- New speech-to-text APIs that enables us doing language model post processing
- APIs for Image captioning
- More example notebooks
- Updated weights (located: http://support.sas.com/documentation/prod-p/vdmml/zip/index.html) supporting new version of ONNX
- Various bug fixes
- New layers: transpose convolution and group convolution
- New architectures: U-Net, MobileNet, and ShuffleNet
- Supporting FasterRCNN for object detection
- Supporting image segmentation
- Various bug fixes
- Improved documentation
- Improved code coverage
- Various bug fixes
- A new way of creating models
- Multi-step forecasting
- Various bug fixes
- ONNX import and export support
- Re-organization of examples
- Text, audio, and time series support in addition to image
- APIs for RNN based tasks: text classification, text generation, and sequence labeling
- APIs for object detection
- APIs for time series processing and modeling
- APIs for processing audio files and creating speech recognition models
- Additional pre-defined network architectures such as DenseNet, DarkNet, Inception, and Yolo with enhanced data visualization and metadata handling