loudmax

27 ms·

331.
by loudmax 3y ago
Really wish some manufacturer would copy Apple's strategy of providing long term software support for their phones.
332.
by loudmax 3y ago
> I suppose the problem is that, if you're not a car enthusiast, a $30k Toyota hybrid is still a better deal, even if it drives like shit. I'm absolutely happy with my plug-in Prius. I have a short commute, and it gets
333.
by loudmax 3y ago
This is how I colorize with grep: cat <file> | grep -E --color "<regex>|$" Where `<regex>` is your pattern.
334.
by loudmax 3y ago
I've had the same experience. One of the things I like most about llama.cpp is the relatively straightforward build process, at least when compared to the mess of Python library requirements you run into if you want to experiment with
335.
by loudmax 3y ago
Perhaps the incentive is to reflect the world we live in?
336.
by loudmax 3y ago
Oxide's goal is that they, and by extension their customers, have as much visibility and control over the software stack in these racks as possible, and that includes firmware. They started developing these systems before the current
337.
by loudmax 3y ago
This looks very promising and the sample images are impressive, especially for the small model size. Unfortunately, according to their Readme, "Inference requires at least 23GB of GPU memory." So not something you can run on cons
338.
by loudmax 3y ago
You typically want an Nvidia GPU with at least 8GB of VRAM to get started with this stuff. You can get away with less, but it will be slow going. More VRAM is better. If you're serious about learning how to use these tools, it's
339.
by loudmax 3y ago
I second downpour.com for audiobooks. You can download files in mp3 or m4b formats and listen to them in whatever player you like. I hadn't heard of libro.fm. I'll have to give them a try, thanks for pointing them out.
340.
by loudmax 3y ago
Yes, there's a lot of hype and buzzwords. Certainly companies are slapping the "AI" label onto their products and jumping on the bandwagon in order to gin up interest. But there's a real technological change going on.
341.
by loudmax 3y ago
This is a subjective and semi-informed opinion, but I don't really see that RAG is "a hack". The value of LLMs isn't their ability to regurgitate information. Computers could do that already, with databases, or even fl
342.
by loudmax 3y ago
It would be very surprising. Mistral AI's github page has more information on their sliding window attention method to achieve this performance: https://github.com/mistralai/mistral-src If Mistral 7b lives up to t
343.
by loudmax 3y ago
Agreed. It's great that they're model available under a permissive license. But without more information about the training data and methodology, it isn't exactly "open source".
344.
by loudmax 3y ago
At first glance, I thought this was another Linux distro with a custom window manager, not a new operating system. But it is its own operating system! It's got a custom kernel and everything. Built from scratch, indeed. I'm alw
345.
by loudmax 3y ago
The world's greatest engineering talent won't save you if the management is incompetent. Google may yet experience a comeback, but it will start with throwing out the current C-suite.
346.
by loudmax 3y ago
The `sway` window manager is an attempt to clone i3 for Wayland. If you do have to switch, that part at least, shouldn't be too painful.
347.
by loudmax 3y ago
The Huggingface Leaderboard is mostly dominated by Llama 2 variants: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderb... It depends a lot on what you're trying to do. If have a focused use case o
348.
by loudmax 3y ago
You can run the smaller Llama variants on consumer grade hardware, but people typically rent GPUs from the cloud to run the larger variants. It is possible to run even larger variants on a beefy workstation or gaming rig, but the performan
349.
by loudmax 3y ago
You can get RISC-V based single board computers today. I think the meaningful question is when we'll see RISC-V CPUs that are performance competitive with x86 or ARM CPUs. In that case, probably not in the near future, at least not fo
350.
by loudmax 3y ago
I think FOSS vs proprietary software is sort of orthogonal to what kind restrictions you should probably impose as a parent. It might be more useful to think about what kind of online environments your kids are engaging in. Anything monet
351.
by loudmax 3y ago
You can absolutely run LLMs without a GPU, but you need to set expectations for performance. Some projects to look into are * llama.cpp - https://github.com/ggerganov/llama.cpp * KoboldCpp - https://gith
352.
by loudmax 3y ago
Windows gets hate because it's a de facto monopoly on enterprise computing. Most medium sized business or organizations don't really have much choice but to go with the industry standard. It's natural to feel resentment aga
353.
by loudmax 3y ago
Kind of surprising they didn't go with "Another Brick in the Wall, Part 2" instead. Part 1 is very low key and doesn't feature much of a recognizable melody. Musically, it doesn't really stand on its own, it only
354.
by loudmax 3y ago
"Breakthroughs" are inherently hard to predict. This field is advancing at a very rapid pace. A lot of the improvements are incremental, but even these are coming very fast, and moving in different directions at once. I don&#x27
355.
by loudmax 3y ago
In this context I'd assume SBC means Single Board Computer, such as a Raspberry Pi or one of the many imitators. The article itself mentions running LLaMa on a Pi 4. The interesting implication about running an LLM on a single board c
356.
by loudmax 3y ago
I presume by "National socialists," you mean the Nazis. They weren't particular advocates of the free market, but they weren't socialist in a meaningful economic sense. The Nazis did consider Communists as enemies. Th
357.
by loudmax 3y ago
My understanding of ghost attention is that the interface will inject periodic reminders into the conversation. These are seen by the model but hidden from the user. They do use up some of the tokens available in the context window.
358.
by loudmax 3y ago
My experience of learning Unix from the installation process back in 2001 or was kind of the reverse. I installed Red Hat on an old machine and it the desktop GUI worked out of the box. That wasn't really what I was going for, so I i
359.
by loudmax 3y ago
What actual policies are you opposed to? And how are you ranking these policies? You mentioned states seizing kids if you wont allow children to begin transgender transitioning. What is the actual policy here, and when would it apply? Th
360.
by loudmax 3y ago
I recommend tracking actual policies instead of asinine culture war nonsense on social media. Never mind liking or disliking politicians on a personal level, look at what policy objectives a politician is trying to achieve. Don't tak

More ›