deepmodeling / deepmodeling/deepmd-kit
Re-enable this in GitHub CUDA CI once the hosted/self-hosted
Open
branch: master
lang: py
todo comment
type: todo
- Dominant language
- Python
- Stars
- 2k
- Forks
- 649
- Avg merge
- 6d 18h
- Merged PRs (30d)
- 15
Description
runner JAX/PJRT abort is understood. The same test passes on a
local GPU, but the GitHub Actions CUDA job can terminate with
CUDA_ERROR_LAUNCH_FAILED while PJRT releases device buffers.
Line: 120
https://github.com/deepmodeling/deepmd-kit/blob/20874166a20a3b9dc77e3a5a68f6a1018dce37a6/source/tests/jax/test_training.py#L117-L127
Contributor guide
Assessment
This issue has not been assessed yet.