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

[Bugfix] Fix more multimodal tests for transformers V5 (#34334)

Browse Source 
Signed-off-by: raushan <raushan@huggingface.co>
This commit is contained in:
Raushan Turganbay
2026-02-11 22:02:05 +01:00
committed by GitHub
parent 5458eb835d
commit 527ca32197
5 changed files with 18 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/model_executor/models/glmasr_utils.py
View File

@@ -18,8 +18,8 @@ def _calculate_conv_output_length(
input_length: torch.Tensor, padding: int, kernel_size: int, stride: int
) -> torch.Tensor:
"""Calculate Conv1d output length using standard formula."""
# Standard formula: floor((input + 2*padding - kernel_size) / stride) + 1
return (input_length + 2 * padding - kernel_size) // stride + 1
# in sync with `hf_processor._get_audio_token_length`
return (input_length + 2 * padding - (kernel_size - 1) - 1) // stride + 1
def _as_list_chunk_counts(