And, I guarantee you, “trade deficit” and “budget deficit”; hence, tariffs.

Also hasn’t she been linked with Mark Sanford? Given that, drawing attention to hiking is an interesting choice.

I think I’m “five closet doors(?) built into floor of Main Room annex”

peoples-front-of-judea-vs-judean-peoples-front-dot-gif

To be fair, Pink Lady is unwise marketing if you’re in the Grease Extended Universe.

… loe

Please share if you get an answer. I have a friend in a similar situation, except the dead end they hit is “your account is suspended due to violating our terms of use”. Well, duh, that’s what they’re trying to fix.

Absolutely. Gaiman in particular wrote so many things that have become core to who I am now. I reread Sandman every couple of years and have it in multiple editions. Denying that would abnegate the responsibility of thinking through a real ethical question.

Jesus Christ. Note the interaction with Gaiman was long before his behavior was made public. But thanks for my first opportunity to engage the BlueSky Block.

No, simply responding to it in a separate skeet. Relax.

And, yes, this might seem like a minor point in the context of the compelling big picture, but any fallacious bit of argumentation is at best a bad look, and at worst will be used by bad actors to discredit the whole thing.

I guarantee anyone doing real NLP work would not disagree with me. One might decide the payoff of a particular cleaning step isn’t worth the effort, but if anyone doing NLP tries to tell you that clean data isn’t a prerequisite for improved model performance, hire someone else.

The “training in what to ignore” stuff happens at a much later stage in the process, and is nowhere near as effective as cleaning the data in the first place. This is an absolutely standard data cleaning approach, zero evidence of nefariousness in and of itself.

I’m not sure whether you’re intending to imply that the removal of copyright notices is itself nefarious or just evidence they knew the material was copyrighted. Removing boilerplate is standard NLP data cleaning so I’m skeptical of the former.

Jesus Christ you people are actually going to make me google “Robbie Williams monkey” aren’t you. I lived in the U.K. for four years a couple decades ago and haven’t thought about that guy since but here I go

ACPRAB

I mean a sense it’s obvious what it means: no new models will be trained directly on the material after three years. But in an important sense the cat is out of the bag at that point.