It seems it's because OP is generating the whole screen every frame / every move. Of course that will give inconsistent results.

I wonder if this approach would work better:

1. generate the whole screen once

2. on update, create a mask for all changed elements of the underlying data

3. do an inpainting pass with this mask, with regional prompting to specify which parts have changed how

4. when moving the camera, do outpainting

This might not be possible with cloud based solutions, but I can see it being possible locally.

So something that is not a thing yet, and is being actively campaigned against?

Yes, we normally act when the vehicle speeding towards us is seen far away already.

Most member states are for it.

This prompted me to check how long ago I was 13 years old.

It was 26 years ago.

Are we old yet?

"Old" is a state of mind. If you find yourself believing that everything new is terrible and you want to go back to "the good old days," then yeah, you're old. If you're still open to learning new things and adapting to and exploring the world as it changes around you, then you're not old yet.

I remember getting this bug on accident when I was a kid. I was building a tunnel, but clicked the wrong tile. It popped up a huuuge sum of money and I got scared that I messed up and I'll go bankrupt, only to realize I now had 2B dollars...

> using iPhones for 20+ years

You probably meant Apple devices in general? The iPhone is not yet 20 years old.

Sorry, that's right. For the iPhone probably about 15 years (ie, since not long after it first came out). Edited my previous comment.

Skateboarding taught me the power of perserverance, and muscle memory. That and Quake 3: Defrag trickjumping.

Llama2 was not released with 30B parameters, or was it?

While the llama2-34b base model hasn't been released, CodeLlama2 is effectively a fine-tuned version of 34b and there are some people working with that.

As Ollama uses a llama.cpp fork on the backend, I'd expect its memory usage to be very similar to that.

Oh nope you are 100% correct, I was thinking of the first llama. My buddy is running the 70B llama 2 on two 3090s and the 30B llama 1 on one 3090.

There's also https://en.wikipedia.org/wiki/Enceladus so it's not just one random moon.

Sticking with just the Jovian moons, both Ganymede and Callisto are thought to have subsurface oceans too.

Ganymede's oceans might be the largest in the entire Solar System.

1. https://en.wikipedia.org/wiki/Ganymede_(moon)#Subsurface_oce...

2. https://en.wikipedia.org/wiki/Callisto_(moon)#Internal_struc...

800km deep. Oof.

Seems like an ocean world could develop intelligence but technology would be quite difficult.

Our technology would be quite difficult, we might not even be able imagine what form an alien race's might take.

Technology is technology, it's about exploiting the laws of nature to do things.

If you want to create refined metals you'll need to heat it, you'll need a source of heat, and you'll need to isolate it from the surrounding water long enough to finish the process.

Being underwater makes a lot things much much more difficult or impossible.

That's even ignoring your species won't be incidentally exposed to sources of easy energy like fire occurring naturally before learning to create it yourself.

There's certainly workarounds and alternatives that could make it _possible_ to develop technologies like fire, metallurgy, complex chemistry, etc. but they'll all be so much harder to discover and do that no early civilisation would consider the required experimentation worthwhile.

Again, you're thinking human centric. Maybe they could do amazing things with non metallic materials which we haven't needed or which haven't been practical for us due to the different circumstances.

there are some pretty cool scifi books that look at how an aquatic species might develop technology. the "ringworld's children" involves a species that develops on a frozen world with an ocean beneath the ice.

Is watching a twitch stream a social activity? I guess you can participate in the chat, but we know what twitch chat looks like...

I've been playing games on Linux daily for years now (even AAA games like Cyberpunk 2077) and this is why I am still holding out on VR. I just don't want to be forced to go back to Windows just to play the games I want to play.

I'm waiting for Valve to maybe make a successor to Index with proper Linux support, but I'm not holding my breath.

Have you ever compared in-game performances between Windows and Linux?