[MM][Perf][CG] Support ViT full CUDA graph for Qwen3.5 (#42151)

Signed-off-by: shen-shanshan <467638484@qq.com>
This commit is contained in:
Shanshan Shen
2026-05-13 16:00:32 +08:00
committed by GitHub
parent 85b2fecab7
commit 92def124bc
4 changed files with 112 additions and 5 deletions
+2 -1
View File
@@ -85,8 +85,9 @@ Models opt-in to encoder CUDA Graphs by implementing the [SupportsEncoderCudaGra
| Architecture | Models | CG for Image | CG for Video |
| ------------ | ------ | ------------ | ------------ |
| `Qwen3VLForConditionalGeneration` | `Qwen3-VL` | ✅︎ | ✅︎ |
| `Qwen2_5_VLForConditionalGeneration` | `Qwen2.5-VL` | ✅︎ | ✅︎ |
| `Qwen3VLForConditionalGeneration` | `Qwen3-VL` | ✅︎ | ✅︎ |
| `Qwen3_5ForConditionalGeneration` | `Qwen3.5` | ✅︎ | ✅︎ |
!!! note
Encoder CUDA Graphs have currently been tested with `--mm-encoder-attn-backend=FLASH_ATTN` and `--mm-encoder-attn-backend=FLASHINFER` on Blackwell GPUs.