Skip to content

fix: retain explicitly scored execution timeouts in training #3930

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

Draft
hallerite wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

fix: retain explicitly scored execution timeouts in training #3930

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
12 changes: 9 additions & 3 deletions src/prime_rl/orchestrator/dispatcher.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -670,9 +670,15 @@ async def handle_completed_request(self, task: asyncio.Task) -> None:
episode.errors.append(vf.Error(type="EmptyEpisode", message="Episode returned with no traces"))

for trace in episode.traces:
if trace.is_timeout and not trace.has_error and meta.kind == "train":
# Training keeps a timed-out rollout out of the batch: an error,
# like any other failure. Eval keeps it as a scored outcome.
termination = getattr(trace, "termination", None)
scored_termination = (
getattr(termination, "valid_sample", False)
and getattr(termination, "reward", None) is not None
and trace.reward == termination.reward
)
if trace.is_timeout and not trace.has_error and meta.kind == "train" and not scored_termination:
# An explicit execution policy may retain a timeout as a scored
# training sample. Unclassified timeouts remain errors.
trace.errors.append(vf.Error(type="Timeout", message=f"Trace stopped by {trace.stop_condition}"))
trace.ok = False
episode.ok = False
Expand Down
Loading