yano

Mavis Beacon, Windows 3.1, yeah somehow i never liked that program much. Though how you see things in hindsight is very different today. Also building a complex skill takes thousands of hours before it really sticks and becomes second nature.

Though could be worse.... like kids hen-pecking on phones using their thumbs.

Other variations, Mario Teaches Typing, and then the 8bit games like word defenders, or dreamcast's 'Typing of the Dead' (arcade port using typing instead of fake guns, funny as hell).

Either way, if it's not engaging or fun, or feels preachy people will refuse it. Gotta Mary Poppins it: Find the fun, and SNAP, the job's a game.

I learned more typing from TTT (Touch Type Tutor) which was a Dos program back in the day. But naturally that's not how it went in school, in school they wanted you to memorize the location of letters on the keyboard and covered your hands with a length of cardboard so you couldn't see it rather than learning naturally...

But that's where 'type tutor' phrase i associate came from.

Typing tutor, but in reverse 😛

No it really sounds good. Would likely be great for home-schooling as well.

I saw QAT stuff going on for a week or two. I wonder if finetunes going forward will all use QAT as a base?

Or will they do Loras?

I'd wondered for a bit why i see some Lora's where that are almost as big as the base model, vs say SDXL where a Lora is 200Mb vs a 6Gb base.

Either way, I'll be keeping eye eyes peeled for interesting releases everywhere.

I mean, like, we are talking about a 50M model here.

On models smaller than say 256Mb there's no reason to use anything but full precision.

Though with Prism-ml/Bonsai showing you can get stuff done on 1 and 1.58bit, we may see some very small things in the near future... Just hope nothing that would bork at important times.

I've wondered, if a sufficiently reasoned model can get away with missing a lot of data because it can reason it out, and thus the smaller size may be worth it to be faster/leaner (even if it needs more tokens to reason the solution).