Appearance
1 entry · Building products and tooling on top of language models.
Reusing a stable prompt prefix across requests to cut cost and latency.