clj-djl.training.dataset — clj-djl 0.1.9

array-dataset clj

(array-dataset {:keys [data labels sampler batchsize shuffle droplast]})

array-dataset-builder clj

(array-dataset-builder)

batch-sampler clj

(batch-sampler subsampler batchsize)

(batch-sampler subsampler batchsize droplast)

Creates a new instance of BatchSampler that samples from the given SubSampler, and yields a mini-batch of batchsize, with optional droplast(true, false) to drop the remaining samples.

build clj

(build builder)

close clj

(close batch)

get-data clj

(get-data batch)

get-data-iterator clj

(get-data-iterator ds manager)

get-labels clj

(get-labels batch)

iter-seq clj

(iter-seq iterable)

(iter-seq iterable iter)

new-array-dataset-builder clj

opt-labels clj

(opt-labels builder & labels)

opt-usage clj multimethod

prepare clj

(prepare ds)

(prepare ds progress)

random-sampler clj

(random-sampler)

(random-sampler seed)

Creates a new instance of RandomSampler with an optional seed

random-split clj

(random-split dataset & ratios)

sequence-sampler clj

(sequence-sampler)

SequenceSampler samples the data from [0, dataset.size) sequentially.

set-data clj

(set-data builder & data)

set-sampling clj

(set-sampling builder sampler)

(set-sampling builder batch-size random)

(set-sampling builder batch-size random drop-last)

to-apair clj

(to-apair dataset)

Convert dataset to a pair of two arrays. First item is the data, and the second item is the labels