Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Fix ndarray remainder broadcasting#5002

Merged
laggui merged 1 commit into
tracel-ai:mainfrom
puneetdixit200:fix/4712-ndarray-remainder-broadcast
May 25, 2026
Merged

Fix ndarray remainder broadcasting#5002
laggui merged 1 commit into
tracel-ai:mainfrom
puneetdixit200:fix/4712-ndarray-remainder-broadcast

Conversation

@puneetdixit200
Copy link
Copy Markdown
Contributor

Summary

  • broadcast ndarray remainder operands to their common shape before zipping
  • reuse the shared binary broadcast helper for remainder and comparison-style operations
  • add an ndarray backend regression test for [1, 3] remainder [1, 1]

Fixes #4712

Tests

  • cargo test -p burn-backend-tests --features ndarray --test tensor remainder_broadcast -- --nocapture
  • cargo test -p burn-backend-tests --features ndarray --test tensor remainder -- --nocapture
  • cargo fmt --all --check
  • cargo clippy -p burn-ndarray -- -D warnings
  • git diff --check

Copy link
Copy Markdown
Member

@laggui laggui left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for addressing this issue!

@laggui laggui merged commit 70660fe into tracel-ai:main May 25, 2026
7 of 9 checks passed
@puneetdixit200
Copy link
Copy Markdown
Contributor Author

The current red jobs look unrelated to the remainder change:

  • linux-std-tests (stable) passed the test suites and was canceled during cargo xtask coverage generate with The runner has received a shutdown signal / The operation was canceled.
  • macos-std-tests (stable) starts failing in burn-backend-tests --test autodiff after a WGPU/DSD shader module validation error (Device::create_shader_module_passthrough, labels like unary_float_f_f32_n_2 / matmul_entry...). The remaining autodiff failures look like a cascade after that backend/device error.

I tried to rerun the failed jobs, but GitHub rejected it with Must have admin rights to Repository. Could a maintainer rerun the failed jobs when convenient?

@laggui
Copy link
Copy Markdown
Member

laggui commented May 26, 2026

Totally unrelated, that's why I merged it. The macos CI runners have been unable to run for at least a week 😅 looks like they're back now but we might have to fix some issues. The linux-std-tests fail on code coverage (the runner gets killed, probably using too much memory or disk space).

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.

ndarray: remainder does not support broadcasting

2 participants