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
+7 -3
View File
@@ -87,6 +87,7 @@ def deep_select_topk(
Returns:
The (num_rows, topk) indices tensor.
"""
assert input.dim() == 2 and input.stride(1) == 1
assert (
@@ -162,7 +163,8 @@ class SparseIndexerTopk(torch.nn.Module):
) -> str:
"""Resolve the decode top-k implementation from the configured
backend ("auto" = the pre-existing chain, or a validated explicit
value)."""
value).
"""
if self._backend == "auto":
return self._resolve_auto(logits, topk_tokens, num_rows)
@@ -209,7 +211,8 @@ class SparseIndexerTopk(torch.nn.Module):
self, logits: torch.Tensor, topk_tokens: int, num_rows: int
) -> str:
"""The pre-existing priority chain: cooperative -> persistent ->
per_row. deep_select/flashinfer/torch are opt-in only."""
per_row. deep_select/flashinfer/torch are opt-in only.
"""
if not self._cooperative_constraints(logits, topk_tokens, num_rows):
return "cooperative"
if self._is_cuda and topk_tokens in (512, 1024, 2048):
@@ -271,7 +274,8 @@ class SparseIndexerTopk(torch.nn.Module):
max_seq_len: int,
) -> None:
"""Run the resolved decode top-k implementation, writing into
topk_indices (int32, -1 fill for rows shorter than topk_tokens)."""
topk_indices (int32, -1 fill for rows shorter than topk_tokens).
"""
backend = self.resolve_backend(logits, topk_tokens, logits.shape[0])
if backend == "deep_select":
row_ends = self._row_ends(seq_lens, next_n, logits.shape[0])