Skip to content

perf: Submit Tinker training requests in the same clock cycle #781

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
vivekkalyan merged 1 commit into from
Jul 28, 2026
+8 −9
Merged

perf: Submit Tinker training requests in the same clock cycle #781

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
17 changes: 8 additions & 9 deletions src/art/tinker_native/backend.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -424,16 +424,15 @@ def remove_mask(datum: tinker.Datum) -> tinker.Datum:
model_input=datum.model_input, loss_fn_inputs=loss_fn_inputs
)

forward_output = await self._tinker_train_call(
"forward_backward",
state.training_client.forward_backward(
[remove_mask(datum) for datum in datums],
loss_fn=loss_fn,
loss_fn_config=loss_fn_config,
),
forward_future = state.training_client.forward_backward(
[remove_mask(datum) for datum in datums],
loss_fn=loss_fn,
loss_fn_config=loss_fn_config,
)
optim_output = await self._tinker_train_call(
"optim_step", state.training_client.optim_step(adam_params)
optim_future = state.training_client.optim_step(adam_params)
forward_output, optim_output = await asyncio.gather(
self._tinker_train_call("forward_backward", forward_future),
self._tinker_train_call("optim_step", optim_future),
)

if forward_output.metrics:
Expand Down
Loading