What shipped in the workspace, newest first. Also in the Atom feed.
Reward functions can now call sandboxed verifiers (SQL execution, unit tests, schema validation) with per-call timeouts. The run view plots the share of zero-variance groups next to mean reward, and dynamic sampling can skip them.
Eval comparisons now report paired differences with 95% intervals per slice, and release gates can be written against the interval's lower bound.
Calibrating a judge against human labels now reports sensitivity, specificity and κ with bootstrap intervals, and pass rates can be shown bias-corrected.
B200 quotes appear alongside H100 and H200, and the broker can rank by estimated cost per run instead of price per hour.
Runs launched from a Databricks workspace register models in Unity Catalog with dataset versions and eval results attached.
Tell us what the model should get better at. A post-training engineer replies within one business day with a plan, a compute estimate and an eval design.