Skip to content

[Feature] - Mini batch training

Mohib Mustafa requested to merge dev_mm into main
  • Training can be done in mini batches to accomodate cpu or gpu memmory restrictions.
  • Option to force select gpu or cpu.
  • Storage of normalized padded data (Total, testing and training) changed from 'CollectedData' to 'RNNSurrogate'.
  • Driver script is appended with 2 control variables, 'default' which controls choice of gpu or cpu and 'mini_batch' which controls the size of mini batches
  • Updated readme.
  • Removed useless comments

Merge request reports