iHore/vLLM
8
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Fix Qwen3-VL M-RoPE on the Transformers modeling backend (grids + compile) (#49292)

Browse Source 
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Aritra Roy Gosthipaty
2026-07-21 18:28:39 +00:00
committed by GitHub
co-authored by Harry Mellor
parent b2b8f679d0
commit 33178f9006
1 changed files with 3 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
vllm/model_executor/models/transformers/multimodal.py
+3 -7
View File
@@ -337,7 +337,7 @@ class MultiModalMixin(SupportsMultiModal, SupportsMRoPE):
# Positions shape handling for MRoPE models
if self.model_config.uses_mrope:
# [3, seq_len] -> [3, 1, seq_len]
positions = positions[:, None]
positions = positions[:, None].contiguous()
model_output = super().forward(
input_ids, positions, intermediate_tensors, inputs_embeds
)
@@ -481,12 +481,8 @@ class MultiModalMixin(SupportsMultiModal, SupportsMRoPE):
image_grid_thw = kwargs.get("image_grid_thw", [])
video_grid_thw = kwargs.get("video_grid_thw", [])
image_grid_thw = (torch.stack if image_grid_thw else torch.tensor)(
image_grid_thw
)
video_grid_thw = (torch.stack if video_grid_thw else torch.tensor)(
video_grid_thw
)
image_grid_thw = torch.stack(image_grid_thw) if image_grid_thw else None
video_grid_thw = torch.stack(video_grid_thw) if video_grid_thw else None
# `get_rope_index` doesn't always accept arbitrary `kwargs`
kwargs = {}