Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add a faster activation path #558

Closed
wants to merge 6 commits into from
Closed

Add a faster activation path #558

wants to merge 6 commits into from

Conversation

avik-pal
Copy link
Member

@avik-pal avik-pal commented Mar 21, 2024

  • sigmoid_fast fails on GPU as pointed out in the NNlib PR. Need to work around that
  • Specialize WrappedFunction on Broadcast.BroadcastFunction --> can't do this mutation will violate purity wrt input/output arrays

Remaining Problems:

  • Type Stability failure for DEQs
  • NeuralPDE failure
  • Baaad rebase

@avik-pal avik-pal linked an issue Mar 21, 2024 that may be closed by this pull request
@avik-pal avik-pal force-pushed the ap/fast_broadcast branch 16 times, most recently from 3622541 to a8f04bb Compare March 26, 2024 18:59
@avik-pal avik-pal force-pushed the ap/fast_broadcast branch 2 times, most recently from 617c7d2 to 64029bb Compare April 4, 2024 03:25
@avik-pal
Copy link
Member Author

See LuxDL/LuxLib.jl#53 for a cleaner approach

@avik-pal
Copy link
Member Author

Absorbed into #591

@avik-pal avik-pal closed this Apr 20, 2024
@avik-pal avik-pal deleted the ap/fast_broadcast branch April 28, 2024 01:38
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Improve broadcasting via FastBroadcast.jl
1 participant