diffusers/main/id/api/models/stable_cascade_unet.md

StableCascadeUNet

Model UNet dari pipeline Stable Cascade.

StableCascadeUNet[[diffusers.models.StableCascadeUNet]]

diffusers.models.StableCascadeUNet[[diffusers.models.StableCascadeUNet]]

diffusers.models.StableCascadeUNet(in_channels: int = 16, out_channels: int = 16, timestep_ratio_embedding_dim: int = 64, patch_size: int = 1, conditioning_dim: int = 2048, block_out_channels: tuple[int, ...] = (2048, 2048), num_attention_heads: tuple[int, ...] = (32, 32), down_num_layers_per_block: tuple[int, ...] = (8, 24), up_num_layers_per_block: tuple[int, ...] = (24, 8), down_blocks_repeat_mappers: tuple[int] | None = (1, 1), up_blocks_repeat_mappers: tuple[int] | None = (1, 1), block_types_per_layer: tuple[tuple[str]] = (('SDCascadeResBlock', 'SDCascadeTimestepBlock', 'SDCascadeAttnBlock'), ('SDCascadeResBlock', 'SDCascadeTimestepBlock', 'SDCascadeAttnBlock')), clip_text_in_channels: int | None = None, clip_text_pooled_in_channels = 1280, clip_image_in_channels: int | None = None, clip_seq = 4, effnet_in_channels: int | None = None, pixel_mapper_in_channels: int | None = None, kernel_size = 3, dropout: float | tuple[float] = (0.1, 0.1), self_attn: bool | tuple[bool] = True, timestep_conditioning_type: tuple[str, ...] = ('sca', 'crp'), switch_level: tuple[bool] | None = None)

Sumber

forward[[diffusers.models.StableCascadeUNet.forward]]

forward(sample, timestep_ratio, clip_text_pooled, clip_text = None, clip_img = None, effnet = None, pixels = None, sca = None, crp = None, return_dict = True)

Sumber

Parameter:

sample (torch.Tensor) : Sampel input berisik.

timestep_ratio (torch.Tensor) : Rasio timestep yang digunakan untuk menghitung embedding timestep.

clip_text_pooled (torch.Tensor) : Embedding teks CLIP yang dipool.

clip_text (torch.Tensor, opsional) : Embedding teks CLIP tingkat urutan.

clip_img (torch.Tensor, opsional) : embedding gambar CLIP.

effnet (torch.Tensor, opsional) : Peta fitur EfficientNet yang digunakan sebagai kondisioning tambahan.

pixels (torch.Tensor, opsional) : Tensor kondisioning tingkat piksel. Jika None, tensor berisi nol akan digunakan.

sca (torch.Tensor, opsional) : Nilai kondisioning sca opsional yang digunakan untuk membangun embedding timestep.

crp (torch.Tensor, opsional) : Nilai kondisioning crp opsional yang digunakan untuk membangun embedding timestep.

return_dict (bool, opsional, defaultnya True) : Apakah akan mengembalikan StableCascadeUNetOutput alih-alih tuple biasa.

Mengembalikan:

Jika return_dict bernilai True, sebuah StableCascadeUNetOutput akan dikembalikan, selain itu sebuah tuple di mana elemen pertama adalah tensor sampel.