1. Pass recurse=True to _convert_all_gradient_functions inside builder() so nested custom gradients inside gradient functions are registered lazily on demand. - #283
Open
copybara-service[bot] wants to merge 1 commit into
Conversation
…uilder()` so nested custom gradients inside gradient functions are registered lazily on demand. 2. Apply `jax.lax.stop_gradient` to `_eval_subgraph` inputs in `_Subgraph.__call__.fn_fwd` so higher-order autodiff does not attempt to differentiate through `XlaCallModule` (`mhlo_apply`), while keeping un-stopped `args` in `lookup` so captured residuals retain their tracers for `fn_bwd`. PiperOrigin-RevId: 982462168
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
recurse=Trueto_convert_all_gradient_functionsinsidebuilder()so nested custom gradients inside gradient functions are registered lazily on demand.jax.lax.stop_gradientto_eval_subgraphinputs in_Subgraph.__call__.fn_fwdso higher-order autodiff does not attempt to differentiate throughXlaCallModule(mhlo_apply), while keeping un-stoppedargsinlookupso captured residuals retain their tracers forfn_bwd.