sci-ml/torchmetrics
Machine learning metrics for distributed PyTorch applications
ChangeLog
commit 8285b31253ced9e74ae4eef9a616ff0e8d36205b
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sat May 16 10:08:42 2026 +0200
sci-ml/torchmetrics: new package, add 1.9.0
Machine learning metrics for distributed PyTorch applications.
Used by sci-ml/lightning and (transitively) by sci-ml/pyannote-audio
for diarization error rate, etc.
SINGLE_IMPL via pytorch. Upstream's CI-only torch<2.9.0 cap is
stripped from published PyPI metadata; the ebuild deliberately
omits the cap so the package solves against sci-ml/pytorch 2.10+.
Tests restricted (long tail of audio/text/image extras).
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sat May 16 10:08:42 2026 +0200
sci-ml/torchmetrics: new package, add 1.9.0
Machine learning metrics for distributed PyTorch applications.
Used by sci-ml/lightning and (transitively) by sci-ml/pyannote-audio
for diarization error rate, etc.
SINGLE_IMPL via pytorch. Upstream's CI-only torch<2.9.0 cap is
stripped from published PyPI metadata; the ebuild deliberately
omits the cap so the package solves against sci-ml/pytorch 2.10+.
Tests restricted (long tail of audio/text/image extras).


View
Download
Browse