evaluation criteria
Precise, measurable definitions of what success looks like for an AI output; specific requirements that the final product must meet to be considered acceptable.
As defined in Stop Prompting Claude. Use Karpathy's Method Instead.
Where it comes up · 1
Most important claims · 3 of 3Search all claims for “evaluation criteria” →
- The first of three verification focus areas is to set the evaluation criteria upfront and with precision before Claude touches anything; vague criteria like 'make this report look good' should be replaced with precise criteria like 'the report must have three sections, each ending with a recommendation.'Stop Prompting Claude. Use Karpathy's Method Instead.
- Precise evaluation criteria are similar to Layer 1 (specs) in that the more precise you are upfront, the less room Claude will have to make mistakes.Stop Prompting Claude. Use Karpathy's Method Instead.
- If Claude has a feedback loop (combining all three verification focus areas: precise evaluation criteria, second-model criticism, external signal), it will 2-3x the quality of the final result.Stop Prompting Claude. Use Karpathy's Method Instead.