Avik Pal
Selecting an algorithm for SVDFactorization will cause it to fail for CUDA. Possible solutions: 1. Keep it as `nothing` and use the default versions used in the downstream package
1. We can't apply it to the entire Lux module. Some of the code exists for debugging and is intentionally type unstable. 2. I like the default of erroring, but...
Fixes #735 #682
Pulling out training logic from #665, because this is simpler to implement and merge. ## Example Usage We can simply replace `AutoEnzyme` or any of those backends with `AutoReactant` and...
## Features - [ ] Move Training utilities out of Experimental. One major recent push has been to clean up the training and ensure it is used everywhere in tutorials....
## TODOs - [ ] Write a script to consolidate the data from bson files into a dataframe or something easy to extend later on - [ ] Write a...
## Example Usage This follows the same structure as SimpleChains. User demands a conversion and provides an input prototype. ```julia using Reactant, Lux, Random model = Chain(Dense(10, 5, tanh), Dense(5,...
## Functions - [ ] `batched_jacobian` - [ ] `Enzyme.Forward` - [ ] `Enzyme.Reverse` - [ ] `vector_jacobian_product` - [ ] `jacobian_vector_product` ## Nested Autodiff Enzyme needs us to use...