You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
thirdshift experiment: the Command, the Bar and running Trials #1003
The Command: it writes a Command record naming both Arms and the Bar, a Command log and Activity log lines. They live in thirdshift's own repository's logs folder, in an experiment folder. It shows in the Inbox and can be Stopped.
The Bar: the standing default is a 10-point margin, a 95% credible level, the prior, the seed and draw count, the Trials per Eval task, the per-Trial timeout and the Void Trial rules. It's recorded before the first Trial, and refused if it changes afterwards.
Resuming: running the same Experiment again resumes it, skipping finished Trials and re-running the ones cut short. A finished Experiment reports its results and doesn't run again.
An integration test with fake Harnesses and a fixture suite checks the Command record (kind, both Arms, the Bar and the stages), the Command log, the Activity log lines and one result per Trial.
At most two Trials run at once, and the Arms are interleaved.
A changed Bar is refused after the first Trial.
An interrupted Experiment resumes when run again, without re-running finished Trials.
Stop ends it, from the CLI as for any Command.
thirdshift experiments --json lists and shows Experiments, and the guide describes the command.
Activity
JacobStephens2 commentedon Oct 23, 2026
Closed by #1128, merged into issue-977 by a thirdshift Merge run.