Super, thanks again for taking the time to do so.

I can’t remember if I shared this earlier but I’m jolting down notes on the topic in https://fabien.benetou.fr/Content/SelfHostingArtificialIntelligence so I do also invest time on the topic. Yet my results have also been… subpar so I’m asking as precisely as I can how others actually benefit from it. I’m tired of seeing posts with grand claims that, unlike you, only talk about the happy path in usage. Still, I’m digging not due to skepticism as much as trying to see what can actually be leveraged, not to say salvaged. So yes, genuine feedback like yours is quite precious.,

I do seem to hear from you and others that to kickstart what would be a blank project and get going it can help. Also that for whatever is very recurrent AND popular, like common structures, it can help.

My situation though is in prototyping where documentation is sparse, if even existent, and working examples are very rare. So far it’s been a bust quite often.

Out of curiosity, which AI tools specifically do you use and do you pay for them?

PS: you mention documentation is both cases, so I imagine it’s useful when it’s very structured and when the user can intuit most of how something works, closer to a clearly named API with arguments than explaining the architecture of the project.

Thanks for that, was quite interesting and I agree that completion too early (even… in general) can be distracting.

I did mean about AI though, how you manage to integrate it in your workflow to “automate the boring parts” as I’m curious which parts are “boring” for you and which tools you actual use, and how, to solve the problem. How in particular you are able to estimate if it can be automated with AI, how long it might take, how often you are correct about that bet, how you store and possibly share past attempts to automate, etc.

Mind explaining a bit your workflow at the moment?

You’re right, it’s not “just” Proton. I also tried recently GoG for Pod and… it just worked! From buying the (sigh) Windows game to playing on Linux in literally minutes. Amazing.

For WMR I don’t know unfortunately. Monado does work though and I would check https://lvra.gitlab.io as it’s a great starting point, maybe starting with the Monado SteamVR plugin.

to steam deck.

to SteamOS

So… FWIW I post often about I have a painless NVIDIA experience, including playing Windows only games, including VR games.

I thought “Damn… how did I get so lucky?” and yesterday while tinkering with partitions (as one does…) I decided I’d try a “speed run” to go from no system to a VR Windows only game running on Linux.

I started from Debian 12 600Mb ISO and ~1h later I was playing.

I’m not saying everybody should have a perfect experience playing games on Linux with an NVIDIA but … mine was again pretty straightforward.

I’d argue it’s easier with Ubuntu and accepting non-free repository, probably having the same result, ~1hr from 0 to play, without even using the command line once.

I don’t think I understand your point, are you saying there is no benefit in running locally and that Websites or APIs are more convenient?

improves my experience coding in unfamiliar languages

Alan Perlis said “A programming language that doesn’t change the way you think is not worth learning.”

So… if you code in another language without actually “getting it”, solely having a usable result, what is actually the point of changing languages?

FWIW I did try a lot (LLMs, code, generative AI for images, 3D models) in a lot of ways (CLI, Web based, chat bot) both locally and using APIs.

I don’t use any on a daily basis. I find it exciting that we can theoretically do a lot “more” automatically but… so far the results have not been worth the efforts. Sadly some of the best use cases are exactly what you highlighted, i.e low effort engagement for spam. Overall I find that either working with a professional (script writer, 3D modeler, dev, designer, etc) is a lot more rewarding but also more efficient which itself makes it cheaper.

For use cases where customization helps while quality does matter much due to scale, i.e spam, then LLMs and related tools are amazing.

PS: I’d love to hear the opinion of a spammer actually, maybe they also think it’s not that efficient either.

I like Ollama, and recommend it to tinker, but I admit this “LLM Explorer” is quite neat thanks to sections like “LLMs Fit 16GB VRAM”

Ollama just works but it doesn’t help to pick which model best fits your needs.

Yes I’m talking about DeArrow. Well yes but to be more precise they initially “block” the addon from working for few hours then they let you use it without paying. Slightly different, again I’m not criticizing just highlighting this is not how most add-ons do work.

Interesting that this extension is pay only, first time I see this. Again makes sense to go against a business model of “free” of cost but too expensive for sanity.

I find YouTube itself to be so adversarial that I don’t even use it anymore.

Still, I’m installing both this and SponsorBlock to symbolically show support to this of projects that IMHO show that I want the Web MY way. I don’t want to browse in whatever way maximizes attention and distraction to increase profit margin of surveillance capitalism.

You’re just making another assumption, maybe the dorm has optic fiber with a big bandwidth and a lower latency that most home and business connection. Maybe OP doesn’t care about 120hz and only heat. I don’t think you are getting my point if you are pointing out imperfection about the current technology : it’s possible.

Right, and I mentioned CUDA earlier as one of the reason of their success, so it’s definitely something important. Clients might be interested in e.g Google TPU, startups like Etched, Tenstorrent, Groq, Cerebras Systems or heck even design their own but are probably limited by their current stack relying on CUDA. I imagine though that if backlog do keep on existing there will be abstraction libraries, at least for the most popular ones e.g TensorFlow, JAX or PyTorch, simply because the cost of waiting is too high.

Anyway what I meant isn’t about hardware or software but rather ROI, namely when Goldman Sachs and others issue analyst report saying that the promise itself isn’t up to par with actual usage for paying customers.

I’m also no stockologist and I agree but I that’s not my point. The stock should be high but that might already have been factored in, namely this is not a new situation, so theoretically that’s been priced in since investors have understood it. My point anyway isn’t about the price itself but rather the narrative (or reason, as the example you mention on backlog and lack of competition) that investors themselves believe.

I’m not sure if you played PCVR in the Summer but imagine that in a tiny room… it’s just way too hot. Again I’m NOT saying it’s good, or bad, I’m only saying you made assumption about OP usage. I’m not sure if you tried CloudXR but basically, it works and it’s not that complex to setup (e.g 1h) so it’s relatively faster and cheaper than building and owning a gaming PC.

I don’t understand why you are even arguing about a legitimate usage.

Sure yet it’s a perfectly legitimate one. I’m not OP, it might be exactly their use case.

You do if you are rendering in the cloud, e.g NVIDIA CloudXR. Not sure what OP plans to do.

Not a lawyer but if you have an email that says you can, I’d argue it’s override the ToS assuming the person giving permission actually legally can.

Anyway I bet what they avoid is reselling access so I believe as long as you don’t pay for yourself then resell to others you’ll be OK.