Hi everyone, thanks to Anastasia Golovina you can ...
# datascience
р
Hi everyone, thanks to Anastasia Golovina you can now use a variety of optimisers (including Adam, AdamW, RMS, Adagrad and SGD) to train TorchScript models from Kotlin - see tests for usage examples.
🙌 1
a
PR into dev and integration with optimization API would be nice.
р
I don't the resources for a cross-platform integration and packaging on the JNI side
a
You don't have to implement it for all platforms. It is OK to do it only for one platform.
If you want a separate release cycle, we can move it to a separate project like @Iaroslav Postovalov’s https://github.com/mipt-npm/kmath-gsl
р
that might be a way forward