FluxTraining.jl

			
			
			
			module
			
			 
			FluxTraining
			
			

			

			

			
			using
			
			 
			Graphs
			

			
			
			using
			
			 
			BSON
			:
			
			 
			@
			load
			,
			
			 
			@
			save
			

			
			using
			
			 

	
			Flux
			

			
			
			using
			
			 

	
			Flux
			:
			
			 
			Params
			,
			
			 
			onecold
			

			
			
			using
			
			 

	
			Flux
			.
			Optimise
			:
			
			 
			update!
			

			
			using
			
			 
			ImageCore
			

			
			using
			
			 
			InlineTest
			

			
			using
			
			 
			Glob
			

			
			module
			
			 

	
			ES
			
			
    
			
			using
			
			 
			Reexport
			
    
			
			@
			reexport
			 
			
			using
			
			 

	
			EarlyStopping
			

			end
			

			
			import
			
			 
			OnlineStats
			

			
			
			using
			
			 
			OnlineStats
			:
			
			 
			EqualWeight
			,
			
			 
			Mean
			,
			
			 
			OnlineStat
			

			
			import
			
			 
			Optimisers
			

			
			using
			
			 
			Parameters
			

			
			
			using
			
			 
			ProgressMeter
			:
			
			 
			Progress
			,
			
			 
			next!
			

			
			
			using
			
			 
			Statistics
			:
			
			 
			mean
			

			
			using
			
			 
			UUIDs
			

			
			using
			
			 
			Zygote
			

			
			using
			
			 
			ChainRulesCore
			

			
			using
			
			 
			ParameterSchedulers
			

			
			
			using
			
			 
			TensorBoardLogger
			:
			
			 
			TBLogger
			,
			
			 
			log_value
			,
			
			 
			log_image
			,
			
			 
			log_text
			,
			
			 
			log_histogram
			,
			
			 
			tb_overwrite
			

			
			
			using
			
			 
			Zygote
			:
			
			 
			Grads
			,
			
			 
			gradient
			

			
			using
			
			 
			ValueHistories
			

			
			
			using
			
			 
			DataStructures
			:
			
			 
			DefaultDict
			

			
			using
			
			 
			PrettyTables
			

			
			
			using
			
			 
			Setfield
			:
			
			 
			@
			set
			

			

			
			import
			
			 
			SnoopPrecompile
			

			

			# functional
			

			
			include
			(
			
			"
			./functional/metrics.jl
			"
			)
			

			

			# callback system
			

			
			include
			(
			
			"
			./callbacks/protect.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/phases.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/events.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/callback.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/graph.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/execution.jl
			"
			)
			

			

			# logging
			

			
			include
			(
			
			"
			./callbacks/logging/Loggables.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/logging/logger.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/logging/tensorboard.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/logging/checkpointer.jl
			"
			)
			

			

			

			# callback implementations
			

			
			include
			(
			
			"
			./callbacks/conditional.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/callbacks.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/earlystopping.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/custom.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/metrics.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/recorder.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/trace.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/sanitycheck.jl
			"
			)
			

			

			# hyperparameter scheduling
			

			
			include
			(
			
			"
			./callbacks/hyperparameters.jl
			"
			)
			

			
			include
			(
			
			"
			./callbacks/scheduler.jl
			"
			)
			

			

			

			# learner
			

			
			include
			(
			
			"
			./learner.jl
			"
			)
			

			
			include
			(
			
			"
			./callbackutils.jl
			"
			)
			

			

			
			include
			(
			
			"
			./training.jl
			"
			)
			

			

			
			include
			(
			
			"
			testutils.jl
			"
			)
			

			

			

			
			
			SnoopPrecompile
			.
			
			@
			precompile_all_calls
			 
			
			begin
			
    
			
			learner
			 
			=
			 
			

	
			testlearner
			(
			)
			
    
			

	
			fit!
			(
			learner
			,
			 
			1
			)
			

			end
			

			

			

			
			export
			 

	
			AbstractCallback
			,
			
    

	
			Loss
			,
			
    

	
			ConditionalCallback
			,
			
    

	
			CancelStepException
			,
			
    

	
			CancelEpochException
			,
			
    

	
			CancelFittingException
			,
			
    

	
			Checkpointer
			,
			
    

	
			CustomCallback
			,
			
    

	
			EarlyStopping
			,
			
    

	
			ToDevice
			,
			
    

	
			ToGPU
			,
			
    

	
			GarbageCollect
			,
			
    

	
			Learner
			,
			
    

	
			Metric
			,
			
    

	
			Recorder
			,
			
    

	
			ProgressPrinter
			,
			
    

	
			Metrics
			,
			
    

	
			MetricsPrinter
			,
			
    

	
			Traces
			,
			
    

	
			TrainingPhase
			,
			
    

	
			ValidationPhase
			,
			
    
			Schedule
			,
			
    

	
			Scheduler
			,
			
    

	
			LogMetrics
			,
			
    

	
			SmoothLoss
			,
			
    

	
			LogTraces
			,
			
    

	
			LogHistograms
			,
			
    

	
			LogHyperParams
			,
			
    

	
			LogVisualization
			,
			
    

	
			TensorBoardBackend
			,
			
    

	
			StopOnNaNLoss
			,
			
    

	
			LearningRate
			,
			
    

	
			throttle
			,
			
    

	
			SanityCheck
			,
			
    

	
			accuracy
			,
			
    

	
			fit!
			,
			
    

	
			epoch!
			,
			
    

	
			step!
			,
			
    

	
			onecycle
			,
			
    

	
			loadmodel
			,
			
    

	
			savemodel
			

			end