Skip to content

Use save_weights_and_get_sampling_client_async in rl/train.py #84

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
merged 2 commits into from
Nov 11, 2025
Merged

Use save_weights_and_get_sampling_client_async in rl/train.py #84

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
707210d
Use save_weights_and_get_sampling_client_async in rl/train.py
andriigrynenko Nov 11, 2025
8d027c6
Update tinker dependency version in pyproject.toml
andriigrynenko Nov 11, 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
2 changes: 1 addition & 1 deletion pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@ dependencies = [
"numpy",
"rich",
"termcolor",
"tinker",
"tinker>=0.3.0",
"torch",
"transformers",
"blobfile",
Expand Down
19 changes: 11 additions & 8 deletions tinker_cookbook/rl/train.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -686,14 +686,17 @@ async def save_checkpoint_and_get_sampling_client(
) -> tuple[tinker.SamplingClient, dict[str, Any]]:
metrics = {}
with timed("save_checkpoint", metrics):
path_dict = await checkpoint_utils.save_checkpoint_async(
training_client=training_client,
name=f"{i_batch:06d}",
log_path=log_path,
loop_state={"batch": i_batch},
kind="both" if (i_batch > start_batch and i_batch % save_every == 0) else "sampler",
)
return training_client.create_sampling_client(path_dict["sampler_path"]), metrics
if i_batch > start_batch and i_batch % save_every == 0:
path_dict = await checkpoint_utils.save_checkpoint_async(
training_client=training_client,
name=f"{i_batch:06d}",
log_path=log_path,
loop_state={"batch": i_batch},
kind="both",
)
return training_client.create_sampling_client(path_dict["sampler_path"]), metrics
else:
return await training_client.save_weights_and_get_sampling_client_async(), metrics


@scope
Expand Down