# Selman + Felix masters thesis meeting * https://github.com/NVIDIA-Merlin/dataloader * https://xarray.dev/blog/xarray-kvikio * Felix * Loading data via Merlin into pytoch cuDF * Number of epochs * Uber petastorm (something on spark) * Felix – description of tasks * Benchmarking the loading * Making sure loading is the issue * How are we loading, what are the mechanisms of loading? * Model performance, effect of loading by batch - need for randomizing * Randomization of chunks * https://github.com/libffcv/ffcv * Masters thesis *