Add pydocstyle to the ruff rules (#52136)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Signed-off-by: Harry Mellor <hej.mellor@gmail.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Harry Mellor
2026-09-16 16:44:29 +00:00
committed by GitHub
co-authored by Claude Opus 5
parent 804e5377aa
commit 8c1557a79c
2002 changed files with 15024 additions and 14269 deletions
+17 -16
View File
@@ -62,8 +62,7 @@ logger = init_logger(__name__)
class BartScaledWordEmbedding(VocabParallelEmbedding):
"""
This module overrides VocabParallelEmbedding's
"""This module overrides VocabParallelEmbedding's
forward by multiplying with embeddings scale.
"""
@@ -138,12 +137,13 @@ class BartDecoderLayer(nn.Module):
decoder_hidden_states: torch.Tensor,
encoder_hidden_states: torch.Tensor | None = None,
) -> torch.Tensor:
r"""
Args:
r"""Args:
decoder_hidden_states: torch.Tensor of *decoder* input embeddings.
encoder_hidden_states: torch.Tensor of *encoder* input embeddings.
Returns:
Decoder layer output torch.Tensor
"""
residual = decoder_hidden_states
@@ -218,12 +218,13 @@ class MBartDecoderLayer(BartDecoderLayer):
class MBartDecoderNoPos(nn.Module):
"""
Transformer decoder consisting of *config.decoder_layers* layers.
"""Transformer decoder consisting of *config.decoder_layers* layers.
Each layer is a [`BartDecoderLayer`]
Args:
config: BartConfig
embed_tokens (nn.Embedding): output embedding
"""
def __init__(
@@ -274,13 +275,13 @@ class MBartDecoderNoPos(nn.Module):
inputs_embeds: torch.Tensor | None = None,
**kwargs,
) -> torch.Tensor:
r"""
Args:
r"""Args:
decoder_input_ids: Indices of *decoder* input sequence tokens in the
vocabulary. Padding will be ignored by default should you provide it.
encoder_hidden_states: Tensor of encoder output embeddings
Returns:
Decoder output torch.Tensor
"""
if inputs_embeds is None:
inputs_embeds = self.embed_tokens(decoder_input_ids)
@@ -339,12 +340,11 @@ class MBartDecoderNoPos(nn.Module):
class NemotronParsePixelInputs(TensorSchema):
"""
Dimensions:
- b: Batch size
- c: Number of channels (3)
- h: Height
- w: Width
"""Dimensions:
- b: Batch size
- c: Number of channels (3)
- h: Height
- w: Width
"""
type: Literal["pixel_values"]
@@ -630,14 +630,15 @@ class NemotronParseForConditionalGeneration(nn.Module, SupportsMultiModal):
encoder_outputs: list[torch.Tensor] | None = None,
**kwargs,
) -> torch.Tensor:
r"""
Args:
r"""Args:
input_ids: torch.Tensor of *decoder* input token ids.
positions: torch.Tensor of *decoder* position indices.
encoder_outputs: List of encoder output tensors (vision embeddings).
During profiling, this may be None or empty.
Returns:
Output torch.Tensor
"""
inputs_embeds = None
if encoder_outputs: