For more information, do take a look at Intel® Distribution of OpenVINO™ Toolkit
Updated, Cleaner API
- The new OpenVINO API 2.0 was introduced, which aligns OpenVINO inputs and outputs with frameworks. Input and output tensors use native framework layouts and element types.
- The API parameters in Model Optimizer have been reduced to minimize complexity. Performance has been significantly improved for model conversion on Open Neural Network Exchange (ONNX*) models.
Broader Model Support
- With Dynamic Input Shapes capabilities on CPU, OpenVINO is able to adapt to multiple input dimensions in a single model providing more complete NLP support. Support for Dynamic Shapes on additional XPUs is expected in a future dot release.
- New models with a focus on NLP and a new category, Anomaly Detection, and support for conversion and inference of select PaddlePaddle* models:
- Pretrained models for anomaly segmentation focus on industrial inspection making speech denoising trainable, plus updates on speech recognition and speech synthesis
- Combined demonstration that includes noise reduction, speech recognition, question answering, translation, and text to speech
- Public models with a focus on NLP ContextNet, Speech-Transformer, HiFi-GAN, Glow-TTS, FastSpeech2, and Wav2Vec
Portability and Performance
- New AUTO plug-in self-discovers available system inferencing capacity based on model requirements so applications no longer need to know their compute environment in advance.
- Automatic batching functionality via code hints automatically scale batch size based on XPU and available memory.
- Built with 12th generation Intel® Core™ processors (formerly code named Alder Lake) in mind. Supports the hybrid architecture necessary to deliver enhancements for high performance inferencing on CPUs and integrated GPUs.