[Fusion] Support 3D activation quantization fusion and enable it for Pixtral (#59612)

Signed-off-by: Jakub Zakrzewski <jzakrzewski@nvidia.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Misha Goin <mgoin64@gmail.com>
This commit is contained in:
Jakub Zakrzewski
2026-10-09 15:12:03 +08:00
committed by GitHub
co-authored by Codex Misha Goin
parent dd30786fef
commit 19c0bed584
6 changed files with 101 additions and 29 deletions
@@ -21,6 +21,7 @@ from vllm.model_executor.kernels.linear.nvfp4.base import (
NvFp4LinearLayerConfig,
)
from vllm.model_executor.kernels.linear.nvfp4.flashinfer import (
FlashInferCuteDslNvFp4LinearKernel,
FlashInferCutlassNvFp4LinearKernel,
FlashInferTrtllmNvFp4LinearKernel,
)
@@ -59,6 +60,7 @@ from vllm.platforms import current_platform
SUPPORTING = {
CutlassFP8ScaledMMLinearKernel,
FlashInferFP8ScaledMMLinearKernel,
FlashInferCuteDslNvFp4LinearKernel,
FlashInferCutlassNvFp4LinearKernel,
PerTensorTorchFP8ScaledMMLinearKernel,
AiterHipbMMPerTokenFp8ScaledMMLinearKernel,