@kellogh@hachyderm.io
@kellogh@hachyderm.io avatar

kellogh

@kellogh@hachyderm.io

I'm a software engineer and sometimes manager. Currently #Raleigh but also #Seattle. Building ML platform for a healthcare startup. Previously, built an IoT platform for one of "those" companies.

Open source: dura, fossil, Jump-Location, Moq.AutoMock, others

Do I have other interests? No, but I do have kids and they have interests. I think that counts for something. I can braid hair and hunt unicorns!

I put the #rust in frustrate

He/Him

#metal #science #python

This profile is from a federated server and may be incomplete. Browse more on the original instance.

dfeldman, to random
@dfeldman@hachyderm.io avatar

Google apparently accidentally posted docs on the roughly 14,000 features that go into search results

Lots of somewhat surprising/interesting ones in there

For example, author names of blog posts, articles, etc. are special; it seems to have different tags for site types (personal, work, etc); lots and lots of stuff for filtering abusive/sexual/violent material

kellogh,
@kellogh@hachyderm.io avatar

@dfeldman you have to go to duck duck go to search for the news article

eleventy, to random
@eleventy@fosstodon.org avatar

objection.

kellogh,
@kellogh@hachyderm.io avatar

@eleventy @zlatiah you didn’t specify the performance dimension

kellogh, to random
@kellogh@hachyderm.io avatar

it’s funny that “legalistic” means a very strict interpretation, but in 2024 general society has become so saturated with computers and technology that actual legal interpretation seems vague, muddy and subjective

kellogh,
@kellogh@hachyderm.io avatar

in the web3 days, “code as law” seemed like a good idea, probably because people fundamentally misunderstood law

kellogh,
@kellogh@hachyderm.io avatar

those days were wild. people would say, “what if we…” and then outline the plot to a black mirror episode

one of the side effects of working in software is you most often observe software in a failing state, so replacing fundamental societal structures with code seems like a really bad idea

dev, to random
@dev@discuss.systems avatar

Squirrels seem to be the apex predators around these parts

kellogh,
@kellogh@hachyderm.io avatar

@dev they eat nuts, which is pretty threatening imho

kellogh, to random
@kellogh@hachyderm.io avatar

i know “ginormous” didn’t used to be a word, but is it a word now?

kellogh,
@kellogh@hachyderm.io avatar

@bobmcwhirter you probably say jithub too

kellogh,
@kellogh@hachyderm.io avatar

@okohll that's ridonkulous

renwillis, to ai
@renwillis@mstdn.social avatar

This is super interesting. It's just a conversation with a research about a study on LLM #AI actual potential. It's only a 13min video, but basically the study shows that most likely AI can't have infinite growth because there is a limited amount of data available to train on.

There's a finite amount of data & no matter how they mathed it out, it's not a J-curve or linear growth of AI's capabilities, but an S-curve every time.

Worth a watch: https://youtu.be/dDUC-LqVrPU?si=Tcuh4NAwsxInvoev

#tech #technology

kellogh,
@kellogh@hachyderm.io avatar

@renwillis more specifically, there’s strong evidence that exponential curves never exist in nature, only logistic curves in disguise

carnage4life, to random
@carnage4life@mas.to avatar

Another example of how Apple is getting left behind in the AI wars is that both Windows PCs and Chromebooks are going to have AI assistants, generative AI image tools and writing tools.

Apple will probably just announce the thinnest MacBook ever. Again.

https://blog.google/products/chromebooks/chromebook-plus-google/

kellogh,
@kellogh@hachyderm.io avatar

@carnage4life i’m a huge advocate for LLMs, but i’m not sure Windows and Google are good examples for how to deploy them. imo Apple has been quietly killing it in terms of downsizing models, and I have a feeling they’re going to suddenly announce something soon that’s a gigantic UX improvement, and leaves everyone in the dust

kellogh, to random
@kellogh@hachyderm.io avatar

people have been saying they’re getting a lot of spam from gmail.com, which is strange, i get most of mine from joebiden.com

kellogh, to rainbow
@kellogh@hachyderm.io avatar

the other day we had a #rainbow in our front yard. absolutely crazy seeing a rainbow so close. oddly there was no pot of gold or unicorns. the 3yo was ultimately disappointed
#rain #outdoors

kellogh,
@kellogh@hachyderm.io avatar

@dragonsidedd right, i imagine they used spectrum shifters to only show up as infrared

micah, to 3DPrinting
@micah@ilbery.family avatar

@kim is going to see her nieces tomorrow so I busted out the 3D printer so she could be the best bad influence possible :neofox_wink_blep: :neofox_peek_knife:
#3DPrinting #sword

kellogh,
@kellogh@hachyderm.io avatar

@micah @kim if you don’t like her, print out hundreds of beads and give them to her kids

kellogh, to random
@kellogh@hachyderm.io avatar

calling that MS Recall thing a “keylogger” is simply unfair to keyloggers

kellogh,
@kellogh@hachyderm.io avatar

@patrizia right, and a keylogger doesn’t capture what’s on screen

kellogh, to random
@kellogh@hachyderm.io avatar

i’m thinking about unfollowing all the threads accounts because replies don’t seem to work, afaict. not a fan of one-way bridges

kellogh,
@kellogh@hachyderm.io avatar

threads+fedi could have been good. so could’ve the bluesky bridge. but fedi people get so hostile toward established networks that we can’t have nice things

futurebird, (edited ) to random
@futurebird@sauropods.win avatar

Without looking it up what is closest to 8oz?

(Curious how well the relation between oz and metric is generally known.)

kellogh,
@kellogh@hachyderm.io avatar

@futurebird my rule of thumb is that a liter is about the same as a quart. a nalgene bottle has quart and liter markings at the top, almost the same. so 8oz is ~1/4 of a liter

foone, to random
@foone@digipres.club avatar

Terrible idea: an author writes a book, but instead of releasing it, they train an LLM on it, and release the model.

So readers can ask the LLM to generate text based on it, but can never know the original text.

kellogh,
@kellogh@hachyderm.io avatar

@foone this seems spiritually similar to releasing a story as a video game. lemme think…

fine tune a model to act kinda like a dungeons and dragons dragon master. apply control vectors so it’s nearly impossible to have it act out of character. integrate it into a game, such that the LLM can issue commands and participate. prompt it with a storyline and let it go. have game mechanics (e.g. dice rolls + cards) determine the actual flow

kellogh, to random
@kellogh@hachyderm.io avatar

the 8yo tried sounding out “Raleigh” and gave me a death stare when i told her how it’s actually pronounced

twilliability, to random
@twilliability@genart.social avatar

Maybe humans need friction; maybe part of civilizational harm is removing too much of it!

If I can drive or ride up escalators, I stop moving my ass. It's more comfy, but I get fat and die early of heart attack.

If Google AI parses the top search result and goes straight to answering my query, I no longer pore through pages of links myself. It's more comfy, but I stop weighing contextual clues for trustworthiness. I get bad answers and stop even noticing.

kellogh,
@kellogh@hachyderm.io avatar

@twilliability continuation of that thought — if i can drive or ride escalators, i also have to go to the gym to stay in shape.

if AI removes the mental friction, i have to get it some other way in order to maintain mental strength. do i play more chess? work, but at a higher level? play music instead?

marcel, to ai
@marcel@waldvogel.family avatar

Modern #AI text generators create randomized output with no prior planning. They resist to be quality-checked by tools and processes established in the software industry.

Given this, the results are amazing. However, companies are selling the idea that these assistants will do quality checking themselves soon™.

This is mass delusion. But hey, the perks for managers/investors are worthwhile 🤷.

#LLM #ChatGPT #Google #Gemini
https://www.theverge.com/2024/5/24/24164119/google-ai-overview-mistakes-search-race-openai

kellogh,
@kellogh@hachyderm.io avatar

@marcel @TheServitor correctness is a little wonky to talk about, because the comparison isn’t direct. in engineering, we assume that 100% correctness isn’t achievable, if you haven’t found bugs, you just haven’t used it long enough. but with a program, every time you find a bug and fix it, you’re permanently closer to correct, whereas with an LLM, each inference is a new case. all you can do is optimize the entire model. so LLM correctness is statistically extremely difficult

anmey, to llm
@anmey@social.anoxinon.de avatar

I think one of the biggest fears people have about AI is that it isn't perfect as assumed, but that, like us humans, it takes the given information, assumes the most likely outcome, and presents it plausibly.

#ki #LLM #AI

kellogh,
@kellogh@hachyderm.io avatar

@anmey yeah, there’s this paradox — we kinda want computers to think like humans, but when they get plausibly good at it, we complain that they don’t think like computers anymore

  • All
  • Subscribed
  • Moderated
  • Favorites
  • JUstTest
  • rosin
  • mdbf
  • tacticalgear
  • osvaldo12
  • InstantRegret
  • DreamBathrooms
  • cubers
  • magazineikmin
  • thenastyranch
  • khanakhh
  • Youngstown
  • slotface
  • kavyap
  • megavids
  • ethstaker
  • tester
  • GTA5RPClips
  • Durango
  • modclub
  • Leos
  • ngwrru68w68
  • everett
  • anitta
  • cisconetworking
  • provamag3
  • normalnudes
  • lostlight
  • All magazines