docs: update whitepaper for the programmable/deep/CI-enforced trainer - #1876
Merged
Conversation
The whitepaper undersold the product (2-layer-only, multi-output/depth pending). Section 4 now reflects the current reality: (b) one-shared-multiplier microsequencer; (c) fully programmable trainer -- any feed-forward topology, free input/output/hidden width AND arbitrary depth (2/3/4-layer), trainable biases, learning real tasks (~97% 2-layer / 98% 3-layer nonlinear, 93% multi-class); (d) correctness as a CI-enforced invariant (bit-exact spec->RTL over a full training run + synthesizability + one-shared-multiplier datapath invariant, every PR). Section 5 honesty updated (cell counts version-specific = trend; the multi-output bug the gate caught+fixed; newest stack pending one JTAG re-connect, earlier 2-layer trainers already validated). Docs only. 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The whitepaper undersold the product (described a 2-layer-only trainer with multi-output/depth pending). Section 4 now reflects the current reality: (c) fully programmable trainer — any feed-forward topology, free input/output/hidden width AND arbitrary depth (2/3/4-layer), trainable biases, learning real tasks (~97% 2-layer / 98% 3-layer nonlinear, 93% multi-class); (d) correctness as a CI-enforced invariant (bit-exact spec→RTL over a full training run + synthesizability + one-shared-multiplier datapath invariant, on every PR). Section 5 honesty updated (version-specific cell counts = trend; the multi-output bug the gate caught+fixed; newest stack pending one JTAG re-connect). Docs only. Refs #1764