Skip to content

Conversation

@danieldk
Copy link
Contributor

This PR adds various bounds checks to NumpyOps to prevent out-of-bound reads and writes:

  • Check that which indices are in bounds for backprop_{maxout, reduce_max}.
  • Check that dY and X shapes match in backprop for elementwise functions.
  • Hook up ReLU to the PyTorch comparison tests.
  • Check that lengths are valid in backprop_reduce_{mean, max, sum}.

@danieldk danieldk marked this pull request as draft March 17, 2022 12:27
@danieldk danieldk marked this pull request as ready for review March 18, 2022 08:38
@danieldk
Copy link
Contributor Author

@explosion-bot please test_gpu

@explosion-bot
Copy link
Collaborator

explosion-bot commented Mar 18, 2022

🪁 Successfully triggered build on Buildkite

URL: https://buildkite.com/explosion-ai/thinc-gpu-test-suite/builds/34

@honnibal
Copy link
Member

Nice! Love the advanced syntax for the nogil except. That's either relatively new, or I just never knew about it 🤯 . Either way, neat.

@svlandeg svlandeg added enhancement Feature requests and improvements feat / ops Backends and maths labels Mar 29, 2022
@svlandeg svlandeg merged commit ed41ef9 into explosion:develop Mar 29, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

enhancement Feature requests and improvements feat / ops Backends and maths

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants