Speculative prompt caching
Explores speculative prompt caching patterns and compares performance against standard caching.
- Upstream notebook:
misc/speculative_prompt_caching.ipynb
What to focus on
- When speculation helps (high reuse, predictable branches)
- Failure modes (low hit-rate, wasted tokens)
- How to evaluate the tradeoffs with real traffic
Run locally
make test-notebooks NOTEBOOK=misc/speculative_prompt_caching.ipynb