Lightning-AI / Lightning-AI/litgpt
Implement prompt caching to speed up inference
Open
Nobody has claimed this yet.
enhancement
- Dominant language
- Python
- Stars
- 13.7k
- Forks
- 1.5k
- Avg merge
- 15h 37m
- Merged PRs (30d)
- 1
Description
In addition to kv-caching, it makes sense to also add prompt caching. I.e., instead re-computing the system-prompt, we can cache these prefilled prompts to avoid recalculations
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No files, tests, or specific entry points are named. Start by tracing the inference path and existing KV-caching implementation, then clarify the intended prompt-cache scope and invalidation behavior. Done should mean system prompts are not unnecessarily recomputed and inference speed is demonstrably improved.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- ai, performance
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100