I didn't read the full paper yet, but a quick ctrl+f shows no mention of Activation Atlas for vision neural network, a collaboration between Google AI and OpenA

I didn't read the full paper yet, but a quick ctrl+f shows no mention of Activation Atlas for vision neural network, a collaboration between Google AI and OpenAI : https://ai.googleblog.com/2019/03/exploring-neural-networks.html?m=1 https://ai.googleblog.com/2019/03/exploring-neural-networks.... Also no mention of the "Unforeseen Attack Robustness" metric by OpenAI : https://openai.com/blog/testing-robustness/ https://openai.com/blog/testing-robustness/ There are probably other publications for NLP models, all the big players are aware that explanation is key.