We lifted 15 functions out of this paper's own repositories and ran 0 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 |
|---|---|---|
| pwlo3k46/vivit | canonical | 0 of 9 |
| jbzre7bp/vivit | canonical | 0 of 4 |
| zv5dmjq5/vivit | canonical | 0 of 2 |
| Function | Status | Where it lives |
|---|---|---|
| approximate_boundaries | Not yet run | jbzre7bp/vivit/vivit/hessianfree/lanczos.py code served (permissive licence) · get_code("8bcff92887677b71") |
| approximate_boundaries_abs | Not yet run | jbzre7bp/vivit/vivit/hessianfree/lanczos.py code served (permissive licence) · get_code("e4d94e72d53429eb") |
| fast_lanczos | Not yet run | jbzre7bp/vivit/vivit/hessianfree/lanczos.py code served (permissive licence) · get_code("c38616bc34dad46d") |
| get_letters | Not yet run | pwlo3k46/vivit/vivit/utils/gram.py code served (permissive licence) · get_code("3b29da7c5bb6f1c4") |
| is_subset | Not yet run | pwlo3k46/vivit/vivit/utils/subsampling.py code served (permissive licence) · get_code("b4944136575f91af") |
| make_top_k_criterion | Not yet run | jbzre7bp/vivit/exp/utils/vivit.py code served (permissive licence) · get_code("e157c276ee749806") |
| merge_subsamplings | Not yet run | pwlo3k46/vivit/vivit/utils/subsampling.py code served (permissive licence) · get_code("26c378565ba49153") |
| pairwise2_dot | Not yet run | pwlo3k46/vivit/vivit/utils/gram.py code served (permissive licence) · get_code("0049e73ca8f9bee3") |
| pairwise_dot | Not yet run | pwlo3k46/vivit/vivit/utils/gram.py code served (permissive licence) · get_code("42d68d5c80b5db4e") |
| sample_output_mapping | Not yet run | pwlo3k46/vivit/vivit/utils/subsampling.py code served (permissive licence) · get_code("2e09290d345742fd") |
| shift_diag | Not yet run | pwlo3k46/vivit/vivit/utils/eig.py code served (permissive licence) · get_code("facdac15f2c804c0") |
| stable_symeig | Not yet run | pwlo3k46/vivit/vivit/utils/eig.py code served (permissive licence) · get_code("e044b617cfd2c895") |
| stable_symeig | Not yet run | zv5dmjq5/vivit/vivit/utils/eig.py code served (permissive licence) · get_code("7cf1a1d2a94fd903") |
| symeig_psd | Not yet run | pwlo3k46/vivit/vivit/utils/eig.py code served (permissive licence) · get_code("6667b9a2a60b7fe1") |
| symeig_psd | Not yet run | zv5dmjq5/vivit/vivit/utils/eig.py code served (permissive licence) · get_code("484f66d87c0c4751") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Curvature in form of the Hessian or its generalized Gauss-Newton (GGN) approximation is valuable for algorithms that rely on a local model for the loss to train, compress, or explain deep networks. Existing methods based on implicit multiplication via automatic differentiation or Kronecker-factored block diagonal approximations do not consider noise in the mini-batch. We present ViViT, a curvature model that leverages the GGN's low-rank structure without further approximations. It allows for efficient computation of eigenvalues, eigenvectors, as well as per-sample first- and second-order directional derivatives. The representation is computed in parallel with gradients in one backward pass and offers a fine-grained cost-accuracy trade-off, which allows it to scale. We demonstrate this by conducting performance benchmarks and substantiate ViViT's usefulness by studying the impact of noise on the GGN's structural properties during neural network training.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2106.02624")
get_code_for_paper("2106.02624")
have("2106.02624")
Connect an agent — have() is free.