twitter-archive / torch-dataset

An extensible and high performance method of reading, sampling and processing data for Torch
Apache License 2.0
76 stars 24 forks source link

Grouped partitions #10

Closed clementfarabet closed 8 years ago

clementfarabet commented 8 years ago

It would be nice to have a system of grouped partitions, useful for creating splits on the fly (training/validation):

trainingPartition = {
   partitions = 10
   partition = {1,2,3,4,5,6,7,8,9}
}
validationPartition = {
   partitions = 10
   partition = {10}
}
egonina commented 8 years ago

@clementfarabet do you want the files to be round-robined to partitions or is it ok to split them into chunks sequentially?

clementfarabet commented 8 years ago

Closing – this overloading of partitions is too confusing.