Conversation started Jun 4, 2021 at 2:32.
Jun 4, 2021 02:32
CMQ: What is a recent conversation you want to see auto-completed by gtp3?
Try the start of the hivemind
This one's a short one with lots of repetition if you just want to test something
And this one's a longer one with lots of terminology GPT probably won't know
maybe it can finally tell us whether or not it was funny :P
What is gpt3?
> Generative Pre-trained Transformer 3 is an autoregressive language model that uses deep learning to produce human-like text. It is the third-generation language prediction model in the GPT-n series created by OpenAI, a San Francisco-based artificial intelligence research laboratory.
It (or something similar) powers a lot of the things like code autocompletion
is this the thing people use to extend music past where it actually stops, or is that another algorithm/model
Jun 4, 2021 02:37
@RedwolfPrograms I just gotta make it into a nicer format first (the original conversation)
@hyper-neutrino Probably not GPT
It's meant for text
right lol
The extent of my knowledge on this stuff is some Robert Miles and Tom Scott videos though, so don't trust what I say 100% :p
Okay right off the bat I can tell y'all are gonna like the autocompletion
I'm excited
Jun 4, 2021 02:48
gist.github.com/Lyxal/da8877ebed81e370ca476c5d243e3a6b (disclaimer: the dialogue I linked in the two gists contain text generated by OpenAI. After the indicated line, the text present is not at all representative of the respective users)
Lyxal: @user how do you feel about being a tool of the redwolf program?
user: I'm not sure.
user: But I'm already used to being a tool of the FDIC.
Wow!
I just want to be able to pay for my teeth.
> user: But I'm already used to being a tool of the FDIC.
> Lyxal: @user well you'll be happy to know that you are a member of the FDIC
> user: The what?
(sorry for the ping, I realized after pasting that I think that still pings you)
@Ausername that line was actually said though
Wait what
Jun 4, 2021 02:50
--- everything after this is Ai generated
L40 is where the real convo and AI gen split
Line 40
ninja'd
> user: Is this conversation being recorded?
Lyxal: @/user yes
Lyxal: @/user by the redwolf program
the AI knows too much
I told you guys this was a bad idea. We're all going to die now. When it takes over we're all going to die.
The hivemind is the AI's biggest enemy.
We need to prepare to fight it, and the more technology we rely on, the faster it gets us.
Throws laptop out window
What does the redwolf program do? Does it rickroll?
Jun 4, 2021 02:52
That snippet cost $0.27 BTW. Luckily I've got $18 in free credits
I want to see if it's funny now :p
Jun 4, 2021 03:14
Is it funny? I can't tell, I'm 5.8k.
there's a screenshot right after the bookmark
Jun 4, 2021 03:29
Specifically, it seems there's good support for this, but as a normal feed or only as a ticker feed?
Unfortunately I don't think that post will be a good way to decide, since there's a pretty big time difference between the two proposals
Wait it's five days
0
A: Sandbox for Proposed Challenges

EasyasPiWrite a C++ demangler Ever look at an objdump and see gibberish like this? _ZN3foo3bar3bazIRKPKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEEE3barISt6vectorIP8_IO_FILESaISG_EEEEvSB_PT_RSo This is a C++ mangled symbol (with the Itanium ABI) of the following beautiful function: void foo::bar...

But still, the other one had gathered quite a few votes before the second was posted :p
Yeah. We could repost and dupe hammer the old one into it but like, idk. i'd just like some feedback, even downvotes are better than nothing :p
i've prompted at least a couple times here (and I think specifically pointed to your answer to get any votes on it) and nothing's changed since like posting day
I support a normal feed, since a ticker feed would be a bit inconsistent
And honestly a bit more annoying IMO
Since we can't just 11 the annoying ones
Jun 4, 2021 03:38
Well, why not try both? :P
yeah. and the other point i think i mentioned at some point was that an interesting hello world will probably prompt discussion on it regardless and having no record as to what people are talking about or why makes context worse
a single onebox is much less space than the confusion from "wait why did people suddenly start talking about some weird language with literally no prompt"
i like the idea of a single bold link better because i logread
that'd probably need to be done by a bot then
if redwolf is willing that is an option to implement :p
Problem is, that adds another bot to the sidebar
An RSS feed might work, but I'm not sure how well they work without oneboxes
well this could go in new posts potentially
Jun 4, 2021 03:43
A bit too much overloading on one bot though, imo
It would work if HN 11s every message to un-onebox it :P
I guess just a normal feed with onebox is a good compromise then
it would make sense for something to post like
Oh, I like that format
Unrelated, I've set up a log of TNB through NP for the next few days
CMQ: User whose name is the most accurate.
but parsing the language and its attached link might be challenging
@Ausername yours since your username is, indeed, a username, or user because user is a user
Jun 4, 2021 03:46
@hyper-neutrino Luckily I have spent way too much time developing a method to fairly reliably do just that :p
@RedwolfPrograms 2021 TNB bot incident
SimplyBeautifulArt maybe?
@RedwolfPrograms omg redwolf surveillance this is literally 1984
I so want to log in as NP and post "Big brother is always watching" then leave
it's strange (for me) that we have three bots operating in this room now. bots (that post) were once strictly forbidden in any regard :p
Jun 4, 2021 03:48
You're the third I presume?
OSP: am I a joke to you
(You mean four right?)
(not counting me)
I'm probably going to rewrite NP/SP, the code and logging is getting insanely messy
the thing about logging is I can't nuke things out of logs, but that's not really much of a problem (or at least not something that would make logging disallowed) because I can't nuke messages out of (most) people's minds anyway if they've already seen it, and one can't really detect logging anyway
Jun 4, 2021 03:51
I'm thinking I'll finish my open source library, then use that instead of having bits and pieces of it strewn though the current program
@hyper-neutrino I'll upgrade it soon to use the socket to detect when a message is deleted and remove it from the logs
I can see the history anyway, and I'm probably the only person who'll be using this, but it's probably a good idea anyway
the main thing is just so long as you are not reposting/spreading nuked content it's w/e :P
any malicious thing you could do with deleted content would be malicious and bad regardless of whether it was deleted anyway, if that makes sense
but yeah if you have auto deletion that would be very cool
also the thing is yes you can see the history but that's why i will fully redact and purge sensitive things that i delete
(which is quite rare, fortunately)
Oh, I didn't know you could do that. Good point, I wasn't even thinking about accidental (or malicious) sensitive information, just the "typical" spam and offensive stuff.
Ah. Yeah, there's a button to purge the revision history of a message
like if you look at ^ it's been edited once but there's no message history
Interestingly, it still tells me it's been edited once (but not the history). (Might be caching though).
Unfortunately, a deleted message will show the last revision, and you can't edit a deleted message, so if you accidentally delete and forget to redact the message itself, you can no longer redact or purge its content.
So doing things in the wrong order will actually block you from redacting the content permanently.
@RedwolfPrograms I think that's expected behavior. I mean, it has been edited once, the important thing is just that you can't tell why
I edited it again, and it says "edited twice", and if you look at the history, it shows one revision and then "older data not available"
one thing I don't like about the purge history button is it doesn't come with confirmation, it just immediately applies
and unlike suspensions, it's not undoable (also suspensions aren't really either because they go on your permanent record)
Jun 4, 2021 04:15
Could someone format the "funny" conversation into name: line for me?
Sure just a sec
It makes the Ai process way easier
Dang, ninja'd
I automated it though, I'll put the code here
CMP: would y'all rather have Elendia's TNBDE (SQL) or an API endpoint you can get TNB messages and whatever else from?
Jun 4, 2021 04:19
var contents = [...document.querySelectorAll(".content")];

console.log(contents.map(c => c.parentNode.parentNode.parentNode.children[0].children[0].children[1].children[0].title + ": " + c.textContent.trim()).join("\n"));
i just copy pasted the content and typed all of the names manually
@hyper-neutrino Api
@lyxal How's it going?
gist.github.com/Lyxal/d0666ff7f02f35525632e7b2a506d32a (disclaimer: the dialogue I linked in the two gists contain text generated by OpenAI. After the indicated line, the text present is not at all representative of the respective users)
Redwolf is very insistent that they're not a joke.
We still don't have an answer lol
Jun 4, 2021 04:22
> hyper-neutrino: it's not funny enough to be funny
> Redwolf Programs: Well, it's not really funny, but it's funny in some other way
> user: Note to self: I have no idea what's going on
user: Note to self: I'm
Redwolf Programs: You're not
pretty accurate
User's trying remember that they are funny
Jun 4, 2021 04:23
GPT gives up near the end lol
HN becomes quite insistent they are not a joke though
legend has it that user is still noting to self to this day
@RedwolfPrograms well, I'm not, I'm hyper-neutrino
Is it bizzare how saying sentences backwards creates backwards sentences saying how bizarre it is?
@lyxal when the
Just to be extremely very clear: the dialogue I linked in the two gists contain text generated by OpenAI. After the indicated line, the text present is not at all representative of the respective users
As soon as you observe whether it is funny or not, it becomes the opposite of what you want it to be.
Jun 4, 2021 04:28
this is a pretty accurate reflection of CGCC. get access to a very good (top-quality?) AI and immediately use it to extend absolutely pointless funny conversations in TNB :P
@lyxal But it is :p
Answer: no. You can never have enough paperclips.
@RedwolfPrograms I'm required to say that it isn't
do you want me to 11 that disclaimer into both of the gist link messages?
Jun 4, 2021 04:38
Sure
Also, I'm only allowed 4 posts per day of this kind
So make your suggestions count
how much did the funny conversation cost :P
Less than 1 cenr
*cent
And we didn't even get to find out if it was funny D:
I made sure to use the cheaper model :p
What a waste of an insignificant amount of money
:p
Jun 4, 2021 04:40
lol
The top model would probably have given the same results anyway
it's not exactly a complex conversation :P
Were any of y'all offended by the two gists?
I wasn't in them so no
Asking because not offending people is part of the toc I agreed to
Jun 4, 2021 04:43
Of course not, dw :p
I was not offended :P your fridge is safe for now
3
@lyxal The AI would be offensive, not you...
I personally don't care if y'all are offended, it's that I don't want to risk my beta status :p
bruh moment
Anyone want a third conversation done?
Jun 4, 2021 04:47
Sure, try the war :p
Might be a bit expensive though so it's your choice
(Not that it isn't anyway :p)
The expense comes from the generated text
Not the prompt
feed it the entirety of TNB then :P /s
Give me TNB in the nice format and I will :p
might take a minute or two :P
I'm on mobile btw
Jun 4, 2021 04:51
i'm sorry for your loss
That's why I need y'all to do the formatting
I can convert the war if you want
Please do
Link it in a gist
oh i didn't send much. might not get anything autocompleted under my name :P
Jun 4, 2021 04:53
Dang
Max 1k word prompt
Oh well
Give me a suitable cut off and I'll try again
Slice it off
Jun 4, 2021 04:56
I suppose to assist in lyxal wanting to protect their access and TOS agreement, if anyone is ever offended by any content produced by the AI posted here, feel free to ping me so I can airstrike it if lyxal can't anymore :P
Oh
Oh this is good
Y'all are gonna love it
wowoowowwwhgasd
Give me a few more moments to add a few more bits on
A suitable point from the start would be user: we can f*ck with their brains
Wow its actually juicy
Jun 4, 2021 04:57
link plz?
@Ausername be patient :p
Trying
I'll go get some paperclips
Oh
Nvm
I added too much
Made my phone crash
Jun 4, 2021 04:58
D:
Couldn't handle all the text
Hopefully you can still access it without paying more?
I'll try again
 
Conversation ended Jun 4, 2021 at 4:59.