Skip to content
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

Skip non-selected experts for mixtral and qwen2_moe #32429

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Skip non-selected experts for mixtral and qwen2_moe #32429

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
00e19b4
Skip non-selected experts for mixtral and qwen2_moe
Coco58323 Aug 5, 2024
774f2c8
Fix: tensor tolist()
Coco58323 Aug 5, 2024
7c092a6
WIP: tokenization test
Coco58323 Aug 5, 2024
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
3 changes: 2 additions & 1 deletion src/transformers/models/mixtral/modeling_mixtral.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -728,7 +728,8 @@ def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
expert_mask = torch.nn.functional.one_hot(selected_experts, num_classes=self.num_experts).permute(2, 1, 0)

# Loop over all available experts in the model and perform the computation on each expert
for expert_idx in range(self.num_experts):
expert_hitted = (expert_mask.sum(dim=(-1, -2)) > 0).nonzero(as_tuple=True)[0].tolist()
for expert_idx in expert_hitted:
expert_layer = self.experts[expert_idx]
idx, top_x = torch.where(expert_mask[expert_idx])

Expand Down
3 changes: 2 additions & 1 deletion src/transformers/models/qwen2_moe/modeling_qwen2_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -719,7 +719,8 @@ def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
expert_mask = torch.nn.functional.one_hot(selected_experts, num_classes=self.num_experts).permute(2, 1, 0)

# Loop over all available experts in the model and perform the computation on each expert
for expert_idx in range(self.num_experts):
expert_hitted = (expert_mask.sum(dim=(-1, -2)) > 0).nonzero(as_tuple=True)[0].tolist()
for expert_idx in expert_hitted:
expert_layer = self.experts[expert_idx]
idx, top_x = torch.where(expert_mask[expert_idx])

Expand Down