A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 10, 16:30 UTC
[Chore] Remove more V0 dead code from `sequence.py` (#31783)
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Cyrus Leung committed
Jan 6, 2026 at 18:25 UTC
97ca4c3b60fee59850f5a0d83fc8d899ae4dd0e6
2 files changed
+1
-36
vllm/outputs.py
+1
-2
@@ -13,7 +13,6 @@ from vllm.logger import init_logger
13
from vllm.logprobs import PromptLogprobs, SampleLogprobs
14
from vllm.lora.request import LoRARequest
15
from vllm.multimodal.inputs import MultiModalPlaceholderDict
16
-from vllm.sequence import RequestMetrics
16
from vllm.v1.metrics.stats import RequestStateStats
17
18
logger = init_logger(__name__)
@@ -113,7 +112,7 @@ class RequestOutput:
112
prompt_logprobs: PromptLogprobs | None,
113
outputs: list[CompletionOutput],
114
finished: bool,
116
- metrics: RequestMetrics | RequestStateStats | None = None,
115
+ metrics: RequestStateStats | None = None,
116
lora_request: LoRARequest | None = None,
117
encoder_prompt: str | None = None,
118
encoder_prompt_token_ids: list[int] | None = None,
vllm/sequence.py
-34
@@ -12,40 +12,6 @@ if TYPE_CHECKING:
12
else:
13
KVConnectorOutput = Any
14
15
-VLLM_TOKEN_ID_ARRAY_TYPE = "l"
16
-
17
-VLLM_INVALID_TOKEN_ID = -1
18
-
19
-
20
-@dataclass
21
-class RequestMetrics:
22
- """Metrics associated with a request.
23
-
24
- Attributes:
25
- arrival_time: The time when the request arrived.
26
- first_scheduled_time: The time when the request was first scheduled.
27
- first_token_time: The time when the first token was generated.
28
- time_in_queue: The time the request spent in the queue.
29
- finished_time: The time when the request was finished.
30
- scheduler_time: The time spent in the scheduler when this request was
31
- being considered by the scheduler.
32
- model_forward_time: The time spent in the model forward pass when this
33
- request was in the batch.
34
- model_execute_time: The time spent in the model execute function. This
35
- will include model forward, block/sync across
36
- workers, cpu-gpu sync time and sampling time.
37
- """
38
-
39
- arrival_time: float
40
- last_token_time: float
41
- first_scheduled_time: float | None
42
- first_token_time: float | None
43
- time_in_queue: float | None
44
- finished_time: float | None = None
45
- scheduler_time: float | None = None
46
- model_forward_time: float | None = None
47
- model_execute_time: float | None = None
48
-
15
16
# cannot use msgspec.Struct here because Dynamo does not support it
17
@dataclass