
Eufloria 2?!
That’s a name I haven’t heard in a long time, was very fun when.
Autocorrect hates me, I am sorry.

Eufloria 2?!
That’s a name I haven’t heard in a long time, was very fun when.

It’s not trained for this, it’s just Gemini and maybe a fine tune, probably running on Google adk or lang chain.
It’s a Google partnership, so that’s what they’d be using.

Why did they skip the 255 versions?!
/s

I’ve always chafed with the rails design philosophies.
I’m fine with the idea of a standard set of tools used across projects (if you’re in the ecosystem), but I hate the centralization of thought and implicit magic’s that RoR embraces.
I’m very much an explicit is better than implicit person.

I would have thought they’d prefix the hash with “sha256-“ or something, or even just 256 prefix the hash.

The killer app is “you can’t build your own cloud because we’re hoarding all the compute”

This was several years ago, context windows were 4k, so the whole thing was unstable.
I should see if that repo still exists and modernize it though.

Before tool use was an official thing, I built a harness for LLMs to generate XML based tool calls, then I had them run in a reality tv game where they had to murder each other…
It was fun, but none of the models were tool trained and so it didn’t work.

It’s not so cut and dry
Apparently someone used it on a clean machine and it still read all their messages and did this kind of shit.
You should have expected nothing less from meta.
This looks like Pokemon: Breath of the Wild
Nice art direction, I’m unclear on gameplay

This is how you make your CISO cry.

Unsure how I feel about this
I don’t like steams discovery system. It’s not good, and the numbers show that most devs don’t get any exposure.
Steam has a Pareto skew problem: the top games get all the attention, and the long tail of games get none.
Steam also has a large undiagnosed observer bias inspection bias problem, where there’s a large contingent of people who spend a lot of time on games but their recommendations are absolutely god awful for normal players. But they make up all the metrics (play time, games played, achievements, etc), so in most algorithms they’ll be over represented. This effect gets stronger the more niche a game is.
Steams user tags are also kinda awful. They’re a total kitchen sink: I played Crimson Desert, now steam recommends the “dragon” tag for me. Well, the top “dragon” games are not at all like an action game that happens to have dragons. Same with “horses” tag. This is somewhere that better tag weighting could help, but also the combination of tags is hugely important.
Sometimes I don’t open steam for more than a year, my recommendations in discovery queue are so far removed from what I want it’s insane.
Is there a steam tag database? Would be fun to play with their data.
Edit: inspection bias not observer bias

Maybe they should have the lower resource abusing version on all the time?

The only thing is this doesn’t yet do online user directories, local users only
Which is a big constraint right now

This wouldn’t happen if he wasn’t intimately involved with, a frequent visitor of, and best friends with the largest pedophile network in modern history.
Two of my favourite actors, I bet they’re chill in real life.

This is almost definitely intentional
I think they’re building hacking models for the US government and masking as this when caught.

Isn’t this their third attempt?
I’m pretty behind what I want for drives, and modern dev practices of storing tons of intermediate container build artifacts, and a resurgence of arrrborists will probably soak up drive demand even if prices fall.