Skip to content

Update tile size #747

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

Merged
copybara-service merged 1 commit into from
Jul 3, 2024
Merged

Update tile size #747

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
5 changes: 1 addition & 4 deletions MaxText/layers/linears.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -354,10 +354,7 @@ def unpermute(self, intermediate, sorted_selected_experts, weights):
return output.reshape(int(self.config.per_device_batch_size), -1, self.config.emb_dim).astype(self.dtype)

def megablox(self, inputs, gate_logits, config, w0_kernel, w1_kernel, wo_kernel):
# TODO(ranran): need to changes in JAX repo to enable optimized tile_size
# instead of the static default tile_size (512, 512, 512)
tile_size = (512, 512, 512)

tile_size = (512, 1024, 1024)
def gmm(inputs, kernel, group_sizes):
hs_shape = inputs.shape
# pad length is the 1st dimension of tiling size in gmm call
Expand Down