[tmva][sofie] Support Clad forward mode in generated inference code - #23077
Merged
guitargeek merged 1 commit intoAug 14, 2026
Merged
Conversation
Emit "_pushforward" custom derivatives for the standalone helper functions (Gemm_Call, Copy, Fill, Relu) next to the existing pullbacks, so that clad::differentiate() works on the generated code — and with it Hessians, which Clad computes by reverse-differentiating the forward derivative code. Also generate a Session::SetWeightsToZero() method for building the zero-weight tangent Sessions that forward-mode differentiation needs (the weights are constants, not inputs). Add a TestCladAutodiff test on a Gemm+Sigmoid model that validates forward-mode directional derivatives against the reverse-mode gradient, and a full Hessian, computed via exact Hessian-vector products (reverse-over-forward), against finite differences of the exact gradient. The Clad limitations and workarounds all this rests on are documented in the emitted code and in the test. 🤖 Done with the help of AI
Test Results 23 files 23 suites 3d 15h 12m 33s ⏱️ For more details on these failures, see this check. Results for commit 90026a6. |
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Emit "_pushforward" custom derivatives for the standalone helper functions (Gemm_Call, Copy, Fill, Relu) next to the existing pullbacks, so that clad::differentiate() works on the generated code — and with it Hessians, which Clad computes by reverse differentiating the forward derivative code.
Also generate a Session::SetWeightsToZero() method for building the zero-weight tangent Sessions that forward-mode differentiation needs (the weights are constants, not inputs).
Add a TestCladAutodiff test on a Gemm+Sigmoid model that validates forward-mode directional derivatives against the reverse mode gradient, and a full Hessian, computed via exact Hessian-vector products (reverse-over-forward), against finite differences of the exact gradient. The Clad limitations and workarounds all this rests on are documented in the emitted code and in the test.
🤖 Done with the help of AI
This is needed for full analytic Hessian support of RooFit models that include ONNX functions.