Unexpected behaviour of helper function `_get_feat_extract_output_lengths` in qwen3_omni_moe

[CYQFWang]

System Info

• transformers version: 5.0.0
• Platform: Linux-6.6.113+-x86_64-with-glibc2.35
• Python version: 3.12.13
• Huggingface_hub version: 1.7.1
• Safetensors version: 0.7.0
• Accelerate version: 1.13.0
• Accelerate config: not found
• DeepSpeed version: not installed
• PyTorch version (accelerator?): 2.10.0+cpu (NA)
• Using distributed or parallel set-up in script?: No

Who can help?

No response

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• My own task or dataset (give details below)

Reproduction

transformers/src/transformers/models/qwen3_omni_moe/modular_qwen3_omni_moe.py

Line 117 in 9a9997f

def _get_feat_extract_output_lengths(input_lengths):

The implementation of above function computing the output length of the audio encoder does not align with the official formula of pytorch Conv2d.
The audio encoder convolution is defined in

transformers/src/transformers/models/qwen3_omni_moe/modular_qwen3_omni_moe.py

Line 871 in 9a9997f

self.conv2d1 = nn.Conv2d(1, config.downsample_hidden_size, 3, 2, padding=1)

Expected behavior

Current implementation is

def _get_feat_extract_output_lengths(input_lengths):
    """
    Computes the output length of the convolutional layers and the output length of the audio encoder
    """

    input_lengths_leave = input_lengths % 100
    feat_lengths = (input_lengths_leave - 1) // 2 + 1
    output_lengths = ((feat_lengths - 1) // 2 + 1 - 1) // 2 + 1 + (input_lengths // 100) * 13
    return output_lengths

and the expected implementation is

def _get_feat_extract_output_lengths(input_lengths):
    """
    Computes the output length of the convolutional layers and the output length of the audio encoder
    """

    feat_lengths = (input_lengths- 1) // 2 + 1
    output_lengths = ((feat_lengths - 1) // 2 + 1 - 1) // 2 + 1 
    return output_lengths