Closed ruicoelhopedro closed 6 months ago
Attention: Patch coverage is 60.60606%
with 13 lines
in your changes are missing coverage. Please review.
Project coverage is 93.61%. Comparing base (
91c0501
) to head (3f7e2e8
).
Files | Patch % | Lines |
---|---|---|
piglot/optimisers/botorch/dataset.py | 60.00% | 8 Missing :warning: |
piglot/optimisers/botorch/bayes.py | 61.53% | 5 Missing :warning: |
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Allow specifying the compute device for the PyTorch objects in BoTorch. This allows running the optimisation in the GPU with CUDA which, locally, has been observed to be almost 6x faster than the CPU version.
However, GPU memory may be a limitation. Thus, when a CUDA out-of-memory error is raised, the optimisation falls back to the CPU automatically.
While there, fix an issue with standardising the test dataset in BoTorch.