← Curriculum track ← Learn hub
Quanta GenAI Curriculum · Generative AI · Advanced

GenAI Advanced — 053: debate multimodal quotas on `Evaluation harness depth` — memo `618014 [53]`

Lesson 053: Evaluation harness depth

Focus

Bias toward observable metrics, not model marketing. Token Evaluation harness depth:53 keeps neighbouring lessons differentiable.

Key ideas

Deep dive notebook

Synthetic drill artefacts

Token CFO scratchpad

- prompt_budget_tokens: 1645
- completion_budget_tokens: 590
- cache_signature: `6ac7`

Hypothesis: halving completions moves P95 `8`%; record actuals.

Practice

Practice Simulate degraded retrieval once; screenshot graceful degradation copy. — 53 Bump literals mindset by 33.