# 1.0.0-beta ## Highlights - 1.0.0-beta Release * Performance and memory optimizations for DL4J ## Deeplearning4J ### Deeplearning4J: New Features * New or enhanced layers: * Added Cropping1D layer [Link](https://github.com/eclipse/deeplearning4j/blob/master/deeplearning4j/deeplearning4j-nn/src/main/java/org/deeplearning4j/nn/conf/layers/convolutional/Cropping1D.java) * Added Convolution3D, Cropping3D, UpSampling3D, ZeroPadding3D, Subsampling3D layers (all with Keras import support): [Link](https://github.com/eclipse/deeplearning4j/blob/master/deeplearning4j/deeplearning4j-nn/src/main/java/org/deeplearning4j/nn/conf/layers/Convolution3D.java) [Link](https://github.com/eclipse/deeplearning4j/pull/5026) * Added EmbeddingSequenceLayer (EmbeddingLayer for time series) [Link](https://github.com/eclipse/deeplearning4j/blob/master/deeplearning4j/deeplearning4j-nn/src/main/java/org/deeplearning4j/nn/conf/layers/EmbeddingSequenceLayer.java) * Added OCNNOutputLayer (one-class neural network) - implementation of [this paper](https://arxiv.org/pdf/1802.06360.pdf) - [Link](https://github.com/eclipse/deeplearning4j/blob/master/deeplearning4j/deeplearning4j-nn/src/main/java/org/deeplearning4j/nn/conf/ocnn/OCNNOutputLayer.java) * Added FrozenLayerWithBackprop layer [Link](https://github.com/eclipse/deeplearning4j/blob/master/deeplearning4j/deeplearning4j-nn/src/main/java/org/deeplearning4j/nn/conf/layers/misc/FrozenLayerWithBackprop.java) * Added DepthwiseConvolution2D layer [Link](https://github.com/eclipse/deeplearning4j/blob/master/deeplearning4j/deeplearning4j-nn/src/main/java/org/deeplearning4j/nn/conf/layers/DepthwiseConvolution2D.java) * Added ComputationGraph.output(DataSetIterator) method [Link](https://github.com/eclipse/deeplearning4j/issues/4965) * Added MultiLayerNetwork/ComputationGraph.layerInputSize methods [Link](https://github.com/eclipse/deeplearning4j/issues/4670) [Link](https://github.com/eclipse/deeplearning4j/pull/5018) * Added SparkComputationGraph.feedForwardWithKey overload with feature mask support [Link](https://github.com/eclipse/deeplearning4j/issues/4984) * Added MultiLayerNetwork.calculateGradients method (for easily getting parameter and input gradients, for example for some model interpretabilithy approaches) [Link](https://github.com/eclipse/deeplearning4j/pull/5018) [Link](https://github.com/eclipse/deeplearning4j/issues/2866) * Added support to get input/activation types for each layer from configuration: `ComputationGraphConfiguration.getLayerActivationTypes(InputType...)`, `ComputationGraphConfiguration.GraphBuilder.getLayerActivationTypes()`, `NeuralNetConfiguration.ListBuilder.getLayerActivationTypes()`, `MultiLayerConfiguration.getLayerActivationTypes(InputType)` methods [Link](https://github.com/eclipse/deeplearning4j/pull/5031) * Evaluation.stats() now prints confusion matrix in easier to read matrix format, rather than list format [Link](https://github.com/eclipse/deeplearning4j/issues/5096) * Added ModelSerializer.addObjectToFile, .getObjectFromFile and .listObjectsInFile for storing arbitrary Java objects in same file as saved network [Link](https://github.com/eclipse/deeplearning4j/issues/4957) * Added SpatialDropout support (with Keras import support) [Link](https://github.com/eclipse/deeplearning4j/blob/master/deeplearning4j/deeplearning4j-nn/src/main/java/org/deeplearning4j/nn/conf/dropout/SpatialDropout.java) * Added `MultiLayerNetwork/ComputationGraph.fit((Multi)DataSetIterator, int numEpochs)` overloads [Link](https://github.com/eclipse/deeplearning4j/pull/5118) * Added performance (hardware) listeners: `SystemInfoPrintListener` and `SystemInfoFilePrintListener` [Link](https://github.com/eclipse/deeplearning4j/pull/5151) ### Deeplearning4J: Bug Fixes and Optimizations * Performance and memory optimizations via optimizations of internal use of workspaces [Link](https://github.com/eclipse/deeplearning4j/pull/4900) * Reflections library has entirely been removed from DL4J and is no longer required for custom layer serialization/deserialization [Link](https://github.com/eclipse/deeplearning4j/pull/4950), [Link](https://github.com/eclipse/deeplearning4j/pull/4956) * Fixes issues with custom and some Keras import layers on Android * RecordReaderMultiDataSetIterator will no longer try to convert unused columns to numerical values [Link](https://github.com/eclipse/deeplearning4j/pull/4945) * Added new model zoo models: * (to do) * Fixes for Android compilation (removed duplicate classes, aligned versions, removed some dependencies) [Link](https://github.com/eclipse/deeplearning4j/pull/4955) [Link](https://github.com/eclipse/deeplearning4j/pull/5074) [Link](https://github.com/eclipse/deeplearning4j/issues/5087) * Fix for RecordReaderMulitDataSetIterator where output could be incorrect for some constructors [Link](https://github.com/eclipse/deeplearning4j/issues/4969) * Non-frozen layers before a frozen layer will no longer be skipped during backprop (useful for GANs and similar architectures) [Link](https://github.com/eclipse/deeplearning4j/pull/5009) [Link](https://github.com/eclipse/deeplearning4j/issues/4964) * Fixed issue where ComputationGraph topological sort may not be consistent on all platforms; could sometimes break ComputationGraphs (with multiple valid topological orderings) trained on PC and deployed on Android [Link](https://github.com/eclipse/deeplearning4j/pull/5050) * Fixed issue with CuDNN batch norm using `1-decay` instead of `decay` [Link](https://github.com/eclipse/deeplearning4j/pull/5076) * deeplearning4j-cuda no longer throws exceptions if present on classpath with nd4j-native backend set to higher priority [Link](https://github.com/eclipse/deeplearning4j/issues/5000) * Added RNG control for CifarDataSetIterator [Link](https://github.com/eclipse/deeplearning4j/issues/5067) * WordVectorSerializer now deletes temp files immediately once done [Link](https://github.com/eclipse/deeplearning4j/pull/5166) ### Deeplearning4J: API Changes (Transition Guide): 1.0.0-alpha to 1.0.0-beta * WorkspaceMode.SINGLE and SEPARATE have been deprecated; use WorkspaceMode.ENABLED instead * Internal layer API changes: custom layers will need to be updated to the new Layer API - see built-in layers or custom layer example * Custom layers etc in pre-1.0.0-beta JSON (ModelSerializer) format need to be registered before they can be deserialized due to JSON format change. Built-in layers and models saved in 1.0.0-beta or later do not require this. Use `NeuralNetConfiguration.registerLegacyCustomClassesForJSON(Class)` for this purpose * IterationListener has been deprecated in favor of TrainingListener. For existing custom listeners, switch from `implements TrainingListener` to `extends BaseTrainingListener` [Link](https://github.com/eclipse/deeplearning4j/pull/5014) * ExistingDataSetIterator has been deprecated; use `fit(DataSetIterator, int numEpochs)` method instead ### Deelpearning4J: 1.0.0-beta Known Issues * ComputationGraph TrainingListener onEpochStart and onEpochEnd methods are not being called correctly * DL4J Zoo Model FaceNetNN4Small2 model configuration is incorrect, causing issues during forward pass * Early stopping score calculators with values thar should be maximized (accuracy, f1 etc) are not working properly (values are minimized not maximized). Workaround: override `ScoreCalculator.calculateScore(...)` and return `1.0 - super.calculateScore(...)`. ## Deeplearing4J: Keras Import ### Deeplearning4J: Keras Import - API Changes (Transition Guide): 1.0.0-alpha to 1.0.0-beta ## ND4J ### ND4J: New Features ### ND4J: Known Issues * Not all op gradients implemented for automatic differentiation * Vast majority of new operations added in 1.0.0-beta do NOT use GPU yet. ### ND4J: API Changes (Transition Guide): 1.0.0-alpha to 1.0.0-beta ## DataVec ### DataVec: New Features * ImageRecordReader now logs number of inferred label classes (to reduce risk of users missing a problem if something is misconfigured) [Link](https://github.com/deeplearning4j/DataVec/issues/569) * Added AnalyzeSpark.getUnique overload for multiple columns [Link](https://github.com/deeplearning4j/DataVec/issues/71) * Added performance/timing module [Link](https://github.com/deeplearning4j/DataVec/pull/580) ### DataVec: Optimizations and Bug Fixes * Reduced ImageRecordReader garbage generation via buffer reuse [Link](https://github.com/deeplearning4j/DataVec/pull/573) * Fixes for Android compilation (aligned versions, removed some dependencies) [Link](https://github.com/deeplearning4j/DataVec/pull/567) [Link](https://github.com/deeplearning4j/DataVec/pull/575) * Removed Reflections library use in DataVec [Link](https://github.com/deeplearning4j/DataVec/pull/570) * Fix for TransformProcessRecordReader batch support [Link](https://github.com/deeplearning4j/DataVec/issues/561) * Fix for TransformProcessRecordReader with filter operations [Link](https://github.com/deeplearning4j/DataVec/issues/552) * Fixed issue with ImageRecordReader/ParentPathLabelGenerator incorrectly filtering directories containing `.` character(s) [Link](https://github.com/deeplearning4j/DataVec/issues/273) * ShowImageTransform now initializes frame lazily to avoid blank windows [Link](https://github.com/deeplearning4j/DataVec/pull/579) ### DataVec: API Changes (Transition Guide): 1.0.0-alpha to 1.0.0-beta * DataVec ClassPathResource has been deprecated; use nd4j-common version instead [Link](https://github.com/deeplearning4j/DataVec/issues/521) ## Arbiter ### Arbiter: New Features * Added LayerSpace for OCNN (one-class neural network) ### Arbiter: Fixes * Fixed timestamp issue that could cause incorrect rendering of first model's results in UI [Link](https://github.com/deeplearning4j/Arbiter/pull/163) * Execution now waits for last model(s) to complete before returning when a termination condition is hit [Link](https://github.com/deeplearning4j/Arbiter/pull/162) * As per DL4J etc: use of Reflections library has been removed entirely from Arbiter [Link](https://github.com/deeplearning4j/Arbiter/pull/154) * Remove use of Eclipse Collections library due to issues with Android compilation [Link](https://github.com/deeplearning4j/Arbiter/pull/156) * Improved cleanup of completed models to reduce maximum memory requirements for training [Link](https://github.com/deeplearning4j/Arbiter/pull/160) --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://deeplearning4j.konduit.ai/1.0.0-m2/release-notes/1.0.0-beta.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.