OndaBatches.jl provides tooling to enable local and distributed batch loading of Onda-formatted datasets. In particular, it defines utilites to:
- ...associate existing labels to the corresponding signals to create a hollistic dataset.
- ...systematically construct batches of training / evaluation data from this dataset while being flexible enough in our sampling mechanism so that we can tailor the properties of the outputs.
- ...initiate a batching mechanism, carefully controlled by a scheduler, when working in a distributed environment.
Watch our JuliaCon2023 talk on OndaBatches.jl! Slides (and source + demo)