docs(track-w): EWC hardening + track_w status - #47
Merged
Conversation
CLAUDE.md: track_w/ line updated — 6 substrates implemented + continual/ sub-package (ewc.py + rehearsal.py, PR #46). README.md: new Track-W hardening section — rehearsal 0.034 vs EWC@λ=1000 0.532 vs none 0.767 on HardSplitTask, honest framing, trace to w4-ewc-comparison.json, API summary.
There was a problem hiding this comment.
Pull request overview
Documentation-only update reflecting the EWC continual-learning hardening work merged in PR #46. Updates the track_w/ status note in CLAUDE.md and adds a new "Track-W hardening" section to README.md summarising the honest rehearsal-vs-EWC comparison on HardSplitTask, with numbers traced to docs/superpowers/research/2026-05-30-w4-ewc-comparison.json.
Changes:
- Replace the "future plan" note for
track_w/inCLAUDE.mdwith the current six-substrate inventory and the newtrack_w/continual/sub-package. - Add a README section comparing rehearsal (0.034), EWC@λ=1000 (0.532) and none (0.767) on
HardSplitTask, framed as a diagnostic rather than a new gate. - Summarise the public APIs in
track_w/continual/(estimate_fisher,penalty,RehearsalBuffer) and referencetasks/hard_split.py.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
| CLAUDE.md | Updates the track_w/ line from "future plan" to the implemented substrate list and the new track_w/continual/ sub-package. |
| README.md | Adds a "Track-W hardening — EWC vs rehearsal" section with the multi-seed forgetting table, framing, and pointers to the new continual modules. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+241
to
+243
| Continual-learning hardening of gate W on a deliberately difficult task | ||
| (`HardSplitTask` — shared 12-class head, Tasks 0 and 1 on non-overlapping | ||
| label splits). Three methods compared under matched settings: |
|
|
||
| | Method | Forgetting ↓ (lower = better) | Notes | | ||
| |---|---|---| | ||
| | **Rehearsal** (20 % replay) | **0.034** | Best by a wide margin | |
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.
Summary
Documentation-only update reflecting PR #46 (EWC continual hardening).
track_w/ — future planline — 6substrates are implemented and
track_w/continual/(ewc.py,rehearsal.py) is shipped.rehearsal vs EWC comparison on
HardSplitTask(rehearsal 0.034 ≫EWC@λ=1000 0.532 ≫ none 0.767), framing as a diagnostic (not a new
gate), and API summary for
estimate_fisher,penalty, andRehearsalBuffer. Numbers trace todocs/superpowers/research/2026-05-30-w4-ewc-comparison.json.No code changes.
🤖 Generated with Claude Code