[-] kogasa@programming.dev 9 points 6 months ago

I'm aware. You seem to be equivocating on the word "emulate." Nobody called WINE an emulator. The design and behavior of NTSYNC is meant to mimic that of NT synchronization primitives.

[-] kogasa@programming.dev 9 points 1 year ago
[-] kogasa@programming.dev 9 points 1 year ago* (last edited 1 year ago)

I mean the latter statement is not true at all. I'm not sure why you think this. A basic GPT model reads a sequence of tokens and predicts the next one. Any sequence of tokens is possible, and each digit 0-9 is likely its own token, as is the case in the GPT2 tokenizer.

An LLM can't generate random numbers in the sense of a proper PRNG simulating draws from a uniform distribution, the output will probably have some kind of statistical bias. But it doesn't have to produce sequences contained in the training data.

[-] kogasa@programming.dev 9 points 1 year ago

"put the excess energy into batteries" is an idea, and is already pretty much what is done, but the large scale implementation still requires a lot of time, effort, and expense.

[-] kogasa@programming.dev 9 points 2 years ago

If you fine tune a LLM on math equations, odds are it won't actually learn how to reliably solve novel problems. Just the same as it won't become a subject matter expert on any topic, but it's a lot harder to write simple math that "looks, but is not, correct" than it is to waffle vaguely about a topic. The idea of a LLM creating a robust model of the semantics of the text it's trained on is, at face value, plausible; it just doesn't seem to actually happen in practice.

[-] kogasa@programming.dev 9 points 2 years ago

What's the ROI? If 15% of wild caught fish are used to support fish farms that produce twice as much, it's not as obviously a bad thing. There'd need to be another food source though.

[-] kogasa@programming.dev 9 points 2 years ago* (last edited 2 years ago)

All cloud providers will support budget notifications. That doesn't do much good when you shoot past the budget in a short timespan. I set a Google cloud budget of $20/month and enabled a Tensorboard instance, which had no observable indication that it cost anything except the base cost of the VM, and got notified that I was $280 over budget the next day. Apparently there was an upfront $300/month/user fee for Tensorboard. (Several months later they changed the pricing model to $10 GiB/month with no user fee.)

[-] kogasa@programming.dev 9 points 2 years ago

Do I work with you

[-] kogasa@programming.dev 9 points 2 years ago

This is so dumb you should defed with Blahaj in principle.

Fucksake. Can we please stop escalating petty grievances to instance-wide disruptions? It shouldn't even be on a hypothetical table.

[-] kogasa@programming.dev 9 points 2 years ago

The TV firmware could be used to display pirated content, better shut that off

[-] kogasa@programming.dev 9 points 3 years ago

Use pCloud.

[-] kogasa@programming.dev 9 points 3 years ago

Visual Studio is pretty good.

view more: ‹ prev next ›

kogasa

0 post score
0 comment score
joined 3 years ago