Which prompt technique actually cuts documentation hallucinations?
Nine prompt techniques, each a single controlled change off one baseline, measured on Javadoc, inline comments, OpenAPI, README and four diagram formats.
A running log of interesting AI evals and measurements — methods, numbers, and what they tell us about how models actually behave.
Nine prompt techniques, each a single controlled change off one baseline, measured on Javadoc, inline comments, OpenAPI, README and four diagram formats.
No matching evals.