Sometimes it comes down to having a third party service handle support or provide the framework, so adding it as a subdomain makes it easier to integrate into your site. And Google often treats subdomains as their own site so you can have a well structured support driven layout, rather than a site trying to showcase featured content while battling with support content.
I hate it when a large grocery store doesn't offer self checkout. I don't want to have to stand and wait in line for 15-20 minutes when few lanes are open and people in front of me have full baskets, when I only have a handful of items.
Since the introduction of AI, doctors using AI to help in detection (not as a replacement to their own work) has been the driving force in improving healthcare outcomes. There may come a time where AI can be the sole source in locations that have a deficit of doctors (cause something is better than nothing), but I don't see it changing any time soon.
It's tough to top their arc with Infinity Way/Endgame. They definitely need to regain focus on where they want to go since they are introducing so many new characters.
TorchArrow is a machine learning preprocessing library over batch data, providing performant and Pandas-style easy-to-use API for model development. Currently it provides a Python DataFrame that allows extensible UDFs with Velox, with the following features:
- Seamless handoff with PyTorch or other model authoring, such as Tensor collation and easily plugging into PyTorch DataLoader and DataPipes
- Zero copy for external readers via Arrow in-memory columnar format
- Multiple execution runtimes support:
- High-performance C++ UDF support with vectorization
Yea, and I understand researchers may just be training models but don't intend to move into production or deploy once they fine tune them, especially in an education environment or when using them as a basis for a research paper.
100% look for a new job if the current employer is not willing to enable your success. It's completely reasonable as a developer to expect a work computer that allows you to do your job effectively. If they gave you an older model that doesn't perform well, make sure you document on the performance and detail how its limitations are holding up your work. You don't need to request some supercomputer but a reasonable alternative to what you have now. If they refuse without giving any reasonable explanation, then move on.
Interesting to see a couple use case scenarios that give the win to each, so there doesn't seem to be a clear cut overall winner. Just depends on what you are looking to do.