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 14, 05:30 UTC
[Misc] Remove typing.List (#26150)
Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com> Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Varun Sundar Rabindranath committed
Oct 3, 2025 at 03:00 UTC
8c9117181d58f552f5bdb236f093123b6837dfb0
1 file changed
+3
-6
vllm/model_executor/layers/fused_moe/fused_moe.py
+3
-6
@@ -4,9 +4,6 @@
4
import functools
5
import json
6
import os
7
-# torch.compile needs typing.List. It will fail torch.library.infer_schema
8
-# otherwise
9
-from typing import List # noqa: UP035
7
from typing import Any, Callable, Optional, Union
8
9
import torch
@@ -1229,7 +1226,7 @@ def inplace_fused_experts(
1226
w2_zp: Optional[torch.Tensor] = None,
1227
a1_scale: Optional[torch.Tensor] = None,
1228
a2_scale: Optional[torch.Tensor] = None,
1232
- block_shape: Optional[List[int]] = None, #noqa: UP006
1229
+ block_shape: Optional[list[int]] = None,
1230
w1_bias: Optional[torch.Tensor] = None,
1231
w2_bias: Optional[torch.Tensor] = None,
1232
) -> None:
@@ -1263,7 +1260,7 @@ def inplace_fused_experts_fake(
1260
w2_zp: Optional[torch.Tensor] = None,
1261
a1_scale: Optional[torch.Tensor] = None,
1262
a2_scale: Optional[torch.Tensor] = None,
1266
- block_shape: Optional[List[int]] = None, #noqa: UP006
1263
+ block_shape: Optional[list[int]] = None,
1264
w1_bias: Optional[torch.Tensor] = None,
1265
w2_bias: Optional[torch.Tensor] = None,
1266
) -> None:
@@ -1302,7 +1299,7 @@ def outplace_fused_experts(
1299
w2_zp: Optional[torch.Tensor] = None,
1300
a1_scale: Optional[torch.Tensor] = None,
1301
a2_scale: Optional[torch.Tensor] = None,
1305
- block_shape: Optional[List[int]] = None, #noqa: UP006
1302
+ block_shape: Optional[list[int]] = None,
1303
w1_bias: Optional[torch.Tensor] = None,
1304
w2_bias: Optional[torch.Tensor] = None,
1305
) -> torch.Tensor: