Agustinus Kristiadi
Agustinus Kristiadi
E.g. using https://github.com/danielfrg/mkdocs-jupyter which integrates `ipynb` into `mkdocs`
Use the idiomatic best-practices, e.g. inherit `ABC` in `BaseLaplace` etc.
We can achieve the same thing by switching off the grads of all but the last layer. ```python model = ... for mod_name, mod in model.named_modules(): if mod_name != "last_layer":...
Now that #243 is merged, it's useful to have some examples of how to leverage `la.functional_samples()` in practical applications. My plan is to create two examples (can be merged as...
Latest `uv` version now support easy command to build & publish. https://docs.astral.sh/uv/guides/publish/ The dev guide docs should be updated.
Also, we need to streamline them. For example, currently the functional Laplace example requires `deepobs` and hence `tensorflow`.
Currently `KronLaplace` doesn't work with arbitrary batch shape since `matrix.py` checks for at most 3-dim tensor. This is problematic when multiplying with Jacobians of shape `(..., n_classes, n_params)`. See https://github.com/aleximmer/Laplace/blob/glm-multidim/examples/lm_example.py...