COAGULOPATH

7 ms·

31.
by COAGULOPATH 3y ago
(They patched Gemini an hour after I finished writing this. My complaints of excessive refusals and model deception may no longer apply.) Witness: - A chess match between Ultra and GPT4 (the first one ever, as far as I'm aware) - A Gem
32.
by COAGULOPATH 3y ago
This reminds me of Matt Lakeman's "An Attempt at Explaining, Blaming, and Being Very Slightly Sympathetic Toward Enron" https://mattlakeman.org/2020/04/27/explaining-blaming-and-be... He makes
33.
by COAGULOPATH 3y ago
A lot of this stuff was actually cool...but OpenAI got there first. By the time LaMDA came out, we already had ChatGPT. Palm2 was a serious ChatGPT competitor, but by that point, we had GPT4. Imagen was trained in mid 2022 (!) and could do
34.
by COAGULOPATH 3y ago
Remember how there was a hosting company called "50 Megs" because that was a huge amount of storage back in the day.
35.
by COAGULOPATH 3y ago
if you didn't have frames you weren't shit
36.
by COAGULOPATH 3y ago
It still doesn't look ready for primetime to me. When the characters' hands move, they either gain a billion extra fingers or morph into fleshy shapeless clubs. It's just hard to tell because the size of the video samples are
37.
by COAGULOPATH 3y ago
Incredibly good article. He writes about as clearly as you can about such a knotty subject. Related xkcd: https://xkcd.com/1882/
38.
by COAGULOPATH 3y ago
Yes, it has an invisible "shadow prompt" that gets sent to it when you start a session. It looks something like this: https://www.reddit.com/r/ChatGPT/comments/zo9of4/comment/j0n...
39.
by COAGULOPATH 3y ago
As far as I know, EVERY 1t+ LLM is a MoE. Switch-c-2048, Wi Dao 2.0, GLAM, Pangu-Σ, presumably GPT4. Am I missing any?
40.
by COAGULOPATH 3y ago
I searched for "David Bowie discography" and got "This topic contains or implies content that falls outside acceptable use guidelines." Then I searched for "David Bowie". It slowly generated text, section by se
41.
by COAGULOPATH 3y ago
Of course they would. The internet spent a solid month laughing at the Sonic the Hedgehog movie because Sonic had weird-looking teeth.
42.
by COAGULOPATH 3y ago
> they simultaneously feel crippled and limited by their lack of editing / iteration ability. Yeah. They're not "videos" so much as images that move around a bit. This doesn't really look any better than those Mi
43.
by COAGULOPATH 3y ago
> Nothing permanent has been done that can't be undone fairly simply, if all parties agree to undo it. Sam views this as a stab in back. He doesn't want to work with backstabbers. The board has put down too many chips to back o
44.
by COAGULOPATH 3y ago
>AI though promises to be the universal automation, i.e. it can do any job. An AI can't do any job that requires a body, so clearly that isn't true. If you mean mental jobs, this could be true in the future, but right now, it
45.
by COAGULOPATH 3y ago
Has anyone ever hired a hitman WITHOUT it being an undercover agent?
46.
by COAGULOPATH 3y ago
>I had to be very specific and write a few long and detailed prompts for the more complex parts of a the application. This is my experience. You still have to understand programming: you're just typing it out in Natural English.
47.
by COAGULOPATH 3y ago
What's the benefit of doing this vs copying one of the many (far superior) Javascript mp3 players on the internet, such as here? https://freefrontend.com/javascript-music-players/
48.
by COAGULOPATH 3y ago
But it got 10/10 on pre-2021 questions, with the same prompting method...
49.
by COAGULOPATH 3y ago
>It is measuring how well it does _at REPLACING HUMANS_ It's possible to do well on a test and have no ability to do the thing the job tests for. GPT-4 scores well on an advanced sommelier exam, but obviously cannot replace a human
50.
by COAGULOPATH 3y ago
The idea of telling a human generated "the quick brown fox..." from a machine-generated one was always a fantasy. Text has no birthmark. Current LLMs have stylistic quirks imprinted on them by RLHF (ChatGPT's endless "it
51.
by COAGULOPATH 3y ago
It's absolutely the case that LLMs give output that's inconsistent in quality, and the determining factor is your prompt. This is true for base models, fine tuned models, and everything else. Obviously other things matter too, but
52.
by COAGULOPATH 3y ago
And I can't fine tune GPT-4 yet. It's not a case that I'm ignoring it. I literally can't do it.
53.
GPT-4 is not getting worse(coagulopath.com)
149 points by COAGULOPATH 3y ago|174 comments
54.
by COAGULOPATH 3y ago
AoEII has fantastic pathfinding. That and the formation system was the game's secret sauce. In early RTS games like Warcraft, half the game was micromanaging your units from place to place. You'd send your army to a location, and
55.
An in-depth examination of ElsaGate (gross/disturbing kids' videos)(coagulopath.com)
3 points by COAGULOPATH 4y ago|0 comments
56.
by COAGULOPATH 4y ago
Years of reading cryptocurrency stuff has made me skeptical of articles wholly based on speculation about things that COULD happen, with no details on how to get there. "X might do Y someday" is just the gentleman's way of sa
57.
Is ChatGPT killing StackOverflow? An analysis of four months of data(coagulopath.com)
6 points by COAGULOPATH 4y ago|5 comments
58.
by COAGULOPATH 5y ago
I remember when I made a website with frames and felt like a genius.
59.
by COAGULOPATH 5y ago
Much depends on Zion's technology base, which doesn't make any sense to me - its both absurdly high-tech while weirdly primitive, with rough stone buildings, caves, people using burning torches for light, and so on. They're f
60.
by COAGULOPATH 5y ago
What's your level of knowledge about other things in this space? Do you know what a blockchain is, for example?

More ›