Fix per file ruff ignores related to typing (#26254)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
@@ -14,11 +14,8 @@ from typing import (
|
||||
Annotated,
|
||||
Any,
|
||||
Callable,
|
||||
Dict,
|
||||
List,
|
||||
Literal,
|
||||
Optional,
|
||||
Type,
|
||||
TypeVar,
|
||||
Union,
|
||||
cast,
|
||||
@@ -325,7 +322,7 @@ class EngineArgs:
|
||||
"""Arguments for vLLM engine."""
|
||||
|
||||
model: str = ModelConfig.model
|
||||
served_model_name: Optional[Union[str, List[str]]] = ModelConfig.served_model_name
|
||||
served_model_name: Optional[Union[str, list[str]]] = ModelConfig.served_model_name
|
||||
tokenizer: Optional[str] = ModelConfig.tokenizer
|
||||
hf_config_path: Optional[str] = ModelConfig.hf_config_path
|
||||
runner: RunnerOption = ModelConfig.runner
|
||||
@@ -350,7 +347,7 @@ class EngineArgs:
|
||||
# is intended for expert use only. The API may change without
|
||||
# notice.
|
||||
distributed_executor_backend: Optional[
|
||||
Union[str, DistributedExecutorBackend, Type[ExecutorBase]]
|
||||
Union[str, DistributedExecutorBackend, type[ExecutorBase]]
|
||||
] = ParallelConfig.distributed_executor_backend
|
||||
# number of P/D disaggregation (or other disaggregation) workers
|
||||
pipeline_parallel_size: int = ParallelConfig.pipeline_parallel_size
|
||||
@@ -418,7 +415,7 @@ class EngineArgs:
|
||||
media_io_kwargs: dict[str, dict[str, Any]] = get_field(
|
||||
MultiModalConfig, "media_io_kwargs"
|
||||
)
|
||||
mm_processor_kwargs: Optional[Dict[str, Any]] = MultiModalConfig.mm_processor_kwargs
|
||||
mm_processor_kwargs: Optional[dict[str, Any]] = MultiModalConfig.mm_processor_kwargs
|
||||
disable_mm_preprocessor_cache: bool = False # DEPRECATED
|
||||
mm_processor_cache_gb: float = MultiModalConfig.mm_processor_cache_gb
|
||||
mm_processor_cache_type: Optional[MMCacheType] = (
|
||||
@@ -436,7 +433,7 @@ class EngineArgs:
|
||||
enable_lora_bias: bool = LoRAConfig.bias_enabled
|
||||
max_loras: int = LoRAConfig.max_loras
|
||||
max_lora_rank: int = LoRAConfig.max_lora_rank
|
||||
default_mm_loras: Optional[Dict[str, str]] = LoRAConfig.default_mm_loras
|
||||
default_mm_loras: Optional[dict[str, str]] = LoRAConfig.default_mm_loras
|
||||
fully_sharded_loras: bool = LoRAConfig.fully_sharded_loras
|
||||
max_cpu_loras: Optional[int] = LoRAConfig.max_cpu_loras
|
||||
lora_dtype: Optional[Union[str, torch.dtype]] = LoRAConfig.lora_dtype
|
||||
@@ -446,7 +443,7 @@ class EngineArgs:
|
||||
num_gpu_blocks_override: Optional[int] = CacheConfig.num_gpu_blocks_override
|
||||
num_lookahead_slots: int = SchedulerConfig.num_lookahead_slots
|
||||
model_loader_extra_config: dict = get_field(LoadConfig, "model_loader_extra_config")
|
||||
ignore_patterns: Optional[Union[str, List[str]]] = LoadConfig.ignore_patterns
|
||||
ignore_patterns: Optional[Union[str, list[str]]] = LoadConfig.ignore_patterns
|
||||
|
||||
enable_chunked_prefill: Optional[bool] = SchedulerConfig.enable_chunked_prefill
|
||||
disable_chunked_mm_input: bool = SchedulerConfig.disable_chunked_mm_input
|
||||
@@ -467,7 +464,7 @@ class EngineArgs:
|
||||
|
||||
logits_processor_pattern: Optional[str] = ModelConfig.logits_processor_pattern
|
||||
|
||||
speculative_config: Optional[Dict[str, Any]] = None
|
||||
speculative_config: Optional[dict[str, Any]] = None
|
||||
|
||||
show_hidden_metrics_for_version: Optional[str] = (
|
||||
ObservabilityConfig.show_hidden_metrics_for_version
|
||||
@@ -477,7 +474,7 @@ class EngineArgs:
|
||||
ObservabilityConfig.collect_detailed_traces
|
||||
)
|
||||
scheduling_policy: SchedulerPolicy = SchedulerConfig.policy
|
||||
scheduler_cls: Union[str, Type[object]] = SchedulerConfig.scheduler_cls
|
||||
scheduler_cls: Union[str, type[object]] = SchedulerConfig.scheduler_cls
|
||||
|
||||
pooler_config: Optional[PoolerConfig] = ModelConfig.pooler_config
|
||||
override_pooler_config: Optional[Union[dict, PoolerConfig]] = (
|
||||
|
||||
+21
-21
@@ -3,7 +3,7 @@
|
||||
|
||||
import time
|
||||
from collections import Counter as CollectionsCounter
|
||||
from typing import Dict, List, Optional, Type, Union, cast
|
||||
from typing import Optional, Union, cast
|
||||
|
||||
import numpy as np
|
||||
import prometheus_client
|
||||
@@ -43,7 +43,7 @@ class Metrics:
|
||||
_counter_cls = prometheus_client.Counter
|
||||
_histogram_cls = prometheus_client.Histogram
|
||||
|
||||
def __init__(self, labelnames: List[str], vllm_config: VllmConfig):
|
||||
def __init__(self, labelnames: list[str], vllm_config: VllmConfig):
|
||||
# Unregister any existing vLLM collectors (for CI/CD)
|
||||
self._unregister_vllm_metrics()
|
||||
|
||||
@@ -304,7 +304,7 @@ class _RayGaugeWrapper:
|
||||
self,
|
||||
name: str,
|
||||
documentation: str = "",
|
||||
labelnames: Optional[List[str]] = None,
|
||||
labelnames: Optional[list[str]] = None,
|
||||
multiprocess_mode: str = "",
|
||||
):
|
||||
del multiprocess_mode
|
||||
@@ -330,7 +330,7 @@ class _RayCounterWrapper:
|
||||
prometheus_client.Counter"""
|
||||
|
||||
def __init__(
|
||||
self, name: str, documentation: str = "", labelnames: Optional[List[str]] = None
|
||||
self, name: str, documentation: str = "", labelnames: Optional[list[str]] = None
|
||||
):
|
||||
labelnames_tuple = tuple(labelnames) if labelnames else None
|
||||
self._counter = ray_metrics.Counter(
|
||||
@@ -355,8 +355,8 @@ class _RayHistogramWrapper:
|
||||
self,
|
||||
name: str,
|
||||
documentation: str = "",
|
||||
labelnames: Optional[List[str]] = None,
|
||||
buckets: Optional[List[float]] = None,
|
||||
labelnames: Optional[list[str]] = None,
|
||||
buckets: Optional[list[float]] = None,
|
||||
):
|
||||
labelnames_tuple = tuple(labelnames) if labelnames else None
|
||||
boundaries = buckets if buckets else []
|
||||
@@ -381,17 +381,17 @@ class RayMetrics(Metrics):
|
||||
Provides the same metrics as Metrics but uses Ray's util.metrics library.
|
||||
"""
|
||||
|
||||
_gauge_cls: Type[prometheus_client.Gauge] = cast(
|
||||
Type[prometheus_client.Gauge], _RayGaugeWrapper
|
||||
_gauge_cls: type[prometheus_client.Gauge] = cast(
|
||||
type[prometheus_client.Gauge], _RayGaugeWrapper
|
||||
)
|
||||
_counter_cls: Type[prometheus_client.Counter] = cast(
|
||||
Type[prometheus_client.Counter], _RayCounterWrapper
|
||||
_counter_cls: type[prometheus_client.Counter] = cast(
|
||||
type[prometheus_client.Counter], _RayCounterWrapper
|
||||
)
|
||||
_histogram_cls: Type[prometheus_client.Histogram] = cast(
|
||||
Type[prometheus_client.Histogram], _RayHistogramWrapper
|
||||
_histogram_cls: type[prometheus_client.Histogram] = cast(
|
||||
type[prometheus_client.Histogram], _RayHistogramWrapper
|
||||
)
|
||||
|
||||
def __init__(self, labelnames: List[str], vllm_config: VllmConfig):
|
||||
def __init__(self, labelnames: list[str], vllm_config: VllmConfig):
|
||||
if ray_metrics is None:
|
||||
raise ImportError("RayMetrics requires Ray to be installed.")
|
||||
super().__init__(labelnames, vllm_config)
|
||||
@@ -401,14 +401,14 @@ class RayMetrics(Metrics):
|
||||
pass
|
||||
|
||||
|
||||
def build_buckets(mantissa_lst: List[int], max_value: int) -> List[int]:
|
||||
def build_buckets(mantissa_lst: list[int], max_value: int) -> list[int]:
|
||||
"""
|
||||
Builds a list of buckets with increasing powers of 10 multiplied by
|
||||
mantissa values until the value exceeds the specified maximum.
|
||||
|
||||
"""
|
||||
exponent = 0
|
||||
buckets: List[int] = []
|
||||
buckets: list[int] = []
|
||||
while True:
|
||||
for m in mantissa_lst:
|
||||
value = m * 10**exponent
|
||||
@@ -419,7 +419,7 @@ def build_buckets(mantissa_lst: List[int], max_value: int) -> List[int]:
|
||||
exponent += 1
|
||||
|
||||
|
||||
def build_1_2_5_buckets(max_value: int) -> List[int]:
|
||||
def build_1_2_5_buckets(max_value: int) -> list[int]:
|
||||
"""
|
||||
Example:
|
||||
>>> build_1_2_5_buckets(100)
|
||||
@@ -428,7 +428,7 @@ def build_1_2_5_buckets(max_value: int) -> List[int]:
|
||||
return build_buckets([1, 2, 5], max_value)
|
||||
|
||||
|
||||
def build_1_2_3_5_8_buckets(max_value: int) -> List[int]:
|
||||
def build_1_2_3_5_8_buckets(max_value: int) -> list[int]:
|
||||
"""
|
||||
Example:
|
||||
>>> build_1_2_3_5_8_buckets(100)
|
||||
@@ -442,7 +442,7 @@ def local_interval_elapsed(now: float, last_log: float, local_interval: float) -
|
||||
return elapsed_time > local_interval
|
||||
|
||||
|
||||
def get_throughput(tracked_stats: List[int], now: float, last_log: float) -> float:
|
||||
def get_throughput(tracked_stats: list[int], now: float, last_log: float) -> float:
|
||||
return float(np.sum(tracked_stats) / (now - last_log))
|
||||
|
||||
|
||||
@@ -530,7 +530,7 @@ class PrometheusStatLogger(StatLoggerBase):
|
||||
_gauge_cls = prometheus_client.Gauge
|
||||
|
||||
def __init__(
|
||||
self, local_interval: float, labels: Dict[str, str], vllm_config: VllmConfig
|
||||
self, local_interval: float, labels: dict[str, str], vllm_config: VllmConfig
|
||||
) -> None:
|
||||
super().__init__(local_interval, vllm_config)
|
||||
# Prometheus metrics
|
||||
@@ -558,12 +558,12 @@ class PrometheusStatLogger(StatLoggerBase):
|
||||
for label, count in data.items():
|
||||
counter.labels(**{**self.labels, label_key: label}).inc(count)
|
||||
|
||||
def _log_histogram(self, histogram, data: Union[List[int], List[float]]) -> None:
|
||||
def _log_histogram(self, histogram, data: Union[list[int], list[float]]) -> None:
|
||||
# Convenience function for logging list to histogram.
|
||||
for datum in data:
|
||||
histogram.labels(**self.labels).observe(datum)
|
||||
|
||||
def _log_gauge_string(self, gauge, data: Dict[str, str]) -> None:
|
||||
def _log_gauge_string(self, gauge, data: dict[str, str]) -> None:
|
||||
gauge.labels(**data).set_to_current_time()
|
||||
|
||||
def _log_prometheus(self, stats: Stats) -> None:
|
||||
|
||||
@@ -16,7 +16,6 @@ do this in Python code and lazily import prometheus_client.
|
||||
import time
|
||||
from abc import ABC, abstractmethod
|
||||
from dataclasses import dataclass
|
||||
from typing import List
|
||||
|
||||
from vllm.config import SupportsMetricsInfo, VllmConfig
|
||||
|
||||
@@ -43,26 +42,26 @@ class Stats:
|
||||
num_prompt_tokens_iter: int
|
||||
num_generation_tokens_iter: int
|
||||
num_tokens_iter: int
|
||||
time_to_first_tokens_iter: List[float]
|
||||
inter_token_latencies_iter: List[float]
|
||||
time_to_first_tokens_iter: list[float]
|
||||
inter_token_latencies_iter: list[float]
|
||||
num_preemption_iter: int
|
||||
|
||||
# Request stats (should have _requests suffix)
|
||||
# Latency
|
||||
time_e2e_requests: List[float]
|
||||
time_queue_requests: List[float]
|
||||
time_inference_requests: List[float]
|
||||
time_prefill_requests: List[float]
|
||||
time_decode_requests: List[float]
|
||||
time_e2e_requests: list[float]
|
||||
time_queue_requests: list[float]
|
||||
time_inference_requests: list[float]
|
||||
time_prefill_requests: list[float]
|
||||
time_decode_requests: list[float]
|
||||
# Metadata
|
||||
num_prompt_tokens_requests: List[int]
|
||||
num_generation_tokens_requests: List[int]
|
||||
n_requests: List[int]
|
||||
max_num_generation_tokens_requests: List[int]
|
||||
max_tokens_requests: List[int]
|
||||
finished_reason_requests: List[str]
|
||||
waiting_lora_adapters: List[str]
|
||||
running_lora_adapters: List[str]
|
||||
num_prompt_tokens_requests: list[int]
|
||||
num_generation_tokens_requests: list[int]
|
||||
n_requests: list[int]
|
||||
max_num_generation_tokens_requests: list[int]
|
||||
max_tokens_requests: list[int]
|
||||
finished_reason_requests: list[str]
|
||||
waiting_lora_adapters: list[str]
|
||||
running_lora_adapters: list[str]
|
||||
max_lora: str
|
||||
|
||||
|
||||
@@ -71,8 +70,8 @@ class StatLoggerBase(ABC):
|
||||
|
||||
def __init__(self, local_interval: float, vllm_config: VllmConfig) -> None:
|
||||
# Tracked stats over current local logging interval.
|
||||
self.num_prompt_tokens: List[int] = []
|
||||
self.num_generation_tokens: List[int] = []
|
||||
self.num_prompt_tokens: list[int] = []
|
||||
self.num_generation_tokens: list[int] = []
|
||||
self.last_local_log = time.time()
|
||||
self.local_interval = local_interval
|
||||
|
||||
|
||||
Reference in New Issue
Block a user