How do you decide what a coarse gate should block vs ship on? #846
Unanswered
yashpatel1230
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Saw the release eval-smoke workflow — you label it a coarse gate on a small real-model sample. That tension is exactly what I am digging into: real-model eval is slow and expensive so the gate stays coarse, but a coarse gate misses regressions. For an enterprise-ready agent platform like Pinchy, how do you decide today what is coarse-enough to ship on versus worth blocking a release for?
All reactions