[Model] Support DeepSeek-V4.1-Flash (#56214)

Signed-off-by: Yongye Zhu <zyy1102000@gmail.com>
Signed-off-by: Jee Jee Li <jeejeelee@inferact.ai>
Signed-off-by: andyluo7 <andy.luo@amd.com>
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: Jee Jee Li <jeejeelee@inferact.ai>
Co-authored-by: andyluo7 <andy.luo@amd.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Yongye Zhu
2026-09-11 02:11:19 -07:00
committed by GitHub
co-authored by Jee Jee Li andyluo7 OpenAI Codex zjy0516 Claude Opus 5
parent e6821ceac9
commit e77daef89e
49 changed files with 3996 additions and 138 deletions
@@ -8,7 +8,7 @@ expose_input_quant_key; the kernel validates and reads the activation via
as_quantized_activation.
"""
from dataclasses import dataclass
from dataclasses import dataclass, replace
import torch
@@ -34,6 +34,16 @@ class QuantizedActivation:
orig_shape: torch.Size
quant_key: QuantKey
def weak_ref(self) -> "QuantizedActivation":
"""Return a copy with non-owning tensor references for CUDA graph replay."""
from vllm.utils.torch_utils import weak_ref_tensor
return replace(
self,
data=weak_ref_tensor(self.data),
scale=weak_ref_tensor(self.scale),
)
def expose_input_quant_key(layer: torch.nn.Module, kernel) -> None:
"""Advertise the kernel's pre-quantized input key on the layer, if any.