Replies: 2 comments
-
|
haha... so sad story.. |
Beta Was this translation helpful? Give feedback.
-
|
Apologies for the long silence — a measured 3.1x regression deserved a fast answer, and this thread shaped real changes that just landed. What can make CRG consume MORE tokens, verified:
If you're willing to share the repo's file count, CRG version, and which agent/mode you used, I'd genuinely like to reproduce the 831k vs 269k result — that's exactly what our eval suite should capture, and we just added a grep-baseline benchmark to make this comparison honest. Also worth confirming your build isn't indexing vendored code (node_modules/dist are excluded by default; |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hi, I want to use this tool for across my projects.
To evaluate, I cloned the same repo in 2 different folders and installed code-review-graph in one repo and did not in the other.
I opened both the repos in 2 different cursor windows.
Send the EXACT same request in the one with crg and without crg. The one with crg is consuming 831k tokens and without it consumned 269k tokens. Had multiple similar observations with other repos as well.
I have installed and followed the same steps as mentioned in the README.
What might be going wrong?
Beta Was this translation helpful? Give feedback.
All reactions