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
@@ -39,7 +39,6 @@ def _create_vllm_config_for_dsd(
a schedule covering every query length in ``[1, max_decode_query_len]`` is
generated.
"""
max_decode_query_len = max_spec_tokens + 1
max_capture_tokens = max_num_seqs * max_decode_query_len
@@ -91,7 +90,6 @@ def test_dynamic_sd_full_cudagraph_covers_all_uniform_decode_shapes(monkeypatch)
shapes have been built, dispatch() should pick a FULL graph for every
uniform decode batch shape produced by DSD up to max_num_seqs.
"""
max_num_seqs = 512
max_spec_tokens = 7
max_decode_query_len = max_spec_tokens + 1
@@ -190,7 +188,6 @@ def test_dynamic_sd_non_uniform_batch_falls_back_to_piecewise(monkeypatch):
FULL candidates should be skipped in favor of the mixed-batch PIECEWISE
graph under FULL_AND_PIECEWISE mode.
"""
max_spec_tokens = 4
monkeypatch.setattr(
@@ -236,7 +233,6 @@ def test_prompt_chunks_shaped_like_spec_decode_miss_the_full_graph(monkeypatch):
by coincidence; replaying the FULL graph over them corrupts every request
in the batch. See https://github.com/vllm-project/vllm/issues/49918.
"""
max_spec_tokens = 7
decode_query_len = max_spec_tokens + 1
@@ -307,7 +303,6 @@ def test_basic_sd_does_not_capture_shorter_full_decode_shapes(monkeypatch):
Uniform batches with smaller query lengths should therefore miss the FULL
path entirely when using FULL_AND_PIECEWISE.
"""
max_num_seqs = 512
max_spec_tokens = 7
max_decode_query_len = max_spec_tokens + 1
@@ -368,7 +363,6 @@ def test_dynamic_sd_only_captures_scheduled_query_lengths(monkeypatch):
FULL graphs, while every other query length (e.g. the lower values 1, 2, 3)
must fall back to the mixed-batch PIECEWISE graph.
"""
max_num_seqs = 128
max_spec_tokens = 7
max_decode_query_len = max_spec_tokens + 1