docs: paired-bootstrap CIs on the headline deltas - #68
Merged
Conversation
Same 1,200 paragraphs, zero unpaired, 10k resamples: r7 over r6 (-0.367pp, CI [-0.467,-0.278], p<1e-4); r7 over GLM-5.2 projected (+0.539pp, CI [+0.202,+0.939], p=0.0003); GLM-5.3-Flash over GLM-5.2 (+7.966pp, CI [+7.520,+8.368]) — the frontier regression is overwhelming, not convention noise. Official Misraj aggregates stay canonical; the simplified per-paragraph scorer is disclosed in the section. Policy: deltas carry CIs from here on.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
B1 of the improvement plan (TODO 08): paired bootstrap over the same 1,200 paragraphs for every headline delta — r7 over r6 (-0.367pp, CI [-0.467,-0.278], p<1e-4), our teacher over GLM-5.2 projected (+0.539pp, CI [+0.202,+0.939], p=0.0003), and the GLM-5.3-Flash frontier regression (+7.966pp, CI [+7.520,+8.368], p<1e-4 — overwhelming, not convention noise). The per-paragraph scorer is disclosed as simplified; official Misraj aggregates remain canonical. Policy from here: every quoted delta carries its CI or a note that the run predates the policy.