There's a big difference between runaway costs and these costs over here, which are 10-20% higher than we think makes sense, especially compared to what we're spending over here. Let's spend some time figuring out how to reduce those costs.
You should be doing both - belt and suspenders.
I'd argue that a models ability to ignore/manage/sift through the noise added to the training set from other LLMs increases in importance and value as time goes on.
It's all technical test footage used to test their media pipelines – presumably, they're sharing it to create industry standards, particularly for partner and open-source library implementations.
Another even more modern example is The Martian. Weir published it chapter by chapter on his website, even updating previous chapters based on (mostly technical) feedback from his readers. Once completed, his readers encouraged him to publish an eBook, it took off on Amazon, and the rest is history.
Is there a canonical definition of hacker as described in this post and (at it's best) this community? I always feel strong gravitation to this term, but also feel nervous about embracing it because of it's criminal connotations in common parlance.
Postgres has a module called unaccent[0] that removes diacritics for filtering. I expect your browser is doing something similar. While not appropriate when looking for exact matches, when doing user-input based searches, this should probably be the norm, as the user may be unaware of the accents or how to input them correctly on their keyboards.
Dove deep on this years ago when implementing a filter for wines and wine regions.