Buckets:
| # AllegroTransformer3DModel | |
| [Allegro](https://github.com/rhymes-ai/Allegro)의 3D 데이터를 위한 Diffusion Transformer 모델이 RhymesAI의 [Allegro: Open the Black Box of Commercial-Level Video Generation Model](https://huggingface.co/papers/2410.15458)에서 소개되었습니다. | |
| 다음 코드 스니펫으로 모델을 로드할 수 있습니다. | |
| ```python | |
| from diffusers import AllegroTransformer3DModel | |
| transformer = AllegroTransformer3DModel.from_pretrained("rhymes-ai/Allegro", subfolder="transformer", dtype=torch.bfloat16).to("cuda") # or "mps", "xpu", "cpu" | |
| ``` | |
| ## AllegroTransformer3DModel[[diffusers.AllegroTransformer3DModel]] | |
| #### diffusers.AllegroTransformer3DModel[[diffusers.AllegroTransformer3DModel]] | |
| ```python | |
| diffusers.AllegroTransformer3DModel(patch_size: int = 2, patch_size_t: int = 1, num_attention_heads: int = 24, attention_head_dim: int = 96, in_channels: int = 4, out_channels: int = 4, num_layers: int = 32, dropout: float = 0.0, cross_attention_dim: int = 2304, attention_bias: bool = True, sample_height: int = 90, sample_width: int = 160, sample_frames: int = 22, activation_fn: str = 'gelu-approximate', norm_elementwise_affine: bool = False, norm_eps: float = 1e-06, caption_channels: int = 4096, interpolation_scale_h: float = 2.0, interpolation_scale_w: float = 2.0, interpolation_scale_t: float = 2.2) | |
| ``` | |
| [소스](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/transformers/transformer_allegro.py#L174) | |
| #### forward[[diffusers.AllegroTransformer3DModel.forward]] | |
| ```python | |
| forward(hidden_states: Tensor, encoder_hidden_states: Tensor, timestep: LongTensor, attention_mask: typing.Optional[torch.Tensor] = None, encoder_attention_mask: typing.Optional[torch.Tensor] = None, image_rotary_emb: tuple[torch.Tensor, torch.Tensor] | None = None, return_dict: bool = True) | |
| ``` | |
| [소스](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/transformers/transformer_allegro.py#L305) | |
| **매개변수:** | |
| hidden_states (`torch.Tensor`, shape `(batch_size, num_channels, num_frames, height, width)`) : 입력 `hidden_states`. | |
| encoder_hidden_states (`torch.Tensor`, shape `(batch_size, sequence_len, embed_dims)`) : 사용할 조건 임베딩(프롬프트와 같은 입력 조건에서 계산된 임베딩). | |
| timestep (`torch.LongTensor`) : 노이즈 제거 단계를 나타내는 데 사용됩니다. | |
| attention_mask (`torch.Tensor`, *선택 사항*) : `hidden_states`에 적용되는 자기 주의(self-attention) 마스크입니다. | |
| encoder_attention_mask (`torch.Tensor`, *선택 사항*) : `encoder_hidden_states`에 적용되는 크로스 어텐션 마스크. | |
| image_rotary_emb (`tuple` of `torch.Tensor`, *선택 사항*) : 사전 계산된 회전 위치 임베딩. | |
| return_dict (`bool`, *선택 사항*, 기본값 `True`) : 일반 튜플 대신 `~models.transformer_2d.Transformer2DModelOutput`을 반환할지 여부. | |
| **반환값:** | |
| `return_dict`가 True이면 `~models.transformer_2d.Transformer2DModelOutput`이 반환되고, 그렇지 않으면 첫 번째 요소가 샘플 텐서인 `tuple`이 반환됩니다. | |
| [AllegroTransformer3DModel](/docs/diffusers/main/ko/api/models/allegro_transformer3d#diffusers.AllegroTransformer3DModel)의 forward 메서드입니다. | |
| ## Transformer2DModelOutput[[diffusers.models.modeling_outputs.Transformer2DModelOutput]] | |
| #### diffusers.models.modeling_outputs.Transformer2DModelOutput[[diffusers.models.modeling_outputs.Transformer2DModelOutput]] | |
| ```python | |
| diffusers.models.modeling_outputs.Transformer2DModelOutput(sample: torch.Tensor) | |
| ``` | |
| [소스](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/modeling_outputs.py#L21) | |
| **매개변수:** | |
| sample (`torch.Tensor`, shape `(batch_size, num_channels, height, width)` 또는 [Transformer2DModel](/docs/diffusers/main/ko/api/models/transformer2d#diffusers.Transformer2DModel)이 discrete인 경우 `(batch size, num_vector_embeds - 1, num_latent_pixels)`) : `encoder_hidden_states` 입력에 조건부인 hidden states 출력입니다. discrete인 경우, 노이즈가 제거된 latent 픽셀에 대한 확률 분포를 반환합니다. | |
| [Transformer2DModel](/docs/diffusers/main/ko/api/models/transformer2d#diffusers.Transformer2DModel)의 출력입니다. | |
Xet Storage Details
- Size:
- 4.28 kB
- Xet hash:
- 11f415cce6e1ddfb7cec42ece87ca73d7bbda06eb331b1003fcda2cc56fa0ea5
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.