Docs
From a reward spec to a gated update.
Three commands cover the first run. Everything after that is review, yield, and whether the policy is allowed to leave the lab.
01
Install
> npm install -g smalldataregimeThe CLI talks to a cluster, a lab machine, or a local checkpoint directory.
02
Point at a spec
> smalldataregime spec check reward.mdA spec needs a desired behavior, at least one negative, and a held-out eval command.
03
Train inside a budget
> smalldataregime train --budget 200 --reward prefs.jsonlThe run stops at the preference budget or when the KL cap trips, then waits on the eval gate.