1) What do you think about hybrid approach: hypergraphs + large-scale NLP models (transformers)? 2) How far we're from real self-evolving cognitive architectur

1) What do you think about hybrid approach: hypergraphs + large-scale NLP models (transformers)? 2) How far we're from real self-evolving cognitive architectures with self-awareness features? Is it a question of years, months, or it's already solved problem? 3) Does it make sense to use embeddings like https://github.com/facebookresearch/PyTorch-BigGraph https://github.com/facebookresearch/PyTorch-BigGraph to achieve better results? 4) Why Cycorp decided to limit communication and collaboration with scientific community / AI-enthusiasts at some point? 5) Did you try to solve GLUE / SUPERGLUE / SQUAD challenges with your system? 6) Is Douglas Lenat still contribute actively to the project? Thanks