2 ms·
Are the PBs of free space necessary? I thought training a ML model on a terabyte of data was computation expensive, but not storage intensive.
by HWR_14 5y ago
Are the PBs of free space necessary? I thought training a ML model on a terabyte of data was computation expensive, but not storage intensive.
- a_bonobo 5y agoSometimes! Not for the GPU part, but to make the summary data for the ML data. For example, genome sequencing data and intermediate results are easily in the TB area of space, but the resulting table of genomic variants (k-mers like n-grams in NLP) is only a few hundred GB.