Grokking LLM Evals
Vote

0% completed

​

Evals as Release Gates

The following function decides whether a change to the job ticket assistant may be merged.

def release_gate(passed, total, floor=0.95): low, high = wilson(passed, total) # 95 percent interval for the pass rate if low >= floor: return "pass" if high < floor: return "fail" return "too close to call"

Most gates compare a pass rate with a number. This one compares a range of plausible pass rates with the number. It also has a third answer, for when the range sits on both sides of the number.

.....

.....

.....

Like the course? Get enrolled and start learning!

Reading Progress

0%


Vote for new content