Filippo Vicentini
Filippo Vicentini
I am experimenting with https://github.com/cgarciae/treeo and plum, and found out that since both inject meta classes they don't work together. This PR makes plum work with classes which already have...
I experimented with cythonizing the plum.parametric module to speed it up a bit (for my own personal peace of mind). Other than slightly modifying the test because now `typeof` is...
Rebase of #87
This is a list of all Tensorboard plugins and types that they (potentially) can accept. I want to see the overlap among different types. - Scalar (scalar vs iteration line...
Right now the method that one needs to implement to extend `TBLogger` for his own custom types is `preprocess(tag, value, data)` and it modifies the `data` argument. I find this...
`jax.tree_util.Partial` is a PyTree-aware version of `functors.partial` whose leaves (args and keyword args) are arrays. Therefore, I believe it makes sense to make it `serialization`-aware. This allows to use `Partial` ...
Machine-Learning practitioners are aware that initialisation functions must be properly rescaled depending on the variance of the output of the nonlinearity. Howvever, scientistis like Mathematicians/Biologists/Physicists and many others who start...
This is a sketch of the API I'd like to see for Jax-aware operators. I welcome **constructive** ideas and opinions on how to improve this. ## Objective: The objective is...
`ExactState` is a bit misleading, in my opinion, and we should find a better name. Proposals might be: - `FullSummationState` - `FullSpaceState` But maybesomeone has a better idea?
Set a minimum dt that is 1/100 of starting dt for integrators, and document it