Hey fellow chatters.
I’ve been using the code interpreter for cooperative brainstorming of some unique approaches, and I’m very happy that GPT can use torch.
However, I was quite surprised to find that it didn’t have access to basic “toy” or “learning” datasets. It would not only be useful for many users learning data science but also for trying out some algorithms.
I’d love to see OpenAI making it possible for GPT to load small basic data sets like MNIST, Income Census and other simple basic datasets. I understand network features are disabled for a good reason, but I reckon these could always be made available locally.
What do you think?