Geforce RTX Family CUDA Compute Capability: DNN Training and Inference?

The webpage at [url][/url] doesn’t list any of the RTX family products. These products have tensor cores. Are they CUDA Compute 7.0-compatible? Can they be used for DNN training and inference?

So apparently the RTX family is “Compute 7.5”. This means all the tensor core functionality available on the Volta line is also available within Turing?