Repository navigation
Make Attention's matching, normalization and value mixture visibly causal - #7
Merged
Merged
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
hash-panda
changed the title
hash-panda
marked this pull request as ready for review
October 29, 2026 19:11
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Attention previously switched between textual end states, so learners had to reconstruct the computation from narration. This revision keeps position rows and exact values continuous through Q/K component multiplication, causal masking, shared-denominator formation, normalized fractions, weighted V and cumulative output.
The first 18% of each later chapter carries the previous output into the next input. Result labels wait for their operation; dashed V and colored pV share a 0–20 scale. Front/side/bottom surfaces, component wells and connected output ports add functional structure without skewing quantitative lengths. Mobile SVG text is at least 15.178 CSS px in the inspected frame matrix.
Validation: npm run check passes, including 17,298 Attention poses, independent original goldens, handoff/path/end/reverse checks and retained scene/mechanical checks. Actual cloud Chrome review includes 1×, native 4×, 390×844, 31 frame bounds/type observations, 12 reverse geometry comparisons, counterfactuals, playback lifecycle, keyboard, manual reduced motion, controlled hidden event and prediction→verify flow. Source-matched screenshots and limitations are committed in docs/ai-learning/reviews/2026-10-29-attention-rework.md.
Developer teaching/craft review passes the concrete reopened Attention findings. No learner-effect study, physical phone, native background scheduling or new Skill host compatibility is claimed. Tokenization/Embedding and Skill distribution alignment remain open; no new topic or mechanical source changed.
Release gate: inspect this exact checked branch preview, then merge and verify matching production READY/SHA and custom-domain flow. Original publication receipts are preserved.