Update Optional[x] -> x | None and Union[x, y] to x | y (#26633)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2025-10-12 09:51:31 -07:00
committed by GitHub
parent 9bb38130cb
commit 8fcaaf6a16
944 changed files with 9490 additions and 10121 deletions
+9 -9
View File
@@ -2,7 +2,7 @@
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
# Datastructures defining a TPU input batch
from typing import Optional, cast
from typing import cast
import numpy as np
import torch
@@ -36,7 +36,7 @@ class InputBatch:
self.pin_memory = pin_memory
self.vocab_size = vocab_size
self._req_ids: list[Optional[str]] = []
self._req_ids: list[str | None] = []
self.req_id_to_index: dict[str, int] = {}
# TODO(woosuk): This buffer could be too large if max_model_len is big.
@@ -155,17 +155,17 @@ class InputBatch:
# To accumulate prompt logprobs tensor chunks across prefill steps.
self.in_progress_prompt_logprobs_cpu: dict[str, LogprobsTensors] = {}
self.logit_bias: list[Optional[dict[int, float]]] = [None] * max_num_reqs
self.logit_bias: list[dict[int, float] | None] = [None] * max_num_reqs
self.has_allowed_token_ids: set[str] = set()
# NOTE(lufang): In the mask tensor, if the corresponding token allowed,
# the value is False. Since we use masked_fill_ to set -inf.
self.allowed_token_ids_mask: Optional[torch.Tensor] = None
self.allowed_token_ids_mask_cpu_tensor: Optional[torch.Tensor] = None
self.allowed_token_ids_mask: torch.Tensor | None = None
self.allowed_token_ids_mask_cpu_tensor: torch.Tensor | None = None
# req_index -> bad_words_token_ids
self.bad_words_token_ids: dict[int, list[list[int]]] = {}
self.req_output_token_ids: list[Optional[list[int]]] = []
self.req_output_token_ids: list[list[int] | None] = []
@property
def req_ids(self) -> list[str]:
@@ -176,7 +176,7 @@ class InputBatch:
def add_request(
self,
request: "CachedRequestState",
req_index: Optional[int] = None,
req_index: int | None = None,
) -> None:
if req_index is None:
req_index = self.num_reqs
@@ -296,7 +296,7 @@ class InputBatch:
# No LoRA
self.request_lora_mapping[req_index] = 0
def remove_request(self, req_id: str) -> Optional[int]:
def remove_request(self, req_id: str) -> int | None:
"""This method must always be followed by a call to condense()."""
req_index = self.req_id_to_index.pop(req_id, None)
@@ -580,7 +580,7 @@ class InputBatch:
)
@property
def max_num_logprobs(self) -> Optional[int]:
def max_num_logprobs(self) -> int | None:
return max(self.num_logprobs.values()) if self.num_logprobs else None
@property