Deep learning data sets can be quite large. There is a directory /nobackup/datasets which contains a single resenet50 dataset for the WMLCE tensorflow-benchmarks Resnet50 benchmark.
A Bede user asked on slack if there were any centrally avialable datasets, as these can be very large so common data sets being available from a central accessible location can save time, network and disk.
It may be worth clearly documenting (with a new section/page) the available datasets, where they are stored and (potentially) how to request new central datasets (if Bede support wishes to provide this)?
Deep learning data sets can be quite large. There is a directory
/nobackup/datasets
which contains a single resenet50 dataset for the WMLCEtensorflow-benchmarks
Resnet50 benchmark.A Bede user asked on slack if there were any centrally avialable datasets, as these can be very large so common data sets being available from a central accessible location can save time, network and disk.
It may be worth clearly documenting (with a new section/page) the available datasets, where they are stored and (potentially) how to request new central datasets (if Bede support wishes to provide this)?