zone411

15 ms·

151.
by zone411 3y ago
All creditors are expected to be repaid in full, though based on Nov 2022 crypto prices.
152.
by zone411 3y ago
About Minsky being too optimistic: "I would tell Marvin this whole AI thing is just horrible. Why are we doing it? And he would say it will be effective for getting our grants, so shut up and just play along and in fact, it was true. I
153.
by zone411 3y ago
GPT-4 Turbo: 31.0 Claude 3 Opus: 27.3 Mistral Large: 17.7 Mistral Medium: 15.3 Gemini Pro 1.0: 14.2 Qwen 1.5 72B Chat: 10.7 Claude 3 Sonnet: 7.6 GPT-3.5 Turbo: 4.2 Mixtral 8x7B Instruct: 4.2 Llama 2 70B Chat: 3.5 Nous Hermes 2 Yi 34B: 1.5 T
154.
by zone411 3y ago
I'd expect more configuration issues in getting it to run on them than from a tested llama.cpp version, since this doesn't seem like a polished release. But maybe.
155.
by zone411 3y ago
If you're just looking to test it out, it's probably easiest to wait for llama.cpp to add support ( https://github.com/ggerganov/llama.cpp/issues/6120 ), and then you can run it slowly if you have eno
156.
by zone411 3y ago
It's actually not the largest. https://huggingface.co/google/switch-c-2048 is 1.6T parameters.
157.
by zone411 3y ago
Einops and safetensors are not niche! They are just more technical and well-known by people who do more than GPT wrappers starting in 2022 ;)
158.
by zone411 3y ago
A simplistic economic model that overlooks hundreds of important factors may provide a basic Econ 101 understanding but it does not reflect how the world truly operates and proves nothing.
159.
by zone411 3y ago
Meta is still playing catch-up. Might be hard to believe but according to Reuters they've been trying to run AI workloads mostly on CPUs until 2022 and they had to pull the plug on the first iteration of their AI chip. https:/&#x
160.
by zone411 3y ago
Yi 34B Chat has not done well on my new NYT Connections benchmark and it's only in the 22nd place on the LMSYS Elo-based leaderboard (151 Elo below GPT 4 Turbo). It's doing better in Chinese. When it comes to models with open-sour
161.
by zone411 3y ago
They really should have an API available. It's difficult to get a good idea of the quality of their model otherwise. For example, I just created a benchmark based on NYT Connections, and I don't mind paying for access and I'd
162.
by zone411 3y ago
I've just created a new benchmark to see how top LLMs do on NYT Connections ( https://www.nytimes.com/games/connections ). 267 puzzles, 3 prompts for each, uppercase and lowercase. GPT-4 Turbo: 31.0 Claude 3 Opus: 2
163.
by zone411 3y ago
> Bit of an aside, but that "low-high" context scale with the US proudly at the "tell it like it is" extreme left instantly reduces the credibility of any article where I see this. Read further. Low context is not &qu
164.
by zone411 3y ago
That's not how it works. It's a major independent law firm and there is no chance they'd risk their reputation to cover for FRHC. If there are problems with how FRHC operates (I have no idea), they'd just ask the law fir
165.
by zone411 3y ago
It actually gained a lot of publicity because Hindenburg has a pretty good track record and many people are paying attention to them. There is zero chance financial regulators didn't see it. FRHC got a law firm to review some of their
166.
by zone411 3y ago
The relationship between processing speed and IQ is not so simple. E.g. https://neurosciencenews.com/iq-decision-speed-23377/ "Researchers discovered that people with higher IQs are quicker when solving simple ta
167.
by zone411 3y ago
You can support this error? What? If you're criticizing someone for errors or lack of research, you should strive for more authoritative references than Wikipedia pages.
168.
by zone411 3y ago
He co-authored the "Grabby Aliens" paper that made multiple rounds around HN among other things.
169.
by zone411 3y ago
I got these from a custom model and they didn't seem to be in Google: Yo mama so fat... ...when she breaks a plate it's considered a strategic military strike ...when she sucked in her stomach there was a breeze ...when she fell i
170.
by zone411 3y ago
Get a 43'' center monitor with two LG DualUP 16:18 ratio monitors on the sides. Cheaper and much much better than Vision Pro. You will need a large desk though.
171.
by zone411 3y ago
Video will be especially important for language models to grasp physical actions that are instinctive and obvious to humans but not explicitly detailed in text or video captions. I mentioned this in 2022: https://twitter.com&#x2F
172.
by zone411 3y ago
Yes. Since working on my AI melodies project ( https://www.melodies.ai/ ) two years ago, I've been saying that producing a high-quality, finalized song from text won't be feasible or even desirable for a while, and
173.
by zone411 3y ago
It's about future profitability, though. That's speculative, of course, but that's why the growth rate matters.
174.
by zone411 3y ago
There is a well-known male advantage in spatial object rotation. https://books.google.com/books?id=4s5WCgAAQBAJ&pg=PA356#v=on...
175.
by zone411 3y ago
My initial impression is that Gemini Advanced has more difficulty understanding prompts and following directions than GPT-4. It surprisingly often changes meaning when rewriting things. However, it can be more creative and specific, while G
176.
by zone411 3y ago
And that's exactly why regulators must compel them to release the footage immediately if they wish to continue operating on public roads.
177.
by zone411 3y ago
So what? People can argue about many things, that's not a valid reason against it.
178.
by zone411 3y ago
Good, they should be forced to set a precedent.
179.
by zone411 3y ago
He probably watched one of Apple's immersive videos (I think it was "Highlining") where for some idiotic reason there is some panning.
180.
by zone411 3y ago
We shouldn't have to wait and listen to the spin from both sites. The video should be released immediately. What's Waymo's excuse?

More ›