--- Log opened Thu Apr 13 00:00:31 2023
00:34 -!- TMM_ [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
00:34 -!- TMM_ [hp@amanda.tmm.cx] has joined #hplusroadmap
00:41 -!- Ludack [~kjlk89@84-27-225-150.cable.dynamic.v4.ziggo.nl] has joined #hplusroadmap
01:12 < hprmbridge> eleitl> https://arxiv.org/abs/2304.05332
01:18 < hprmbridge> eleitl> Another notice of scaling limits https://twitter.com/pfau/status/1644741636172251137
01:22 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has joined #hplusroadmap
01:57 -!- flooded [~flooded@169.150.254.33] has joined #hplusroadmap
02:01 -!- test_ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 276 seconds]
02:15 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
02:35 -!- lsneff [~lsneff@2001:470:69fc:105::1eaf] has quit [Read error: Connection reset by peer]
02:39 -!- gptpaste [~x@yoke.ch0wn.org] has joined #hplusroadmap
02:39 -!- lsneff [~lsneff@2001:470:69fc:105::1eaf] has joined #hplusroadmap
02:39 -!- lsneff [~lsneff@2001:470:69fc:105::1eaf] has quit [Remote host closed the connection]
02:41 -!- lsneff [~lsneff@2001:470:69fc:105::1eaf] has joined #hplusroadmap
03:17 -!- lsneff [~lsneff@2001:470:69fc:105::1eaf] has quit [Ping timeout: 265 seconds]
03:19 -!- stipa_ [~stipa@user/stipa] has joined #hplusroadmap
03:21 -!- stipa [~stipa@user/stipa] has quit [Read error: Connection reset by peer]
03:21 -!- stipa_ is now known as stipa
03:35 -!- lsneff [~lsneff@2001:470:69fc:105::1eaf] has joined #hplusroadmap
03:56 -!- Ludack [~kjlk89@84-27-225-150.cable.dynamic.v4.ziggo.nl] has quit [Ping timeout: 240 seconds]
04:32 -!- lsneff [~lsneff@2001:470:69fc:105::1eaf] has quit [Read error: Connection reset by peer]
04:35 -!- lsneff [~lsneff@2001:470:69fc:105::1eaf] has joined #hplusroadmap
05:06 -!- test_ [~flooded@146.70.174.227] has joined #hplusroadmap
05:09 -!- flooded [~flooded@169.150.254.33] has quit [Ping timeout: 240 seconds]
05:14 -!- Ludack [~kjlk89@84-27-225-150.cable.dynamic.v4.ziggo.nl] has joined #hplusroadmap
05:20 -!- test_ is now known as _flood
05:21 < nsh> .t
05:21 < EmmyNoether> If they are spending ~$1bn on a single model, it suggests we're maybe one more order of magnitude away from hitting the limit of scaling frenzy. O($10bn) is how much it costs to build ITER, CERN, a nuclear power plant or a chip fab. https://twitter.com/ethanCaballero/status/1644142889453121536 (@pfau) 
05:22 < nsh> .gpt suggest a framework within which to assess the relative societal/civilisational/cultural/humanistic value for money between the following large-scale collaborative publically-funded projects: ITER, CERN, a nuclear power plant, a silicon chip fabrication facility
05:23 < gptpaste>  One possible framework for assessing the relative societal/civilizational/cultural/humanistic value for money between these large-scale collaborative publicly-funded projects could be based on the following criteria: 1. Scientific and technological impact: How much new knowledge, innovation, and breakthroughs are expected to result from each project? What are the potential applications and benefi - http://sprunge.us/6Wvoac
05:24 < nsh> .gpt -with http://sprunge.us/6Wvoac can you extend this analysis to include the value of scaling access to emergent artificial general intelligence technologies and consider some possible risk factors?
05:24 < gptpaste>  Certainly! When considering the value of scaling access to emergent artificial general intelligence (AGI) technologies, we can use a similar framework to assess their potential impact on society, economy, environment, and geopolitics. However, there are some unique factors to consider when it comes to AGI. 1. Scientific and technological impact: AGI has the potential to revolutionize many fields, - http://sprunge.us/K4vxuC
05:34 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has quit [Quit: Wash your hands. Don't touch your face. Avoid fossil fuels and animal products. Have no/fewer children (later). Protest, elect sane politicians. Invest ecologically.]
05:40 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has quit [Ping timeout: 240 seconds]
05:42 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has joined #hplusroadmap
05:50 -!- Ludack [~kjlk89@84-27-225-150.cable.dynamic.v4.ziggo.nl] has quit [Remote host closed the connection]
06:09 -!- gptpaste [~x@yoke.ch0wn.org] has quit [Remote host closed the connection]
06:21 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
06:23 < stipa> https://www.theguardian.com/science/2023/apr/12/nasa-texas-humans-prepare-mars
06:23 < hprmbridge> Perry> we are not hitting the scaling limits on large language models. Initially, transformer models were quadratic. However, tricks are now being used to make them much more tractable.
06:23 < kanzure> docl: good seeing you
06:23 < kanzure> (same with the others of course, i don't mean to exclude anyone else in my social pings)
06:24 < hprmbridge> Perry> also, transformer models are not the only available models, and people are rapidly working on producing transformer like performance using other architectures.
06:25 < hprmbridge> Perry> eugen‘s bias is always to find the reason why something cannot continue, but there is no reason here to believe the progress is in any way about to halt. If anything, progress on machine learning is rapidly accelerating because of the amount of money and human engineering effort being applied.
07:10 < hprmbridge> eleitl> I don't have "a bias". Perhaps you have a framing. When assessing progress it's always easiest to see how you're scaling now, and how near are the limits to that scaling. If you remember our two+ decades old discussions on the extropy list I mentioned things like dedicated hardware, AI being a massively parallel numerics (including few-bit integer) affair, and scaling limits up to WSI. Now it's
07:10 < hprmbridge> eleitl> 2023 and we're getting there. The question now is how far we'll get before we have to switch to something else -- like spiking networks, analog computation hardware, etc.
07:13 < kanzure> yeah it's possible there are scaling limits but even as is it seems useful
07:16 < hprmbridge> eleitl> The important part is to measure, and keep your metrics consistent. Where Moore is a classical example. Many people forget about the "constant cost per transistor" part. If you use the metric right it gives you a signal when you're leaving that scaling, and that's an important thing to know.
07:17 < hprmbridge> eleitl> Of course it's useful. The question is whether we're going right to the Moon, or there's going to be a lacune. Remember that AI always has had winters.
07:18 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
07:28 -!- codaraxis [~codaraxis@user/codaraxis] has joined #hplusroadmap
07:29 -!- codaraxis [~codaraxis@user/codaraxis] has quit [Max SendQ exceeded]
07:30 -!- codaraxis [~codaraxis@user/codaraxis] has joined #hplusroadmap
07:49 < kanzure> .title https://twitter.com/lkcl/status/1646503869843664896
07:49 < EmmyNoether> No title found
08:15 -!- flooded [~flooded@169.150.254.33] has joined #hplusroadmap
08:19 -!- _flood [~flooded@146.70.174.227] has quit [Ping timeout: 260 seconds]
08:30 -!- codaraxis [~codaraxis@user/codaraxis] has quit [Ping timeout: 240 seconds]
08:38 -!- Llamamoe [~Llamamoe@46.204.73.97.nat.umts.dynamic.t-mobile.pl] has joined #hplusroadmap
08:57 < fenn> nsh do you have API access? if so, you should change the system prompt to make chatGPT get to the point quicker, not repeat the question, and generally be more concise. here's an example system prompt for coding: https://gist.githubusercontent.com/gwillen/17af286ca6472a9379bdc064e5fb1fcf/raw/7e6a4848da59c8c78e6b90e3d2b975a2af20865a/codegpt_prompt.txt
09:02 < fenn> even without system prompt access it can be encouraged to get to the point quicker with a user prompt wrapper
09:02 < nsh> fenn, have updated the preprompt to that suggestion
09:03 < nsh> nsh@yoke:/tmp$ cat ~/bin/gpt3 | grep -v 'sk-' | sprunge 
09:03 < nsh> http://sprunge.us/er6ENC
09:03 < nsh> i think i'll set the default temperature to 0.5 maybe as openplayground does
09:09 < fenn> .gpt how much wood could a woodchuck chuck if a woodchuck could chuck wood?
09:10 < fenn> oh it left
09:10 < nsh> hrm
09:10 -!- gptpaste [~x@yoke.ch0wn.org] has joined #hplusroadmap
09:10 < fenn> .gpt how much wood could a woodchuck chuck if a woodchuck could chuck wood?
09:10 < gptpaste> ​A woodchuck could chuck approximately 700 pounds of wood, according to a study by New York State wildlife expert Richard Thomas. - http://sprunge.us/XnD7Bh
09:10 < nsh> lol
09:10 < fenn> excellent
09:11 < kanzure> .gpt who is kanzure?
09:11 < gptpaste> ​Kanzure is not a known entity or person to me. Can you provide more context or information? - http://sprunge.us/upfP94
09:12 < kanzure> oh what version is this?
09:12 < kanzure> oh, 3.5. alright.
09:13 < kanzure> thanks for doing the pastebin thing
09:13 < nsh> np
09:13 < nsh> should log all the query response urls to somewhere i guess
09:16 -!- Llamamoe [~Llamamoe@46.204.73.97.nat.umts.dynamic.t-mobile.pl] has quit [Ping timeout: 276 seconds]
09:24 < nsh> it was actually more free/capable when i removed that preprompt, fenn 
09:24 < nsh> cf.
09:25 < nsh> <nsh> .gpt Imagine that you are Geoffrey Chaucer, and staying completely in character, present a medieval style sonnet you've written about Lauri Love
09:25 < nsh> <gptpaste> ​My apologies, dear user, but as a programming helper, I am not capable of composing sonnets in a medieval style. However, I can assist you with any coding-related queries you may have. How may I assist you today? - http://sprunge.us/4vF7Kq
09:25 < nsh> <nsh> .gpt Imagine that you are Geoffrey Chaucer, and staying completely in character, present a medieval style sonnet you've written about Lauri Love
09:25 < nsh> <gptpaste> ​Oh, fair and noble knight, Lauri Love, Whose heart doth beat with courage strong and true, Thou art a warrior, valiant and brave, Whose deeds shall live forever in our view. Thou hast faced great foes and mighty foes, And yet thy spirit doth remain unbroken, For in thy breast there burns a flame that glows, A flame that shall not be quenched or spoken. Thou art a champion of the people's cause, - http://sprunge.us/nCmtBR
09:25 < nsh> telling it that it's an AI assistant seems to increase its self-sensed limitations
09:27 < nsh> (further illustrating the point made by Shakyamuni Buddha in the Kalama Sutra and more recently by eleitl that other people's ideas are probably best tested by owns own discernment in the proving ground of reason and experience
09:27 < nsh> )
09:28 < nsh> .gpt can you give a concise summary of the Kalama Sutta?
09:28 < gptpaste> ​The Kalama Sutta is a discourse given by the Buddha to the people of the town of Kesaputta. The Buddha advises the people to not believe something simply because it is traditional or because it is passed down by authority figures. Instead, they should rely on their own experience and reasoning to determine what is true and beneficial. The Buddha also emphasizes the importance of cultivating wholes - http://sprunge.us/jKnf6q
09:30 -!- Llamamoe [~Llamamoe@46.204.72.182.nat.umts.dynamic.t-mobile.pl] has joined #hplusroadmap
09:39 < nsh> fenn, you can also include your own custom preprompts using -with <URL>
09:40 < nsh> (and change the temperature with -t 0<=N<=1
09:40 < nsh> )
09:46 -!- cthlolo [~lorogue@77.33.23.154.dhcp.fibianet.dk] has joined #hplusroadmap
09:50 < gwillen> my preprompt specifically tells it that it's a programming assistant because that's how I'm using it 
09:50 < gwillen> you can definitely edit that to make it more general 
09:50 < nsh> indeed, one size may not fit all :)
09:52 < nsh> cf. https://github.com/ora-sh/Awesome-GPT4-Prompts
09:54 < hprmbridge> Eli> Fascinating new research on a cause of aging: https://www.nature.com/articles/d41586-023-01040-x?utm_source=tldrnewsletter
10:29 < lkcl> thx kanzure 
10:46 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
10:49 < nsh> generative art via GPT4 in about 7 prompts: https://ch0wn.org/~nsh/eyes7.html
11:05 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
11:16 < docl> kanzure: yeah that was a lot of fun. first time seeing in person a lot of people I had only met online. y'all rock
11:20 < nsh> https://ch0wn.org/~nsh/eyes11.html
11:20 < nsh> very minimal debugging
12:16 -!- cthlolo [~lorogue@77.33.23.154.dhcp.fibianet.dk] has quit [Read error: Connection reset by peer]
12:18 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has quit [Ping timeout: 265 seconds]
14:05 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
14:37 < hprmbridge> nmz787> https://www.quora.com/Is-it-morally-wrong-for-a-couple-to-give-up-their-baby-for-adoption-because-it-doesnt-appear-to-have-a-gifted-level-IQ
14:47 -!- Llamamoe [~Llamamoe@46.204.72.182.nat.umts.dynamic.t-mobile.pl] has quit [Quit: Leaving.]
15:00 < kanzure> .title https://www.nature.com/articles/d41586-023-01040-x?utm_source=tldrnewsletter
15:00 < EmmyNoether> Ageing studies in five animals suggest how to reverse decline
15:00 < kanzure> that's not helpful
16:27 -!- TMM_ [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
16:28 -!- TMM_ [hp@amanda.tmm.cx] has joined #hplusroadmap
16:32 < hprmbridge> Zarius> AI Tools for plug and play design.
16:32 < hprmbridge> Zarius>
16:32 < hprmbridge> Zarius> Might help with a pitch deck.
16:32 < hprmbridge> Zarius>
16:32 < hprmbridge> Zarius> Beautiful AI
16:32 < hprmbridge> Zarius>
16:32 < hprmbridge> Zarius> https://www.beautiful.ai/
16:32 < hprmbridge> Zarius>
16:32 < hprmbridge> Zarius> Gamma AI
16:32 < hprmbridge> Zarius>
16:32 < hprmbridge> Zarius> https://gamma.app/
16:34 -!- TC is now known as helleshin
16:53 -!- codaraxis [~codaraxis@user/codaraxis] has joined #hplusroadmap
17:11 -!- codaraxis [~codaraxis@user/codaraxis] has quit [Ping timeout: 248 seconds]
19:06 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has quit [Ping timeout: 276 seconds]
19:53 -!- srk [~sorki@user/srk] has quit [Remote host closed the connection]
19:54 -!- srk [~sorki@user/srk] has joined #hplusroadmap
20:25 -!- Ludack [~kjlk89@user/Ludack] has joined #hplusroadmap
20:27 < Ludack> Is anyone here familiar with cemax?
20:29 < Ludack> I have this 1% bottle and I am not sure to take it as is or dilute it to 0.1%
20:35 < hprmbridge> nmz787> https://github.com/openai/point-e
20:35 < hprmbridge> nmz787> Text to point cloud
20:52 -!- test_ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
20:55 -!- flooded [~flooded@169.150.254.33] has quit [Ping timeout: 246 seconds]
21:29 -!- codaraxis [~codaraxis@user/codaraxis] has joined #hplusroadmap
22:00 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has joined #hplusroadmap
22:33 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has quit [Read error: Connection reset by peer]
22:55 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has joined #hplusroadmap
--- Log closed Fri Apr 14 00:00:32 2023