Scalar reverse-mode autograd engine + SGD#1
Merged
Conversation
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 little scalar autograd engine, basically a Value type that records each op as it happens and then replays the graph backward to fill in gradients via the chain rule. built the usual ops (+, *, **, div) plus relu/tanh/exp/log/sigmoid, and an SGD optimizer with optional momentum plus a minimize loop on top. this is the gradient descent + backprop foundation everything else in the repo will lean on. tests check the grads against numeric finite differences, verify a softmax cross-entropy grad against numpy, and actually train a tiny linear regression through the engine to make sure the whole loop works. also covers the annoying edge cases like a node reused twice needing its grads accumulated. numpy only, no ml libs.