test: gate a 3-class argmax classifier learns -- closes the scaling axes (Refs #1764) - #1916
Merged
Conversation
…xes (Refs #1764) Adds a 3-class argmax classifier (2,8,3) learning angular sectors (one-hot targets, argmax over 3 outputs), held-out 55/60 (92%), deterministic, ~4.5s. The CI learning gate now proves generalization across all four scaling axes: inputs (3-input majority), depth (deep nets), width, and outputs (2- and now 3-class argmax) -- each held-out >=90% per PR. Cements that the spec-verified method learns and generalizes across dimensions; the only limit was the open silicon flow's placement marginality. Refs #1764 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Contributor
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds a 3-class argmax classifier
(2,8,3)learning angular sectors (one-hot targets, argmax over 3 outputs) to the CI learning gate — held-out 55/60 (92%), deterministic, ~4.5s.The gate now proves generalization across all four scaling axes:
(3,5,1);[2,4,3,1]/[2,5,3,1];(2,4,2)and now 3-class argmax(2,8,3).Each held-out ≥ 90%, enforced per pull request. This cements the thesis: the spec-verified method learns and generalizes across dimensions in the model — the only limit was the open silicon flow's placement marginality (which works today via met-timing + seed-search; commercial P&R adds determinism and scale). Refs #1764