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:
co-authored by
Claude Opus 5
parent
804e5377aa
commit
8c1557a79c
@@ -96,6 +96,7 @@ def _make_paged_kv_metadata(
|
||||
kv_indices – CUDA int32, shape [total_blocks]
|
||||
kv_last_page_lens – CPU int32, shape [num_seqs]
|
||||
block_tables – CUDA int32, shape [num_seqs, max_blocks_per_seq]
|
||||
|
||||
"""
|
||||
num_seqs = len(kv_lens)
|
||||
max_blocks = (max(kv_lens) + block_size - 1) // block_size
|
||||
@@ -161,7 +162,8 @@ def test_fast_decode_plan_importable() -> None:
|
||||
@torch.inference_mode
|
||||
def test_fast_plan_decode_warmup_uses_full_plan(dtype: torch.dtype) -> None:
|
||||
"""On the first call fast_plan_decode must route through self.plan() and
|
||||
flip vllm_first_call to False on the wrapper object."""
|
||||
flip vllm_first_call to False on the wrapper object.
|
||||
"""
|
||||
from unittest.mock import patch
|
||||
|
||||
from vllm.v1.attention.backends.flashinfer import fast_plan_decode
|
||||
|
||||
Reference in New Issue
Block a user