Skip to content

Make gradient checkpointing configurable in the assimilate global #1155

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
wants to merge 2 commits into
base: develop
Choose a base branch
from
+7 −2
Open

Make gradient checkpointing configurable in the assimilate global #1155

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
11161e5
add assimilate_global_gradient_checkpoint_mode to config
Oct 25, 2025
cf5b13f
ruff the code
Oct 27, 2025
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
1 change: 1 addition & 0 deletions config/default_config.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,7 @@ ae_global_with_qk_lnorm: True
ae_global_att_dense_rate: 1.0
ae_global_block_factor: 64
ae_global_mlp_hidden_factor: 2
assimilate_global_gradient_checkpoint_mode: False

decoder_type: PerceiverIOCoordConditioning # CrossAttentionAdaNormConditioning
pred_adapter_kv: False
Expand Down
8 changes: 6 additions & 2 deletions src/weathergen/model/model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -787,8 +787,12 @@ def assimilate_global(self, model_params: ModelParams, tokens: torch.Tensor) ->
"""

# global assimilation engine and adapter
for block in self.ae_global_blocks:
tokens = checkpoint(block, tokens, use_reentrant=False)
if self.cf.assimilate_global_gradient_checkpoint_mode:
for block in self.ae_global_blocks:
tokens = checkpoint(block, tokens, use_reentrant=False)
else:
for block in self.ae_global_blocks:
tokens = block(tokens)

return tokens

Expand Down
Loading