We lifted 3 functions out of this paper's own repositories and ran 1 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| f-dangel/hbp | canonical | 1 of 3 |
| Function | Status | Where it lives |
|---|---|---|
| batch_summed_hessian | Ran | f-dangel/hbp/bpexts/hbp/loss.py code served (permissive licence) · get_code("7a9d3012348d3892") |
| hbp_decorate | Not yet run | f-dangel/hbp/bpexts/hbp/module.py code served (permissive licence) · get_code("1cace52fec18df3f") |
| hbp_elementwise_nonlinear | Not yet run | f-dangel/hbp/bpexts/hbp/nonlinear.py code served (permissive licence) · get_code("0043d66c29c327c1") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
We propose a modular extension of backpropagation for the computation of block-diagonal approximations to various curvature matrices of the training objective (in particular, the Hessian, generalized Gauss-Newton, and positive-curvature Hessian). The approach reduces the otherwise tedious manual derivation of these matrices into local modules, and is easy to integrate into existing machine learning libraries. Moreover, we develop a compact notation derived from matrix differential calculus. We outline different strategies applicable to our method. They subsume recently-proposed block-diagonal approximations as special cases, and are extended to convolutional neural networks in this work.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("1902.01813")
get_code_for_paper("1902.01813")
have("1902.01813")
Connect an agent — have() is free.