biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

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

Browse Source 
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2025-10-12 17:51:31 +01:00
committed by GitHub
parent 9bb38130cb
commit 8fcaaf6a16
944 changed files with 9490 additions and 10121 deletions
Download Patch File Download Diff File Expand all files Collapse all files

39
vllm/model_executor/models/qwen3_moe.py
View File

@@ -26,7 +26,7 @@
import typing
from collections.abc import Callable, Iterable
from itertools import islice
from typing import Any, Optional, Union
from typing import Any
import torch
from torch import nn
@@ -84,7 +84,7 @@ class Qwen3MoeMLP(nn.Module):
hidden_size: int,
intermediate_size: int,
hidden_act: str,
quant_config: Optional[QuantizationConfig] = None,
quant_config: QuantizationConfig | None = None,
reduce_results: bool = True,
prefix: str = "",
) -> None:
@@ -215,15 +215,15 @@ class Qwen3MoeAttention(nn.Module):
num_heads: int,
num_kv_heads: int,
rope_theta: float = 10000,
rope_scaling: Optional[dict[str, Any]] = None,
rope_scaling: dict[str, Any] | None = None,
max_position_embeddings: int = 8192,
head_dim: Optional[int] = None,
head_dim: int | None = None,
rms_norm_eps: float = 1e-06,
qkv_bias: bool = False,
cache_config: Optional[CacheConfig] = None,
quant_config: Optional[QuantizationConfig] = None,
cache_config: CacheConfig | None = None,
quant_config: QuantizationConfig | None = None,
prefix: str = "",
dual_chunk_attention_config: Optional[dict[str, Any]] = None,
dual_chunk_attention_config: dict[str, Any] | None = None,
) -> None:
super().__init__()
self.hidden_size = hidden_size
@@ -374,7 +374,7 @@ class Qwen3MoeDecoderLayer(nn.Module):
self,
positions: torch.Tensor,
hidden_states: torch.Tensor,
residual: Optional[torch.Tensor],
residual: torch.Tensor | None,
) -> tuple[torch.Tensor, torch.Tensor]:
# Self Attention
if residual is None:
@@ -432,11 +432,9 @@ class Qwen3MoeModel(nn.Module):
self,
input_ids: torch.Tensor,
positions: torch.Tensor,
intermediate_tensors: Optional[IntermediateTensors] = None,
inputs_embeds: Optional[torch.Tensor] = None,
) -> Union[
torch.Tensor, IntermediateTensors, tuple[torch.Tensor, list[torch.Tensor]]
]:
intermediate_tensors: IntermediateTensors | None = None,
inputs_embeds: torch.Tensor | None = None,
) -> torch.Tensor | IntermediateTensors | tuple[torch.Tensor, list[torch.Tensor]]:
if get_pp_group().is_first_rank:
if inputs_embeds is not None:
hidden_states = inputs_embeds
@@ -647,12 +645,7 @@ class Qwen3MoeForCausalLM(
self.quant_config = quant_config
# Only perform the following mapping when Qwen3MoeMLP exists
if getattr(config, "mlp_only_layers", []):
self.packed_modules_mapping["gate_up_proj"] = (
[
"gate_proj",
"up_proj",
],
)
self.packed_modules_mapping["gate_up_proj"] = ["gate_proj", "up_proj"]
self.model = Qwen3MoeModel(
vllm_config=vllm_config, prefix=maybe_prefix(prefix, "model")
)
@@ -742,9 +735,9 @@ class Qwen3MoeForCausalLM(
self,
input_ids: torch.Tensor,
positions: torch.Tensor,
intermediate_tensors: Optional[IntermediateTensors] = None,
inputs_embeds: Optional[torch.Tensor] = None,
) -> Union[torch.Tensor, IntermediateTensors]:
intermediate_tensors: IntermediateTensors | None = None,
inputs_embeds: torch.Tensor | None = None,
) -> torch.Tensor | IntermediateTensors:
hidden_states = self.model(
input_ids, positions, intermediate_tensors, inputs_embeds
)
@@ -753,7 +746,7 @@ class Qwen3MoeForCausalLM(
def compute_logits(
self,
hidden_states: torch.Tensor,
) -> Optional[torch.Tensor]:
) -> torch.Tensor | None:
logits = self.logits_processor(self.lm_head, hidden_states)
return logits