We've recently released a preprint paper about our TinyTroupe open-source project developed at Microsoft. This is still very much under research and development, but the paper is meant to consolidate much of the principles and findings behind it so far. As this remains a preprint manuscript, any suggestion, important related work or constructive comment would be helpful to improve the paper and is very much welcomed. In particular, if you know of good specialized benchmarks to evaluate this kind of thing, please share. Thank you!
In short, Prompt Engineering is just one of the pieces in building a GPT-3/LLM-based solution. In fact, I'd say a whole new set of Software Engineering best practices is necessary and will gradually emerge. I gave one such approach that has been useful to my related projects.
Indeed. I do know Azure has special provisions for compliant computing for a couple of industries. Regarding GPT-3, besides the original OpenAI offer, it is now possible to consume it from Azure instead. My understanding is that this is not just rebranding, but actually different machines and additional software. Azure folks have their own privacy constraints, which I guess is meant to address PII concerns, among others. Here's the link:https://azure.microsoft.com/en-us/products/cognitive-service...