It could be to reduce discrepancy (and the disappointment) between marketing and reality.
Reminds me of this monologue from the 1993 movie Falling Down [1]:
> See, this is what I'm talking about. Look at that. See what I mean? It's plump, juicy, three inches thick. Look at this sorry, miserable, squashed thing. Can anybody tell me what's wrong with this picture?
In rpn notation you just put the input on the stack, right? The encodings seems like they could get pretty big, and encodings certainly wouldn't be unique, but you should be able to encode pretty much any constant you could think of.
> Nice list! I'd say the SQLite with WAL is the biggest money saver mentioned.
Funny you said that. I migrated an old, Django web site to a slightly more modern architecture (docker compose with uvicorn instead of bare metal uWSGI) the other day, and while doing that I noticed that it doesn't need PostgreSQL at all. The old server had it already installed, so it was the lazy choice.
I just dumped all data and loaded it into an SQLite database with WAL and it's much easier to maintain and back up now.
> That sounds like straight up scammer behavior. "
Microsoft reached out to the police department, then the person went to the local police department to verify who they were. I don't see how this could be a scam.
I think there are a few scammer red flags in this - it stood out to me that they said "support watched me setup 3 different accounts" - not saying MS support couldn't do this, but remoting into the machine and watching a victim enter form details is a very scammer-y thing for sure
Ignoring the training costs, the marginal cost for inference is pretty low for providers. They are estimated to break even or better with their $20/month subscriptions.
That being said, they can't stop launching new models, so training is not a one time task. Therefore one might argue that it is part of the marginal cost.
reply