Neural Network Framework Version 12 completes its high-level neural network framework in terms of functionality, while improving its simplicity and performance. A variety of new layers and encoders have been added, in particular, to handle sequential data such as text or audio. Importantly, a model repository is introduced, bringing a collection of pre-trained networks to be used as is, symbolically manipulated, or fine-tuned to a specific task. Finally, performance has been improved for most hardware configurations, including the possibility to train using multiple graphic cards. This framework is arguably the easiest existing tool to build neural network applications, yet with broad capabilities and without performance compromise. Access pre-trained models. » Define and visualize arbitrary nets. » Manipulate neural net symbolically. » Access detailed training information. » Measure net performances. » Train a network on multiple GPUs. » Train networks on text or audio data. » Create attention mechanisms. » Define custom recurrent layers. » Generate sequences from recurrent layers efficiently. » Train convolution nets on sequences. » Train transformer nets. » Train capsule nets. » Train self-normalizing neural nets. » Access reinforcement learning environments. » Related Examples Obtain and Use Pre-trained Models » Use Pre-trained Models to Visualize Features » Perform Net Surgery for Transfer Learning » Visualize the Insides of a Neural Network » Work Flexibly with Recurrent Nets » Use Transformer Neural Nets » Use Capsule Nets » Train a Self-Normalizing Neural Net » Train an Audio Classifier » Train a Net to Model English » Segment Images Semantically » Train an Agent in a Reinforcement Learning Environment »