feat: GF-T axpy primitive (perceptron/SGD vector update) - #1824
Merged
Conversation
gft_axpy.t27: one-class weight-vector update w' = w +/- eta*x over 2 features (sign=1 boost w+eta*x, sign=0 suppress w-eta*x); returns (w0'<<32)|w1'. Reuses smul/sadd/neg. Building block for multi-class perceptron training. Honest envelope note: a full on-chip 3-class perceptron (GftClassify3 + 2x GftAxpy + register routing) placed on the AX7203 (fasm 22.9M) but exceeds the openXC7 CORRECTNESS envelope -- a faithful GF-T Python sim converges to 100% and iverilog learns, but the silicon bitstream miscomputes. The axpy spec itself is verified (boost/supp tests PASS). On-chip multi-class training needs the model split into smaller separately-verified passes. 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
New spec gft_axpy.t27 — one-class weight-vector update w' = w +/- eta*x over 2 features (sign=1 boost, sign=0 suppress); returns (w0'<<32)|w1'. In-spec tests (boost/supp) PASS. Building block for multi-class perceptron training. Honest note: a full on-chip 3-class perceptron placed on AX7203 (fasm 22.9M) but exceeds the openXC7 CORRECTNESS envelope (GF-T sim converges 100%, iverilog learns, silicon miscomputes) — documented for future split-pass approach. docs/NOW.md updated. Refs #1764