Too bad about no shared code or models, but it is understandable. (I am a deep learning practitioner and I benefit a lot from public models from Google, OpenAI, Facebook, spaCy, and Huggingface - but I understand that some stuff has to be held back for business reasons.)