graphrag
graphrag copied to clipboard
Published
1 month ago
•
microsoft
microsoft
Add LLM Token and Cost Estimation Pre-check Before Indexing
Description
This PR introduces a new feature that estimates the token usage and LLM cost before executing the indexing pipeline. This helps users anticipate costs and make informed decisions before proceeding.
Related Issues
https://github.com/microsoft/graphrag/issues/385 , https://github.com/microsoft/graphrag/pull/487
Proposed Changes
New CLI params:
- --estimate-cost : [True || False] (by default False)
- --average-output-tokens-per-chunk : (by default 500}
graphrag index \
--root ./ragtest \
--estimate-cost \
--average-output-tokens-per-chunk 500
- Util file: estimate_cost.py added for token counting and pricing logic.
- Pricing dynamically fetched from a public JSON source.
- CLI flag --estimate-cost added for optional pre-check.
- Console summary with embedding/chat model breakdown and token counts.
- User confirmation prompt before continuing indexing. = Conservative cost estimation based on --average-output-tokens-per-chunk value.
Checklist
- [x] I have tested these changes locally.
- [x] I have reviewed the code changes.
- [x] I have updated the documentation (if necessary).
- [ ] I have added appropriate unit tests (if applicable).
