feat: trainable biases + real-task generalization for the GF-T trainer - #1867
Merged
Conversation
gen() now emits TRAINABLE hidden biases b_j and output biases bo_o (a proper general 2-layer net, not the XOR-specific fixed-bias hack): forward z_j=W_j.x+b_j, y_o=v_o.relu(z)+bo_o; backprop adds bias grads/updates. Self-tests PASS: (a) generated XOR microcode trains 4/4; (b) a (2,4,1) net trains a NOISY NONLINEAR 2D task (XOR-region, real-valued points) and generalizes to held-out 58/60 (~97%) -- real learning + generalization, not toy XOR. Still one shared smul+sadd datapath, ~constant area. A proper programmable ternary NN trainer that learns real nonlinear tasks with generalization. Refs #1764 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
gen() now emits TRAINABLE hidden + output biases (a proper general 2-layer net). Self-tests PASS: generated XOR microcode trains 4/4; a (2,4,1) net trains a NOISY NONLINEAR 2D task and generalizes to held-out 58/60 (~97%) — real learning, not toy XOR. Same shared smul+sadd datapath, ~constant area. A proper programmable ternary NN trainer that learns real nonlinear tasks with generalization. docs/NOW.md updated. Refs #1764