Skip to content

Fixed hyperparameter scope in Muon Torch #32

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
aahladc merged 1 commit into from
May 21, 2026
Merged

Fixed hyperparameter scope in Muon Torch #32

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
6 changes: 5 additions & 1 deletion submissions/self_tuning/muon_torch/submission.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,7 @@
"warmup_factor": 0.05,
"step_reduce": 1.0
}
hyperparameters = SimpleNamespace(**HPARAMS)
HPARAMS = SimpleNamespace(**HPARAMS)


def _pytorch_cosine_warmup(step_hint: int, hyperparameters, optimizer):
Expand All @@ -65,6 +65,8 @@ def init_optimizer_state(
del model_state
del rng

hyperparameters = HPARAMS

muon_params, adam_params = split_params_muon_adam(model_params)

optimizer_state = {
Expand Down Expand Up @@ -119,6 +121,8 @@ def update_params(
del train_state
del eval_results

hyperparameters = HPARAMS

reduced_steps = int(workload.step_hint * getattr(hyperparameters, "step_reduce", 1.0))
if global_step >= reduced_steps:
raise spec.TrainingCompleteError(
Expand Down
Loading