype | None = None,
):
"""Get prompt embeddings with image support for editing."""
dtype = dtype or self.text_encoder.dtype
@@ -397,11 +397,11 @@ class QwenImageEditPipeline(
def encode_prompt(
self,
prompt: Union[str, list[str]],
image: Optional[torch.Tensor] = None,
prompt: str | list[str],
image: torch.Tensor | None = None,
num_images_per_prompt: int = 1,
prompt_embeds: Optional[torch.Tensor] = None,
prompt_embeds_mask: Optional[torch.Tensor] = None,
prompt_embeds: torch.Tensor | None = None,
prompt_embeds_mask: torch.Tensor | None = None,
max_sequence_length: int = 1024,
):
r"""
@@ -638,24 +638,24 @@ class QwenImageEditPipeline(
def forward(
self,
req: OmniDiffusionRequest,
prompt: Union[str, list[str]] = "",
negative_prompt: Union[str, list[str]] = "",
image: Optional[Union[PIL.Image.Image, torch.Tensor]] = None,
prompt: str | list[str] = "",
negative_prompt: str | list[str] = "",
image: PIL.Image.Image | torch.Tensor | None = None,
true_cfg_scale: float = 4.0,
height: int | None = None,
width: int | None = None,
num_inference_steps: int = 50,
sigmas: Optional[list[float]] = None,
sigmas: list[float] | None = None,
guidance_scale: float = 1.0,
num_images_per_prompt: int = 1,
generator: Optional[Union[torch.Generator, list[torch.Generator]]] = None,
latents: Optional[torch.Tensor] = None,
prompt_embeds: Optional[torch.Tensor] = None,
prompt_embeds_mask: Optional[torch.Tensor] = None,
negative_prompt_embeds: Optional[torch.Tensor] = None,
negative_prompt_embeds_mask: Optional[torch.Tensor] = None,
output_type: Optional[str] = "pil",
attention_kwargs: Optional[dict[str, Any]] = None,
generator: torch.Generator | list[torch.Generator] | None = None,
latents: torch.Tensor | None = None,
prompt_embeds: torch.Tensor | None = None,
prompt_embeds_mask: torch.Tensor | None = None,
negative_prompt_embeds: torch.Tensor | None = None,
negative_prompt_embeds_mask: torch.Tensor | None = None,
output_type: str | None = "pil",
attention_kwargs: dict[str, Any] | None = None,
callback_on_step_end_tensor_inputs: list[str] = ["latents"],
max_sequence_length: int = 512,
) -> DiffusionOutput:
@@ -3,7 +3,7 @@
import functools
from collections.abc import Iterable
from typing import Any, Optional, Union
from typing import Any
import torch
import torch.nn as nn
@@ -26,7 +26,7 @@ logger = init_logger(__name__)
def apply_rotary_emb_qwen(
x: torch.Tensor,
freqs_cis: Union[torch.Tensor, tuple[torch.Tensor]],
freqs_cis: torch.Tensor | tuple[torch.Tensor],
use_real: bool = True,
use_real_unbind_dim: int = -1,
) -> tuple[torch.Tensor, torch.Tensor]:
@@ -393,8 +393,8 @@ class QwenImageTransformerBlock(nn.Module):
encoder_hidden_states: torch.Tensor,
encoder_hidden_states_mask: torch.Tensor,
temb: torch.Tensor,
image_rotary_emb: Optional[tuple[torch.Tensor, torch.Tensor]] = None,
joint_attention_kwargs: Optional[dict[str, Any]] = None,
image_rotary_emb: tuple[torch.Tensor, torch.Tensor] | None = None,
joint_attention_kwargs: dict[str, Any] | None = None,
) -> tuple[torch.Tensor, torch.Tensor]:
# Get modulation parameters for both streams
img_mod_params = self.img_mod(temb) # [B, 6*dim]
@@ -490,7 +490,7 @@ class QwenImageTransformer2DModel(nn.Module):
od_config: OmniDiffusionConfig,
patch_size: int = 2,
in_channels: int = 64,
out_channels: Optional[int] = 16,
out_channels: int | None = 16,
num_layers: int = 60,
attention_head_dim: int = 128,
num_attention_heads: int = 24,
@@ -537,12 +537,12 @@ class QwenImageTransformer2DModel(nn.Module):
encoder_hidden_states: torch.Tensor = None,
encoder_hidden_states_mask: torch.Tensor = None,
timestep: torch.LongTensor = None,
img_shapes: Optional[list[tuple[int, int, int]]] = None,
txt_seq_lens: Optional[list[int]] = None,
img_shapes: list[tuple[int, int, int]] | None = None,
txt_seq_lens: list[int] | None = None,
guidance: torch.Tensor = None, # TODO: this should probably be removed
attention_kwargs: Optional[dict[str, Any]] = None,
attention_kwargs: dict[str, Any] | None = None,
return_dict: bool = True,
) -> Union[torch.Tensor, Transformer2DModelOutput]:
) -> torch.Tensor | Transformer2DModelOutput:
"""
The [`QwenTransformer2DModel`] forward method.
@@ -3,7 +3,7 @@
import math
from collections.abc import Iterable
from typing import Any, Optional, Union
from typing import Any
import torch
import torch.nn as nn
@@ -127,7 +127,7 @@ class WanRotaryPosEmbed(nn.Module):
class WanImageEmbedding(nn.Module):
"""Image embedding module for I2V tasks."""
def __init__(self, in_features: int, out_features: int, pos_embed_seq_len: Optional[int] = None):
def __init__(self, in_features: int, out_features: int, pos_embed_seq_len: int | None = None):
super().__init__()
self.norm1 = FP32LayerNorm(in_features)
@@ -159,8 +159,8 @@ class WanTimeTextImageEmbedding(nn.Module):
time_freq_dim: int,
time_proj_dim: int,
text_embed_dim: int,
image_embed_dim: Optional[int] = None,
pos_embed_seq_len: Optional[int] = None,
image_embed_dim: int | None = None,
pos_embed_seq_len: int | None = None,
):
super().__init__()
@@ -178,9 +178,9 @@ class WanTimeTextImageEmbedding(nn.Module):
self,
timestep: torch.Tensor,
encoder_hidden_states: torch.Tensor,
encoder_hidden_states_image: Optional[torch.Tensor] = None,
timestep_seq_len: Optional[int] = None,
) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, Optional[torch.Tensor]]:
encoder_hidden_states_image: torch.Tensor | None = None,
timestep_seq_len: int | None = None,
) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor | None]:
timestep = self.timesteps_proj(timestep)
if timestep_seq_len is not None:
timestep = timestep.unflatten(0, (-1, timestep_seq_len))
@@ -250,7 +250,7 @@ class WanSelfAttention(nn.Module):
def forward(
self,
hidden_states: torch.Tensor,
rotary_emb: Optional[tuple[torch.Tensor, torch.Tensor]] = None,
rotary_emb: tuple[torch.Tensor, torch.Tensor] | None = None,
) -> torch.Tensor:
# Fused QKV projection
qkv, _ = self.to_qkv(hidden_states)
@@ -296,7 +296,7 @@ class WanCrossAttention(nn.Module):
head_dim: int,
eps: float = 1e-5,
dropout: float = 0.0,
added_kv_proj_dim: Optional[int] = None,
added_kv_proj_dim: int | None = None,
):
super().__init__()
@@ -413,7 +413,7 @@ class WanTransformerBlock(nn.Module):
ffn_dim: int,
num_heads: int,
eps: float = 1e-6,
added_kv_proj_dim: Optional[int] = None,
added_kv_proj_dim: int | None = None,
cross_attn_norm: bool = False,
):
super().__init__()
@@ -528,10 +528,10 @@ class WanTransformer3DModel(nn.Module):
num_layers: int = 40,
cross_attn_norm: bool = True,
eps: float = 1e-6,
image_dim: Optional[int] = None,
added_kv_proj_dim: Optional[int] = None,
image_dim: int | None = None,
added_kv_proj_dim: int | None = None,
rope_max_seq_len: int = 1024,
pos_embed_seq_len: Optional[int] = None,
pos_embed_seq_len: int | None = None,
):
super().__init__()
@@ -598,10 +598,10 @@ class WanTransformer3DModel(nn.Module):
hidden_states: torch.Tensor,
timestep: torch.LongTensor,
encoder_hidden_states: torch.Tensor,
encoder_hidden_states_image: Optional[torch.Tensor] = None,
encoder_hidden_states_image: torch.Tensor | None = None,
return_dict: bool = True,
attention_kwargs: Optional[dict[str, Any]] = None,
) -> Union[torch.Tensor, Transformer2DModelOutput]:
attention_kwargs: dict[str, Any] | None = None,
) -> torch.Tensor | Transformer2DModelOutput:
batch_size, num_channels, num_frames, height, width = hidden_states.shape
p_t, p_h, p_w = self.config.patch_size
post_patch_num_frames = num_frames // p_t
@@ -18,8 +18,8 @@
import inspect
import json
import os
from collections.abc import Iterable
from typing import Any, Callable
from collections.abc import Callable, Iterable
from typing import Any
import torch
import torch.nn as nn
@@ -17,7 +17,6 @@
import math
from collections.abc import Iterable
from typing import Optional
import torch
import torch.nn as nn
@@ -232,7 +231,7 @@ class ZImageTransformerBlock(nn.Module):
x: torch.Tensor,
attn_mask: torch.Tensor,
freqs_cis: torch.Tensor,
adaln_input: Optional[torch.Tensor] = None,
adaln_input: torch.Tensor | None = None,
):
if self.modulation:
assert adaln_input is not None