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:
@@ -1,7 +1,8 @@
|
||||
# SPDX-License-Identifier: Apache-2.0
|
||||
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
|
||||
|
||||
from typing import Any, Callable, Optional, Union
|
||||
from collections.abc import Callable
|
||||
from typing import Any, Optional
|
||||
|
||||
import torch
|
||||
|
||||
@@ -40,7 +41,7 @@ class MoeWNA16Config(QuantizationConfig):
|
||||
group_size: int,
|
||||
has_zp: bool,
|
||||
lm_head_quantized: bool,
|
||||
modules_to_not_convert: Optional[list[str]],
|
||||
modules_to_not_convert: list[str] | None,
|
||||
full_config: dict[str, Any],
|
||||
) -> None:
|
||||
super().__init__()
|
||||
@@ -127,7 +128,7 @@ class MoeWNA16Config(QuantizationConfig):
|
||||
@classmethod
|
||||
def override_quantization_method(
|
||||
cls, hf_quant_cfg, user_quant
|
||||
) -> Optional[QuantizationMethods]:
|
||||
) -> QuantizationMethods | None:
|
||||
can_convert = cls.is_moe_wna16_compatible(hf_quant_cfg)
|
||||
if can_convert and user_quant == "moe_wna16":
|
||||
return cls.get_name()
|
||||
@@ -339,7 +340,7 @@ class MoeWNA16Method(FusedMoEMethodBase):
|
||||
|
||||
def get_fused_moe_quant_config(
|
||||
self, layer: torch.nn.Module
|
||||
) -> Optional[FusedMoEQuantConfig]:
|
||||
) -> FusedMoEQuantConfig | None:
|
||||
weight_bits = self.quant_config.weight_bits
|
||||
has_zp = self.quant_config.has_zp
|
||||
assert weight_bits == 4 or weight_bits == 8
|
||||
@@ -365,21 +366,21 @@ class MoeWNA16Method(FusedMoEMethodBase):
|
||||
top_k: int,
|
||||
renormalize: bool,
|
||||
use_grouped_topk: bool = False,
|
||||
topk_group: Optional[int] = None,
|
||||
num_expert_group: Optional[int] = None,
|
||||
topk_group: int | None = None,
|
||||
num_expert_group: int | None = None,
|
||||
global_num_experts: int = -1,
|
||||
expert_map: Optional[torch.Tensor] = None,
|
||||
custom_routing_function: Optional[Callable] = None,
|
||||
expert_map: torch.Tensor | None = None,
|
||||
custom_routing_function: Callable | None = None,
|
||||
scoring_func: str = "softmax",
|
||||
routed_scaling_factor: float = 1.0,
|
||||
e_score_correction_bias: Optional[torch.Tensor] = None,
|
||||
e_score_correction_bias: torch.Tensor | None = None,
|
||||
apply_router_weight_on_input: bool = False,
|
||||
activation: str = "silu",
|
||||
enable_eplb: bool = False,
|
||||
expert_load_view: Optional[torch.Tensor] = None,
|
||||
logical_to_physical_map: Optional[torch.Tensor] = None,
|
||||
logical_replica_count: Optional[torch.Tensor] = None,
|
||||
) -> Union[torch.Tensor, tuple[torch.Tensor, torch.Tensor]]:
|
||||
expert_load_view: torch.Tensor | None = None,
|
||||
logical_to_physical_map: torch.Tensor | None = None,
|
||||
logical_replica_count: torch.Tensor | None = None,
|
||||
) -> torch.Tensor | tuple[torch.Tensor, torch.Tensor]:
|
||||
assert self.fused_experts is None
|
||||
if enable_eplb:
|
||||
raise NotImplementedError("EPLB not supported for `MoeWNA16Method` yet.")
|
||||
|
||||
Reference in New Issue
Block a user