Sam Anklesaria
Sam Anklesaria
Is there currently a way to cast a funsor with `real` output to `bint`? Or vice versa? This would be particularly useful for indexing.
It seems like Optimisers only works with vectors. ``` struct WithFloat val :: Float64 end @functor WithFloat ``` The call to `Optimisers.trainable(WithFloat(4))` returns `(val = 4.3,)`, but `destructure(WithFloat(4))` produces an...
Might it be possible to add the random Fourier features approximation? Here's a draft pull request.
PLEASE NOTE THAT THE TORCHAUDIO REPOSITORY IS NO LONGER ACTIVELY MONITORED. You may not get a response. For open discussions, visit https://discuss.pytorch.org/. This is a draft PR for testing the...
Update mnist example to use NNX. Closes #5054. Same as #5002, only not coming from copybara (so potentially easier to merge with changers).
Fixes #5130 This does not remove the Fingerprint check (also mentioned in the issue). This just fixes the pop bug.
# What does this PR do? This PR adds a guide that shows some common techniques for working with Flax models during optimization. These include: - Calculation of Exponential Moving...
This PR addresses #5067. Currently, the `tabulate` function tries to `jit` compile every single method of the input object and its decedents so as to support the 'compute_flops' option. If...
This PR improves the variable sharding API. Specifically: - It renames sharding_names to sharding_metadata (still supporting sharding_names as a property) - It generalizes the spmd APIs to allow sharding_metadata to...