I did look and I apologize if I’ve missed this question/answer in another post. In older courses, I’ve found the “deeplearning-platform-release” image on GCP with fastai pre-installed very helpful. I have installed the dev v2 env on GCP on a vanilla “deeplearning-platform-release” image in a separate directory and installed the beta fastbook inside that directory. It works and I can run the v1 and v2 notebooks side by side with Jupyter on port 8080, but it’s not ideal and it’s all beta book/beta code. My question is whether/when there will be a “deeplearning-platform-release” image with V2 on GCP? That would make setup of V2, the new course, and the book very easy. I make enough mistakes that I need to wipe out the instance and have a do-over and the pre-configured instance would really be nice. Thanks so much for all the incredible work you guys have done on the API, the course and the book!
Until the GCP images are updated with newer cuda drivers, the instructions in this post work for the fastai 2 release: