clj-djl.training — clj-djl 0.1.4

accuracy clj

(accuracy)

add-accumulator clj

(add-accumulator acc key)

Adds an accumulator to the accuracy for the results of the evaluation with the given key.

add-evaluator clj

(add-evaluator config evaluator)

add-training-listeners clj

(add-training-listeners config listener)

as-consumer clj macro

(as-consumer f)

attach-gradient clj

(attach-gradient ndarray)

Attaches a gradient NDArray to this NDArray and marks it so GradientCollector.backward(NDArray) can compute the gradient with respect to it.

backward clj

(backward gc target)

close clj

(close batch)

default-training-config clj

(default-training-config {:keys [loss devices data-manager initializer optimizer
                                 evaluator listeners]})

fit clj

(fit trainer nepochs train-iter)

(fit trainer nepochs train-iter test-iter)

forward clj

(forward trainer input)

get-accumulator clj

(get-accumulator acc key)

Returns the accumulated evaluator value.

get-devices clj

(get-devices config)

get-evaluators clj

(get-evaluators trainer)

get-gradient clj

(get-gradient ndarray)

Returns the gradient NDArray attached to this NDArray.

get-loss clj

(get-loss trainer)

get-manager clj

(get-manager trainer)

get-metrics clj

(get-metrics trainer)

Get metrics from trainer, put the metrics to seq of map: [{"train_progress_Accuracy" {:timestamp 1607859588747 :value 0.68125 :unit "count"}}]

get-training-result clj

(get-training-result trainer)

gradient-collector clj

(gradient-collector)

initialize clj

(initialize trainer shapes)

(initialize trainer shape & shapes)

iter-seq clj

(iter-seq iterable)

(iter-seq iterable iter)

iterate-dataset clj

(iterate-dataset trainer ds)

metrics clj

(metrics)

new-accuracy clj

new-default-training-config clj

(new-default-training-config loss)

new-default-training-listeners clj

new-gradient-collector clj

new-progress-bar clj

new-topk-accuracy clj

new-trainer clj

new-training-config clj

notify-listeners clj

(notify-listeners trainer callback)

opt-initializer clj

(opt-initializer config initializer)

opt-optimizer clj

(opt-optimizer config optimizer)

parameter-store clj

(parameter-store manager copy)

progress-bar clj

(progress-bar)

set-metrics clj

(set-metrics trainer metrics)

softmax-cross-entropy-loss clj

(softmax-cross-entropy-loss)

step clj

(step trainer)

topk-accuracy clj

(topk-accuracy topk)

(topk-accuracy index topk)

(topk-accuracy name index topk)

train-batch clj

(train-batch trainer batch)

trainer clj

(trainer {:keys [model loss devices data-manager initializer optimizer]})

(trainer model config)

training-config clj

training-listeners clj

(training-listeners)

update-accumulator clj

(update-accumulator acc key label-list pred-list)

Updates the accuracy with the given key based on a NDList of labels and predictions.

validate-batch clj

(validate-batch trainer batch)