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
@@ -98,7 +98,7 @@ class LogitsProcsRequestParams:
|
||||
self.params = _sampling_params_from_logitproc(logitproc_type)
|
||||
|
||||
def __str__(self):
|
||||
"""For debugging"""
|
||||
"""For debugging."""
|
||||
summ = ", ".join(f"{k}={v}" for k, v in vars(self).items())
|
||||
return f"MyClass({summ})"
|
||||
|
||||
@@ -117,7 +117,7 @@ def _generate_fake_sampling_metadata(
|
||||
vocab_size: int,
|
||||
device: torch.device,
|
||||
) -> SamplingMetadata:
|
||||
"""Generate fake sampling metadata with fake logitsprocs"""
|
||||
"""Generate fake sampling metadata with fake logitsprocs."""
|
||||
output_token_ids: list[list[int]] = []
|
||||
prompt_token_ids: list[list[int]] = []
|
||||
for _ in range(batch_size):
|
||||
@@ -176,7 +176,7 @@ def _generate_fake_sampling_metadata(
|
||||
|
||||
|
||||
def _generate_test_fakes(batch_size: int, device: str) -> LogitsprocsTestFakes:
|
||||
"""Generate fake logits and sampling metadata"""
|
||||
"""Generate fake logits and sampling metadata."""
|
||||
fake_logits = create_fake_logits(batch_size, VOCAB_SIZE)
|
||||
# Create one dominant token per batch, to support min-p test
|
||||
for i in range(batch_size):
|
||||
@@ -192,7 +192,7 @@ def _generate_test_fakes(batch_size: int, device: str) -> LogitsprocsTestFakes:
|
||||
|
||||
|
||||
def _sampling_params_from_logitproc(logitproc_type: LogitprocType) -> SamplingParams:
|
||||
"""Customize request SamplingParams for a specified logitproc"""
|
||||
"""Customize request SamplingParams for a specified logitproc."""
|
||||
# SamplingParams for req with no logitproc
|
||||
kwargs = {"min_p": 0.0, "logit_bias": None, "min_tokens": 0}
|
||||
if fxn := logitsprocs_test_mapping[logitproc_type].gen_request_fxn:
|
||||
@@ -220,6 +220,7 @@ def _generate_mixed_logitsprocs_batch_params(
|
||||
Returns:
|
||||
List of per-request params which configure the engine for that request's
|
||||
enabled logitproc
|
||||
|
||||
"""
|
||||
batch_size = len(logitsprocs_types) * reqs_per_logitproc
|
||||
# Generate multiple repeats of key params for each logitproc;
|
||||
@@ -251,7 +252,7 @@ def _raise_error_invalid(
|
||||
|
||||
|
||||
def _logit_bias_params(kwargs: dict) -> None:
|
||||
"""Logit bias config"""
|
||||
"""Logit bias config."""
|
||||
kwargs["logit_bias"] = {
|
||||
random.randint(0, VOCAB_SIZE - 1): random.choice([-0.1, 0.2])
|
||||
}
|
||||
@@ -265,7 +266,7 @@ def _logit_bias_validate(
|
||||
request_params: LogitsProcsRequestParams,
|
||||
step_idx: int,
|
||||
) -> None:
|
||||
"""Validate logit bias logitproc applied correctly"""
|
||||
"""Validate logit bias logitproc applied correctly."""
|
||||
logit_bias = request_params.params.logit_bias
|
||||
logits_old = test_fakes.logits[persistent_batch[batch_index].workload_index].cpu()
|
||||
logits_new = logits_new[batch_index].cpu()
|
||||
@@ -301,7 +302,7 @@ def _logit_bias_validate(
|
||||
|
||||
|
||||
def _min_p_params(kwargs: dict) -> None:
|
||||
"""Min-p logitproc config"""
|
||||
"""Min-p logitproc config."""
|
||||
kwargs["min_p"] = 0.1
|
||||
|
||||
|
||||
@@ -313,7 +314,7 @@ def _min_p_validate(
|
||||
request_params: LogitsProcsRequestParams,
|
||||
step_idx: int,
|
||||
) -> None:
|
||||
"""Validate min-p logitproc applied correctly"""
|
||||
"""Validate min-p logitproc applied correctly."""
|
||||
for token_id in range(VOCAB_SIZE):
|
||||
logits_for_token = logits_new[batch_index][token_id]
|
||||
if token_id == 0:
|
||||
@@ -347,7 +348,7 @@ def _min_p_validate(
|
||||
|
||||
|
||||
def _min_tokens_params(kwargs: dict) -> None:
|
||||
"""Min-tokens logitproc config"""
|
||||
"""Min-tokens logitproc config."""
|
||||
kwargs["min_tokens"] = MIN_TOKENS_LEN_THRESHOLD
|
||||
kwargs["stop_token_ids"] = [
|
||||
np.random.randint(0, VOCAB_SIZE - 1)
|
||||
@@ -363,7 +364,7 @@ def _min_tokens_validate(
|
||||
request_params: LogitsProcsRequestParams,
|
||||
step_idx: int,
|
||||
) -> None:
|
||||
"""Validate min-tokens logitsproc applied correctly"""
|
||||
"""Validate min-tokens logitsproc applied correctly."""
|
||||
ref_num_out_tokens = len(request_params.out_tokens)
|
||||
min_reached = ref_num_out_tokens >= MIN_TOKENS_LEN_THRESHOLD
|
||||
ref_all_stop_token_ids = request_params.params.all_stop_token_ids
|
||||
@@ -560,7 +561,7 @@ def test_min_tokens_restores_all_masked_structured_output_stop_token_spec_decode
|
||||
|
||||
|
||||
def _thinking_budget_params(kwargs: dict) -> None:
|
||||
"""Set SamplingParams kwargs for thinking token budget tests"""
|
||||
"""Set SamplingParams kwargs for thinking token budget tests."""
|
||||
kwargs["thinking_token_budget"] = THINKING_TOKEN_BUDGET
|
||||
|
||||
|
||||
@@ -661,7 +662,7 @@ def _none_validate(
|
||||
request_params: LogitsProcsRequestParams,
|
||||
step_idx: int,
|
||||
) -> None:
|
||||
"""Validate that no logits processors are applied"""
|
||||
"""Validate that no logits processors are applied."""
|
||||
logits = test_fakes.logits[persistent_batch[batch_index].workload_index].cpu()
|
||||
ref_logits = logits_new[batch_index]
|
||||
if not torch.all(ref_logits == logits):
|
||||
@@ -706,7 +707,7 @@ logitsprocs_test_mapping = {
|
||||
|
||||
|
||||
def _get_test_cases() -> list[list[str]]:
|
||||
"""Each test case is a set of logitsprocs"""
|
||||
"""Each test case is a set of logitsprocs."""
|
||||
logitsprocs_types = list(logitsprocs_test_mapping.keys())
|
||||
|
||||
# Isolate thinking-budget handling from other processors to avoid cross-talk.
|
||||
@@ -1343,7 +1344,8 @@ def test_thinking_budget_invalid_budget_rejected(invalid_budget):
|
||||
|
||||
def test_thinking_budget_long_thinking_section_end_marker_found_at_correct_index():
|
||||
"""Test thinking budget enforced for a long thinking run,
|
||||
then a natural end marker."""
|
||||
then a natural end marker.
|
||||
"""
|
||||
h = ThinkingBudgetStateHolder(
|
||||
MockReasoningConfig(), 8, 0, torch.device("cpu"), False
|
||||
)
|
||||
|
||||
@@ -86,8 +86,8 @@ def _run_test(kwargs: dict, logitproc_loaded: bool) -> None:
|
||||
Args:
|
||||
kwargs: `LLM` constructor kwargs
|
||||
logitproc_loaded: server has loaded dummy logitproc if True
|
||||
"""
|
||||
|
||||
"""
|
||||
# Create a vLLM instance and load custom logitproc
|
||||
llm_logitproc = LLM(
|
||||
model=MODEL_NAME,
|
||||
@@ -131,7 +131,7 @@ def _run_test(kwargs: dict, logitproc_loaded: bool) -> None:
|
||||
@create_new_process_for_each_test()
|
||||
@pytest.mark.parametrize("logitproc_source", list(CustomLogitprocSource))
|
||||
def test_custom_logitsprocs(monkeypatch, logitproc_source: CustomLogitprocSource):
|
||||
"""Test offline Python interface for passing custom logitsprocs
|
||||
"""Test offline Python interface for passing custom logitsprocs.
|
||||
|
||||
Construct an `LLM` instance which loads a custom logitproc that has a
|
||||
well-defined behavior (mask out all tokens except one `target_token`)
|
||||
@@ -158,8 +158,8 @@ def test_custom_logitsprocs(monkeypatch, logitproc_source: CustomLogitprocSource
|
||||
logitproc_source: what source (entrypoint, fully-qualified class name
|
||||
(FQCN), class object, or None) the user pulls the
|
||||
logitproc from
|
||||
"""
|
||||
|
||||
"""
|
||||
# Test that logitproc info is passed to workers
|
||||
monkeypatch.setenv("VLLM_ENABLE_V1_MULTIPROCESSING", "1")
|
||||
set_random_seed(40)
|
||||
@@ -192,7 +192,7 @@ def test_custom_logitsprocs(monkeypatch, logitproc_source: CustomLogitprocSource
|
||||
|
||||
@create_new_process_for_each_test()
|
||||
def test_custom_logitsprocs_req(monkeypatch):
|
||||
"""Test passing request-level logits processor to offline Python interface
|
||||
"""Test passing request-level logits processor to offline Python interface.
|
||||
|
||||
Wrap a request-level logits processor to create a batch level logits
|
||||
processor that has a well-defined behavior (mask out all tokens except one
|
||||
@@ -213,8 +213,8 @@ def test_custom_logitsprocs_req(monkeypatch):
|
||||
|
||||
Args:
|
||||
monkeypatch: for setting env vars
|
||||
"""
|
||||
|
||||
"""
|
||||
# Test that logitproc info is passed to workers
|
||||
monkeypatch.setenv("VLLM_ENABLE_V1_MULTIPROCESSING", "1")
|
||||
set_random_seed(40)
|
||||
@@ -258,6 +258,7 @@ def test_rejects_custom_logitsprocs(
|
||||
logitproc_source: what source (entrypoint, fully-qualified class name
|
||||
(FQCN), or class object) the user pulls the
|
||||
logitproc from
|
||||
|
||||
"""
|
||||
monkeypatch.setenv("VLLM_ENABLE_V1_MULTIPROCESSING", "0")
|
||||
set_random_seed(40)
|
||||
|
||||
@@ -77,7 +77,6 @@ def server(default_server_args, request, monkeypatch):
|
||||
(2) No --logits-processors cli arg; inject a dummy logits processor
|
||||
entrypoint
|
||||
"""
|
||||
|
||||
# Test that logitproc info is passed to workers
|
||||
monkeypatch.setenv("VLLM_ENABLE_V1_MULTIPROCESSING", "1")
|
||||
|
||||
@@ -118,7 +117,7 @@ api_keyword_args = {
|
||||
[MODEL_NAME],
|
||||
)
|
||||
def test_custom_logitsprocs(server, model_name: str):
|
||||
"""Test custom logitsprocs when starting OpenAI server from CLI
|
||||
"""Test custom logitsprocs when starting OpenAI server from CLI.
|
||||
|
||||
Launch vLLM OpenAI-compatible server, configured to load a custom logitproc
|
||||
that has a well-defined behavior (mask out all tokens except one
|
||||
@@ -130,7 +129,6 @@ def test_custom_logitsprocs(server, model_name: str):
|
||||
Validate that requests which activate the custom logitproc, repeat the same
|
||||
token
|
||||
"""
|
||||
|
||||
import asyncio
|
||||
|
||||
async def _async_main(srv, mn):
|
||||
@@ -180,8 +178,7 @@ def test_custom_logitsprocs(server, model_name: str):
|
||||
async def test_invalid_custom_logitsproc_arg(
|
||||
client: openai.AsyncOpenAI, model_name: str
|
||||
):
|
||||
"""Test that request with invalid custom logitsproc is rejected"""
|
||||
|
||||
"""Test that request with invalid custom logitsproc is rejected."""
|
||||
prompt = "Hello, my name is"
|
||||
# Pass invalid (non-int) target_token value to dummy logits processor
|
||||
request_keyword_args: dict[str, Any] = {
|
||||
|
||||
@@ -37,7 +37,7 @@ DUMMY_LOGITPROC_FQCN = f"{DUMMY_LOGITPROC_MODULE}:DummyLogitsProcessor"
|
||||
|
||||
|
||||
class CustomLogitprocSource(Enum):
|
||||
"""How to source a logitproc for testing purposes"""
|
||||
"""How to source a logitproc for testing purposes."""
|
||||
|
||||
LOGITPROC_SOURCE_NONE = auto() # No custom logitproc
|
||||
LOGITPROC_SOURCE_ENTRYPOINT = auto() # Via entrypoint
|
||||
@@ -55,7 +55,7 @@ prompts = [
|
||||
|
||||
|
||||
class DummyLogitsProcessor(LogitsProcessor):
|
||||
"""Fake logit processor to support unit testing and examples"""
|
||||
"""Fake logit processor to support unit testing and examples."""
|
||||
|
||||
@classmethod
|
||||
def validate_params(cls, params: SamplingParams):
|
||||
@@ -73,7 +73,7 @@ class DummyLogitsProcessor(LogitsProcessor):
|
||||
self.req_info: dict[int, int] = {}
|
||||
|
||||
def is_argmax_invariant(self) -> bool:
|
||||
"""Never impacts greedy sampling"""
|
||||
"""Never impacts greedy sampling."""
|
||||
return False
|
||||
|
||||
def update_state(self, batch_update: BatchUpdate | None):
|
||||
@@ -115,7 +115,7 @@ class DummyLogitsProcessor(LogitsProcessor):
|
||||
|
||||
|
||||
class EntryPoint:
|
||||
"""Dummy entrypoint class for logitsprocs testing"""
|
||||
"""Dummy entrypoint class for logitsprocs testing."""
|
||||
|
||||
def __init__(self):
|
||||
self.name = DUMMY_LOGITPROC_ENTRYPOINT
|
||||
@@ -126,7 +126,7 @@ class EntryPoint:
|
||||
|
||||
|
||||
class EntryPoints(list):
|
||||
"""Dummy EntryPoints class for logitsprocs testing"""
|
||||
"""Dummy EntryPoints class for logitsprocs testing."""
|
||||
|
||||
def __init__(self, group: str):
|
||||
# Emulate list-like functionality
|
||||
@@ -138,10 +138,11 @@ class EntryPoints(list):
|
||||
|
||||
class DummyPerReqLogitsProcessor:
|
||||
"""The request-level logits processor masks out all logits except the
|
||||
token id identified by `target_token`"""
|
||||
token id identified by `target_token`
|
||||
"""
|
||||
|
||||
def __init__(self, target_token: int) -> None:
|
||||
"""Specify `target_token`"""
|
||||
"""Specify `target_token`."""
|
||||
self.target_token = target_token
|
||||
|
||||
def __call__(
|
||||
@@ -157,7 +158,8 @@ class DummyPerReqLogitsProcessor:
|
||||
|
||||
class WrappedPerReqLogitsProcessor(AdapterLogitsProcessor):
|
||||
"""Example of wrapping a fake request-level logit processor to create a
|
||||
batch-level logits processor"""
|
||||
batch-level logits processor
|
||||
"""
|
||||
|
||||
def is_argmax_invariant(self) -> bool:
|
||||
return False
|
||||
@@ -178,6 +180,7 @@ class WrappedPerReqLogitsProcessor(AdapterLogitsProcessor):
|
||||
|
||||
Returns:
|
||||
`Callable` request logits processor, or None
|
||||
|
||||
"""
|
||||
target_token: Any | None = params.extra_args and params.extra_args.get(
|
||||
"target_token"
|
||||
|
||||
Reference in New Issue
Block a user