lobofta

Try NixOs. Somewhat of a steep learning curve, but once you're over the hump setting up a machine is about half an hour of work and most of that is just waiting for stuff to download. Set up a new machine the other day and it immediately became my daily driver as everything works nearly exactly as on my other machines.

In my case the possible hallucination was the only one that was 100% factual.

Vegan ML engineer here. In total agreement with you. People are just moving the goal post to keep themselves protected from the obvious conclusion: there is nothing really all that special about us humans. Perhaps subjective experience is simply the internal state of a self supervised continuous learning algorithm and we don't like that conclusion very much.

I use Niri workspaces that way. I name my spaces (usually after branches) and have a browser, editor and usually a few terminals open on a workspace. It's also great that a workspace has infinite space so that I can never have to think about creating workspaces just because some workspace has run out of room.

Right, not having to leave the terminal is a big one for me too. I live there and every time I start some desktop app, I gotta switch gears, potentially grab the mouse, leave my vim keybinds behind and leave my font and color scheme behind. I just feel more productive on the terminal then on desktop apps.

I think also a big problem of desktop apps is that you have to deal with window management. Now that I am on Niri it is really apparent to me how much I hated juggling windows in pretty much any other window manager that I ever used, except, interestingly enough, tmux.

Almost 40 years of software development and I think you're overlooking the strength of Gradle. Before Gradle if you wanted any kind of build step that was slightly outside of the norm you had to roll your own maven plugin or god forbid try to script something with ant in xml. Using a DSL in a scripting language was a smart move.

What I hated most about Gradle is that groovy is untyped. You'd make a change and wait for your build to complete only to find out minutes later that you made a typo or some other innocuous mistake. Glad they introduced Kotlin DSL.

I didn't notice the breaking changes much when we heavily used gradle. Gradle also comes with gradlew, which bootstraps your gradle project with the exact gradle version that is needed to build that project, so you can take an old gradle project and build it regardless of all those breaking changes that you mention.

A bit shocking to see how low people rate factory farming, place 34. Arguably the worst thing happening on this planet right now, the only thing is: not to humans, but to other sentient beings.

Why does it need to be the average? It seems to me more like it models the manifold of human knowledge. However we often query for the average, because that is often good enough and gives us quick results, but there is nothing fundamentally preventing us from sending AI into the deep end of under-explored territory and perhaps coming back with something new. It is ultimately the exploration vs exploitation trade off.

Of course it's possible.

I don't say this, because I know how, but because I see no reason why we will be unable to crack that problem. If our brains can do it, so will AI one day.

Thanks. It is too bad so few people think of fish as the sentient beings that they are.

Arm the teachers!

For sure that is possible.

I don't know how OP manages their secrets, but I am running NixOs and am letting 1Password manage all my secrets. 1Password can manage SSH agents, can inject environment variables and manage passwords/keys in the browser. All I need to do when I setup a new machine with NixOS is connect my 1Password to its account manually, after that it's all automated.

The sad thing about this read is that none of the criticisms are necessarily leveled as Nix as a system, but are purely about how the ecosystem is managed.

Might we ever distinguish what is complex and complicated? Probably not, but I guess the author argues that this gives us a way forward because we can try to distill large models.

Would love a GPU benchmark too especially for training and inference workloads