FluxML/Flux.jl
Allow BatchNorm training on CUDA with `track_stats=false`
Chiusa
#1606 aperta il 31 mag 2021
good first issuehelp wanted
Metriche repository
- Star
- (4725 stelle)
- Metriche merge PR
- (Merge medio 3g 6h) (6 PR mergiate in 30 g)
Descrizione
Gathered from https://discourse.julialang.org/t/batchnorm-only-track-stats-true-supported-on-gpu/62091.
This would most likely require changes in NNlibCUDA as well. I'm not sure how interchangeable the various cudnnBatchNormalizationForward* functions are, so putting a pin in this until someone more knowledgeable can comment.