[Docs] Fix warnings in mkdocs build (continued) (#24740)
Signed-off-by: Zerohertz <ohg3417@gmail.com>
This commit is contained in:
@@ -143,16 +143,8 @@ class AriaProjector(nn.Module):
|
||||
projects ViT's outputs into MoE's inputs.
|
||||
|
||||
Args:
|
||||
patch_to_query_dict (dict): Maps patch numbers to their corresponding
|
||||
query numbers,
|
||||
e.g., {1225: 128, 4900: 256}. This allows for different query sizes
|
||||
based on image resolution.
|
||||
embed_dim (int): Embedding dimension.
|
||||
num_heads (int): Number of attention heads.
|
||||
kv_dim (int): Dimension of key and value.
|
||||
ff_dim (int): Hidden dimension of the feed-forward network.
|
||||
output_dim (int): Output dimension.
|
||||
norm_layer (nn.Module): Normalization layer. Default is nn.LayerNorm.
|
||||
config: [AriaConfig](https://huggingface.co/docs/transformers/main/model_doc/aria#transformers.AriaConfig)
|
||||
containing projector configuration parameters.
|
||||
|
||||
Outputs:
|
||||
A tensor with the shape of (batch_size, query_number, output_dim)
|
||||
@@ -282,8 +274,8 @@ class AriaTextMoELayer(nn.Module):
|
||||
Forward pass of the MoE Layer.
|
||||
|
||||
Args:
|
||||
hidden_states (torch.Tensor): Input tensor of shape (batch_size,
|
||||
sequence_length, hidden_size).
|
||||
hidden_states: Input tensor of shape
|
||||
(batch_size, sequence_length, hidden_size).
|
||||
|
||||
Returns:
|
||||
torch.Tensor: Output tensor after passing through the MoE layer.
|
||||
|
||||
@@ -401,8 +401,7 @@ class BartEncoderLayer(nn.Module):
|
||||
def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
hidden_states
|
||||
torch.Tensor of *encoder* input embeddings.
|
||||
hidden_states: torch.Tensor of *encoder* input embeddings.
|
||||
Returns:
|
||||
Encoder layer output torch.Tensor
|
||||
"""
|
||||
@@ -490,10 +489,8 @@ class BartDecoderLayer(nn.Module):
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
decoder_hidden_states
|
||||
torch.Tensor of *decoder* input embeddings.
|
||||
encoder_hidden_states
|
||||
torch.Tensor of *encoder* input embeddings.
|
||||
decoder_hidden_states: torch.Tensor of *decoder* input embeddings.
|
||||
encoder_hidden_states: torch.Tensor of *encoder* input embeddings.
|
||||
Returns:
|
||||
Decoder layer output torch.Tensor
|
||||
"""
|
||||
@@ -584,12 +581,10 @@ class BartEncoder(nn.Module):
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
input_ids
|
||||
Indices of *encoder* input sequence tokens in the vocabulary.
|
||||
Padding will be ignored by default should you
|
||||
provide it.
|
||||
positions
|
||||
Positions of *encoder* input sequence tokens.
|
||||
input_ids: Indices of *encoder* input sequence tokens in the
|
||||
vocabulary.
|
||||
Padding will be ignored by default should you provide it.
|
||||
positions: Positions of *encoder* input sequence tokens.
|
||||
Returns:
|
||||
Decoder output torch.Tensor
|
||||
"""
|
||||
@@ -663,14 +658,11 @@ class BartDecoder(nn.Module):
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
decoder_input_ids
|
||||
Indices of *decoder* input sequence tokens in the vocabulary.
|
||||
Padding will be ignored by default should you
|
||||
provide it.
|
||||
decoder_positions
|
||||
Positions of *decoder* input sequence tokens.
|
||||
encoder_hidden_states:
|
||||
Tensor of encoder output embeddings
|
||||
decoder_input_ids: Indices of *decoder* input sequence tokens
|
||||
in the vocabulary.
|
||||
Padding will be ignored by default should you provide it.
|
||||
decoder_positions: Positions of *decoder* input sequence tokens.
|
||||
encoder_hidden_states: Tensor of encoder output embeddings.
|
||||
Returns:
|
||||
Decoder output torch.Tensor
|
||||
"""
|
||||
@@ -732,16 +724,13 @@ class BartModel(nn.Module, SupportsQuant):
|
||||
encoder_positions: torch.Tensor) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
input_ids
|
||||
Indices of *decoder* input sequence tokens in the vocabulary.
|
||||
Padding will be ignored by default should you
|
||||
provide it.
|
||||
positions
|
||||
Positions of *decoder* input sequence tokens.
|
||||
encoder_input_ids
|
||||
Indices of *encoder* input sequence tokens in the vocabulary.
|
||||
encoder_positions:
|
||||
Positions of *encoder* input sequence tokens.
|
||||
input_ids: Indices of *decoder* input sequence tokens
|
||||
in the vocabulary.
|
||||
Padding will be ignored by default should you provide it.
|
||||
positions: Positions of *decoder* input sequence tokens.
|
||||
encoder_input_ids: Indices of *encoder* input sequence tokens
|
||||
in the vocabulary.
|
||||
encoder_positions: Positions of *encoder* input sequence tokens.
|
||||
Returns:
|
||||
Model output torch.Tensor
|
||||
"""
|
||||
@@ -848,14 +837,10 @@ class BartForConditionalGeneration(nn.Module, SupportsV0Only, SupportsQuant):
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
input_ids
|
||||
torch.Tensor of *decoder* input token ids.
|
||||
positions
|
||||
torch.Tensor of *decoder* position indices.
|
||||
encoder_input_ids
|
||||
torch.Tensor of *encoder* input token ids.
|
||||
encoder_positions
|
||||
torch.Tensor of *encoder* position indices
|
||||
input_ids: torch.Tensor of *decoder* input token ids.
|
||||
positions: torch.Tensor of *decoder* position indices.
|
||||
encoder_input_ids: torch.Tensor of *encoder* input token ids.
|
||||
encoder_positions: torch.Tensor of *encoder* position indices.
|
||||
Returns:
|
||||
Output torch.Tensor
|
||||
"""
|
||||
@@ -912,8 +897,7 @@ class MBartEncoderLayer(BartEncoderLayer):
|
||||
def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
hidden_states
|
||||
torch.Tensor of *encoder* input embeddings.
|
||||
hidden_states: torch.Tensor of *encoder* input embeddings.
|
||||
Returns:
|
||||
Encoder layer output torch.Tensor
|
||||
"""
|
||||
@@ -1035,12 +1019,10 @@ class MBartEncoder(nn.Module):
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
input_ids
|
||||
Indices of *encoder* input sequence tokens in the vocabulary.
|
||||
Padding will be ignored by default should you
|
||||
provide it.
|
||||
positions
|
||||
Positions of *encoder* input sequence tokens.
|
||||
input_ids: Indices of *encoder* input sequence tokens in the
|
||||
vocabulary.
|
||||
Padding will be ignored by default should you provide it.
|
||||
positions: Positions of *encoder* input sequence tokens.
|
||||
Returns:
|
||||
Decoder output torch.Tensor
|
||||
"""
|
||||
@@ -1116,14 +1098,11 @@ class MBartDecoder(nn.Module):
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
decoder_input_ids
|
||||
Indices of *decoder* input sequence tokens in the vocabulary.
|
||||
Padding will be ignored by default should you
|
||||
provide it.
|
||||
decoder_positions
|
||||
Positions of *decoder* input sequence tokens.
|
||||
encoder_hidden_states:
|
||||
Tensor of encoder output embeddings
|
||||
decoder_input_ids: Indices of *decoder* input sequence tokens
|
||||
in the vocabulary.
|
||||
Padding will be ignored by default should you provide it.
|
||||
decoder_positions: Positions of *decoder* input sequence tokens.
|
||||
encoder_hidden_states: Tensor of encoder output embeddings.
|
||||
Returns:
|
||||
Decoder output torch.Tensor
|
||||
"""
|
||||
@@ -1185,16 +1164,13 @@ class MBartModel(nn.Module, SupportsQuant):
|
||||
encoder_positions: torch.Tensor) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
input_ids
|
||||
Indices of *decoder* input sequence tokens in the vocabulary.
|
||||
Padding will be ignored by default should you
|
||||
provide it.
|
||||
positions
|
||||
Positions of *decoder* input sequence tokens.
|
||||
encoder_input_ids
|
||||
Indices of *encoder* input sequence tokens in the vocabulary.
|
||||
encoder_positions:
|
||||
Positions of *encoder* input sequence tokens.
|
||||
input_ids: Indices of *decoder* input sequence tokens
|
||||
in the vocabulary.
|
||||
Padding will be ignored by default should you provide it.
|
||||
positions: Positions of *decoder* input sequence tokens.
|
||||
encoder_input_ids: Indices of *encoder* input sequence tokens
|
||||
in the vocabulary.
|
||||
encoder_positions: Positions of *encoder* input sequence tokens.
|
||||
Returns:
|
||||
Model output torch.Tensor
|
||||
"""
|
||||
|
||||
@@ -678,7 +678,6 @@ class Blip2ForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsPP,
|
||||
Args:
|
||||
input_ids: Flattened (concatenated) input_ids corresponding to a
|
||||
batch.
|
||||
pixel_values: The pixels in each input image.
|
||||
|
||||
Info:
|
||||
[Blip2ImageInputs][]
|
||||
|
||||
@@ -79,10 +79,8 @@ class DonutLanguageForConditionalGeneration(nn.Module, SupportsV0Only):
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
input_ids
|
||||
torch.Tensor of *decoder* input token ids.
|
||||
positions
|
||||
torch.Tensor of *decoder* position indices.
|
||||
input_ids: torch.Tensor of *decoder* input token ids.
|
||||
positions: torch.Tensor of *decoder* position indices.
|
||||
Returns:
|
||||
Output torch.Tensor
|
||||
"""
|
||||
@@ -351,14 +349,10 @@ class DonutForConditionalGeneration(nn.Module, SupportsMultiModal,
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
input_ids
|
||||
torch.Tensor of *decoder* input token ids.
|
||||
positions
|
||||
torch.Tensor of *decoder* position indices.
|
||||
encoder_input_ids
|
||||
torch.Tensor of *encoder* input token ids.
|
||||
encoder_positions
|
||||
torch.Tensor of *encoder* position indices
|
||||
input_ids: torch.Tensor of *decoder* input token ids.
|
||||
positions: torch.Tensor of *decoder* position indices.
|
||||
encoder_input_ids: torch.Tensor of *encoder* input token ids.
|
||||
encoder_positions: torch.Tensor of *encoder* position indices
|
||||
Returns:
|
||||
Output torch.Tensor
|
||||
"""
|
||||
|
||||
@@ -631,16 +631,14 @@ class Florence2LanguageModel(nn.Module):
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
input_ids
|
||||
Indices of *decoder* input sequence tokens in the vocabulary.
|
||||
input_ids: Indices of *decoder* input sequence tokens
|
||||
in the vocabulary.
|
||||
Padding will be ignored by default should you
|
||||
provide it.
|
||||
positions
|
||||
Positions of *decoder* input sequence tokens.
|
||||
encoder_input_ids
|
||||
Indices of *encoder* input sequence tokens in the vocabulary.
|
||||
encoder_positions:
|
||||
Positions of *encoder* input sequence tokens.
|
||||
positions: Positions of *decoder* input sequence tokens.
|
||||
encoder_input_ids: Indices of *encoder* input sequence tokens
|
||||
in the vocabulary.
|
||||
encoder_positions: Positions of *encoder* input sequence tokens.
|
||||
Returns:
|
||||
Model output torch.Tensor
|
||||
"""
|
||||
@@ -699,14 +697,10 @@ class Florence2LanguageForConditionalGeneration(nn.Module, SupportsV0Only):
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
input_ids
|
||||
torch.Tensor of *decoder* input token ids.
|
||||
positions
|
||||
torch.Tensor of *decoder* position indices.
|
||||
encoder_input_ids
|
||||
torch.Tensor of *encoder* input token ids.
|
||||
encoder_positions
|
||||
torch.Tensor of *encoder* position indices
|
||||
input_ids: torch.Tensor of *decoder* input token ids.
|
||||
positions: torch.Tensor of *decoder* position indices.
|
||||
encoder_input_ids: torch.Tensor of *encoder* input token ids.
|
||||
encoder_positions: torch.Tensor of *encoder* position indices
|
||||
Returns:
|
||||
Output torch.Tensor
|
||||
"""
|
||||
@@ -1068,14 +1062,10 @@ class Florence2ForConditionalGeneration(nn.Module, SupportsMultiModal,
|
||||
) -> torch.Tensor:
|
||||
r"""
|
||||
Args:
|
||||
input_ids
|
||||
torch.Tensor of *decoder* input token ids.
|
||||
positions
|
||||
torch.Tensor of *decoder* position indices.
|
||||
encoder_input_ids
|
||||
torch.Tensor of *encoder* input token ids.
|
||||
encoder_positions
|
||||
torch.Tensor of *encoder* position indices
|
||||
input_ids: torch.Tensor of *decoder* input token ids.
|
||||
positions: torch.Tensor of *decoder* position indices.
|
||||
encoder_input_ids: torch.Tensor of *encoder* input token ids.
|
||||
encoder_positions: torch.Tensor of *encoder* position indices
|
||||
Returns:
|
||||
Output torch.Tensor
|
||||
"""
|
||||
|
||||
@@ -1599,17 +1599,10 @@ class Glm4vForConditionalGeneration(nn.Module, SupportsMultiModal,
|
||||
**NOTE**: If mrope is enabled (default setting for GLM-4V
|
||||
opensource models), the shape will be `(3, seq_len)`,
|
||||
otherwise it will be `(seq_len,).
|
||||
pixel_values: Pixel values to be fed to a model.
|
||||
`None` if no images are passed.
|
||||
image_grid_thw: Tensor `(n_images, 3)` of image 3D grid in LLM.
|
||||
`None` if no images are passed.
|
||||
pixel_values_videos: Pixel values of videos to be fed to a model.
|
||||
`None` if no videos are passed.
|
||||
video_grid_thw: Tensor `(n_videos, 3)` of video 3D grid in LLM.
|
||||
`None` if no videos are passed.
|
||||
second_per_grid_ts: Tensor `(num_videos)` of video time interval (
|
||||
in seconds) for each grid along the temporal dimension in the
|
||||
3D position IDs. `None` if no videos are passed.
|
||||
intermediate_tensors: Optional intermediate tensors for pipeline
|
||||
parallelism.
|
||||
inputs_embeds: Optional pre-computed input embeddings.
|
||||
**kwargs: Additional keyword arguments.
|
||||
"""
|
||||
if intermediate_tensors is not None:
|
||||
inputs_embeds = None
|
||||
|
||||
Reference in New Issue
Block a user