Skip to content

⚡️ Speed up method Qwen3OmniMoeCausalConvNet._get_extra_padding_for_conv1d by 36% #892

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
codeflash-ai wants to merge 1 commit into
base: main
Choose a base branch
from
Open

⚡️ Speed up method Qwen3OmniMoeCausalConvNet._get_extra_padding_for_conv1d by 36% #892

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 9 additions & 2 deletions src/transformers/models/qwen3_omni_moe/modular_qwen3_omni_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1987,8 +1987,15 @@ def __init__(

def _get_extra_padding_for_conv1d(self, hidden_state: torch.Tensor) -> int:
length = hidden_state.shape[-1]
n_frames = (length - self.kernel_size + self.padding) / self.stride + 1
ideal_length = (math.ceil(n_frames) - 1) * self.stride + (self.kernel_size - self.padding)
# Avoid repeated calculations and float divisions by unrolling math.
k = self.kernel_size
p = self.padding
s = self.stride
# Compute numerator as integer for ceil division later
numer = length - k + p
# Use integer math for ceil division to avoid float math.ceil
n_frames = (numer + s - 1) // s + 1
ideal_length = (n_frames - 1) * s + (k - p)
return ideal_length - length

def forward(self, hidden_state):
Expand Down