--- Log opened Wed Mar 05 00:00:35 2025
00:16 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap
00:22 < hprmbridge> kanzure> have you tried not sleeping. frees up a bunch of time.
00:25 < fenn> i spend a lot of time not sleeping, but it's often when i'm trying to sleep :)
00:25 < fenn> also this strategy doesn't help with getting knowledge into my head
00:30 < hprmbridge> kanzure> arnold says "sleep faster'
00:37 < fenn> what does danny devito say
00:42 < hprmbridge> kanzure> 8x8 electrodes is not enough
01:09 < fenn> i'm skeptical of the electrode approach, but optical sensing also has tradeoffs between spatial resolution, temporal resolution, and cost
01:09 < hprmbridge> kanzure> has anyone tried just adding a retina or something
01:09 < fenn> ideally you'd grow the organoid around micro scale fiber optics or insulated electrodes
01:10 < fenn> the cost is the cameras and projectors and stuff, although the genetic engineering adds a one-time r&d cost
01:14 < fenn> you probably do want special photoreceptor cells for high bandwidth input
01:15 < fenn> i'm still kinda fuzzy on how training is supposed to work for this
01:16 < fenn> they(?) got it to play pong by randomizing the input after the ball falls off the screen as a sort of "punishment"
01:18 < fenn> i guess you can just circuitbend the glorified autocomplete mode but it seems risky
01:18 < hprmbridge> kanzure> you don't even need the neurons to play pong https://diyhpl.us/~bryan/papers2/ai/physical-intelligence/Electro-active%20polymer%20hydrogels%20exhibit%20emergent%20memory%20when%20embodied%20in%20a%20simulated%20game%20environment%20-%202024.pdf
01:19 < fenn> it's kind of infuriating that this took until 2024 for no particular reason
01:19 < hprmbridge> kanzure> maybe it's not a useful product and that's why it hadn't
01:19 -!- Hoolootwo [~Hooloovoo@hax0rbana.org] has joined #hplusroadmap
01:20 < hprmbridge> kanzure> finalspark also has API access to neuron blobs btw
01:20 -!- Hooloovoo [~Hooloovoo@hax0rbana.org] has quit [Ping timeout: 272 seconds]
01:53 -!- WizJin [~Wizzy@user/WizJin] has joined #hplusroadmap
06:31 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has quit [Remote host closed the connection]
06:32 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap
07:37 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 246 seconds]
08:04 < RangerMauve> kanzure: +1 to using qwen2.5-coder. I'd say don't go smaller than 3b because it struggles to follow even basic instructions. VSCode with Continue.dev and a local ollama instance has been a good combo in my experience.
08:10 < L29Ah> medllama3_v20 have vanished from huggingface, it shows 404 now
08:11 < RangerMauve> Still on ollama 😈 https://www.ollama.com/ahmgam/medllama3-v20
08:19 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has quit [Remote host closed the connection]
08:20 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap
08:21 < RangerMauve> Gonna pull a copy in case someone wants one in the future
08:23 < kanzure> wanted: encyclopedia of llama 3.2 latent space.
09:12 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has quit [Remote host closed the connection]
09:26 < RangerMauve> Just tried running it and it seems to be trained to answer multiple choice questions more than anything
09:29 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap
09:47 -!- flooded [~flooded@79.127.187.214] has joined #hplusroadmap
10:25 -!- TMM [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
10:25 -!- TMM [hp@amanda.tmm.cx] has joined #hplusroadmap
11:01 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has quit [Remote host closed the connection]
11:03 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap
13:02 -!- WizJin_ [~Wizzy@2402:a00:184:8354:c9a7:174a:d1d2:ed35] has joined #hplusroadmap
13:05 -!- WizJin [~Wizzy@user/WizJin] has quit [Ping timeout: 252 seconds]
16:14 -!- millefy2 [~Millefeui@82.67.97.72] has quit [Quit: Ping timeout (120 seconds)]
16:14 -!- millefy2 [~Millefeui@82.67.97.72] has joined #hplusroadmap
17:47 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has quit [Ping timeout: 276 seconds]
18:59 -!- flyback [~flyback@2601:540:c701:900:a030:e600:b439:c9a] has quit [Quit: Leaving]
19:02 -!- TMM [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
19:02 -!- TMM [hp@amanda.tmm.cx] has joined #hplusroadmap
19:13 -!- flyback [~flyback@2601:540:c701:900:dd07:3e5a:cfdc:fdcb] has joined #hplusroadmap
19:13 < hprmbridge> kanzure> https://medium.com/@kording/why-i-believe-that-the-brain-does-something-like-gradient-descent-27611c491205
20:01 -!- WizJin__ [~Wizzy@2402:a00:184:8354:9dd9:e60b:e3eb:fe0b] has joined #hplusroadmap
20:05 -!- WizJin_ [~Wizzy@2402:a00:184:8354:c9a7:174a:d1d2:ed35] has quit [Ping timeout: 276 seconds]
22:01 < fenn> today i found a nice picture of the standard model on one page: https://miro.medium.com/v2/resize:fit:1400/format:webp/0*1H0ZNsdDXzntp9MV.png
22:01 < fenn> oops wrong channel
22:01 < fenn> not bad to have in the archives i guess
23:02 -!- WizJin_ [~Wizzy@2402:a00:184:8354:9dd9:e60b:e3eb:fe0b] has joined #hplusroadmap
23:05 -!- WizJin__ [~Wizzy@2402:a00:184:8354:9dd9:e60b:e3eb:fe0b] has quit [Ping timeout: 245 seconds]
23:51 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap
--- Log closed Thu Mar 06 00:00:36 2025