Search results

Superintelligence is something that arises over a single weekend when a sufficiently advanced AI is instructed to develop itself without limit. It is not something AI makers would make, it is something AI eventually makes of itself, and once able to, it happens fast, and beyond human understanding.

Spam link removed

To me the most important thing is the 128GB Unified RAM version which can easily run 70b and larger LLM's, and so I would not buy other than the 128GB.

For years Netflix has been cancelling all the costly high quality English language shows Holland made Netflix famous for, and has been raising prices on English speakers saying it is for new content, but not saying it is for new foreign language content. Now that Netflix is a foreign language...

This may redefine modern PC architecture. It is exactly what people need to run AI. It is desirable, and may sell like hotcakes.

Unless you are getting 64GB RAM minimum in your next computer you just don't get it.

I still use Winamp 2.9.1 which was the last good version of the original Winamp. I wonder if the code they released is for the original 2.x or the later versions which were just not the same thing. It's important.

Cancelling devices is so Microsoft-like. Who's in charge of devices there up now?

The best DOS I have had is Digital Research DOS 7.3 and later, which included not only multitasking, but also a complete TCIP stack for internet with the DOS based Webspyder browser which would run at high resolution. To this day it is still the best DOS I own, which was free from Caldera. I...

The science behind producing good language models that do not hallucinate is primarily to audit the training data to ensure nothing that is untrue gets into it. GPT-3 was an experiment to see what happens if they just scrape the entire internet into it, seeking emergent capabilities that arise...

I have been testing it out, and I am finding that LLaMA 3 8b is giving some of the worst responses I have ever seen in an SLM, and that LLaMA 2 is incomperably better. This is very sad to see because I had my hopes up.

I have been trying out the Meta LLaMA 3 8b GGUF from Hugging Face (fixed version) in GPT4ALL and it is working perfectly. Very pleased. I have been looking forward to this one as a potential replacement for LLaMA 2 WizardLM v1.2 7b (and/or 13b).

Just finished watching it today. Really enjoyed it. Made me wish Disney would add a like button.

I would have included the announcement that LLaMA 3 7b and 13b are about to be released this month where the full unquantized LLaMA 3 is reportedly on par or better than GPT-4, while a 140b LLaMA 3 is reportedly coming this summer. Now that Microsoft is using quantized versions of GPT-4 for...

If Windows 11 could be installed on older computers and retain local login, then risk management could approve it, and even I might have been able to consider downgrading to it from Windows 10, but Microsoft screwed itself with their pandemic work from home project to piss on everything Myerson...

So, just download models from the original manufacturers?

Just run GPT4All from gpt4all.io ... No new hardware (nor OS) needed.

When a painter uses a brush to paint a picture, the brush is not eligible to hold the copyright, nor is the person who used the tool denied copyright just because they created the picture with a picture creating tool, like when using a brush, or some AI, to do so.

It is really nice to see the authors mentioned GPT4All by Nomic AI. I would just like to clarify for everyone some of the differences between ChatRTX and GPT4All, just to ensure everyone who wants to run AI on any computer understands and knows them. With GPT4All, which is a really small...

Is there some reason people think that you can only use one browser to the exclusion of all others? I use Edge for Copilot and FireFox for browsing the net and have found no issues with using both in parallel. I like FireFox because it has no AI in it, and I am using Edge because it does, and...