@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.

kellogh, to random
@kellogh@hachyderm.io avatar

imo the war on drugs indirectly caused the current immigration crisis on the southern border. not sure that helps the current situation to acknowledge, but there, i said it

hgrsd, to tech
@hgrsd@hachyderm.io avatar

I'm a few weeks away from returning to work after paternity leave. I'm finding it hard to imagine that I'll care all too much about work upon my return to the world of SaaS companies.

How do other parents deal with this? Any motivational tips and tricks?

kellogh,
@kellogh@hachyderm.io avatar

@hgrsd decide what you want out of life. Your can have family be #1 or #2 priority, and either is fine. But you’ll regret it if you don’t confront that decision directly. Having work be #2 doesn’t mean being bad at it. Lots of good software engineers that prioritize family as #1

adron, to random
@adron@metalhead.club avatar

So just clarifying…

Multiple Republicans have been arrested for child porn lately.

Santos, a Republican, has been arrested for fraud.

Trump got charged with rape and related sexual assault charges.

Am I missing anything?

kellogh,
@kellogh@hachyderm.io avatar

@adron the party’s response to these crimes

xgranade, to random
@xgranade@wandering.shop avatar

Not raising pay in an inflationary economy is a form of pay cut.

kellogh,
@kellogh@hachyderm.io avatar

@xgranade honestly didn’t seem like anyone was trying to hide that fact…

samir, to random
@samir@functional.computer avatar

I used to ask a programming interview question along the lines of "implement FizzBuzz as badly as you can". It helps me understand what horrors they've seen.

I think I need to update it to "convince ChatGPT to spit out something that's as terrible as possible, and explain why it's terrible".

kellogh,
@kellogh@hachyderm.io avatar

@samir ha, it’s not a bad way to interview, although you’d want other data points to go with it. If you do this, though, you’re effectively requiring that they know how to use ChatGPT and I’m not sure, yet, that’s a reasonable thing to require

kellogh,
@kellogh@hachyderm.io avatar

@samir imo its an extremely useful tool, and despite how magical it feels, it takes skill to avoid it's pitfalls and get the most value out of it.

Most of the criticism is people asking it to do dumb things and it does dumb things, which was never interesting

karchie, to random
@karchie@freeradical.zone avatar

Well, that was exciting

kellogh,
@kellogh@hachyderm.io avatar

@karchie i'll show that to my 7yo. she'll be excited to hear that Elsa's magic powers are real

kellogh, to ai
@kellogh@hachyderm.io avatar

if you're interested, I put the code for my 7yo's AI assistant in a gist.

In a nutshell:

  1. ChatGPT with speech recognition
  2. For stories, ChatGPT prompts StableDiffusion to generate an image
  3. For factual stuff, ChatGPT prompts the app to search Wikipedia (via @duckduckgo) and re-prompts for a more factual response

https://gist.github.com/tkellogg/e52aa97ebd73a6d48ac54ba1d7d22912

kellogh,
@kellogh@hachyderm.io avatar

@dcopeland tbh I loved that book, as sad as it was, and it was sort of intentional

kellogh, to random
@kellogh@hachyderm.io avatar

this may be a hot take, but if you find yourself saying "g'night all" at work, you're probably working in a fucking sweat shop

kellogh, to random
@kellogh@hachyderm.io avatar

This is great! We need to be explainable and understandable. But a few problems

  1. Billions of parameters
  2. Nobody knows what an explanation is

https://openai.com/research/language-models-can-explain-neurons-in-language-models

kellogh,
@kellogh@hachyderm.io avatar

When you have billions of anything, most things get difficult. Doesn't really matter what you're talking about. Explanations are very tough as it is, it's going to be very tough to give people what they're looking for.

More on that — Most / explanation algorithms have suffered from people misusing them, to mean things they were never meant to mean

kellogh,
@kellogh@hachyderm.io avatar

Last, most people don't know what an explanation is. Its easy to talk about informally, but it's REALLY hard to define precisely. This 60 page paper gives an overview of all the things we mean when we talk about "explanation", all humanities research. It's worth a read if you want to understand how complex we really are https://arxiv.org/abs/1706.07269

kellogh,
@kellogh@hachyderm.io avatar

@russelsteapot42 exactly!

kellogh, to random
@kellogh@hachyderm.io avatar

people who say things like "biz" annoy me to an irrational extent

kellogh, to random
@kellogh@hachyderm.io avatar

Is this real? Do LLMs report any kind of certainty via the logits? Would be cool if they did, but seems like that would be much bigger news

kellogh,
@kellogh@hachyderm.io avatar

Yesterday I saw this, which is similar. But I don’t think what they’re doing is as simple as “reading the logits”. Seems like they had to train a model to interpret them. So I guess it’s not out of the question https://syncedreview.com/2022/11/16/deepminds-epistemic-neural-networks-enable-large-language-model-fine-tuning-with-50-less-data/

kellogh, to random
@kellogh@hachyderm.io avatar

Bluesky is initialized as BS and I don’t think that’s an accident

kellogh, to random
@kellogh@hachyderm.io avatar

I think it’s safe to say that it’s generally a bad thing when your local news is national news

kellogh, to random
@kellogh@hachyderm.io avatar

I wonder how strong the correlation is between people who grew up in Jewish/Christian homes and fear of snakes

kellogh,
@kellogh@hachyderm.io avatar

I was talking to someone from India and they’re like, “yeah, I don’t know what the big deal is. We had a king cobra live under our porch for 6 years. They’re so calm and docile”

kellogh,
@kellogh@hachyderm.io avatar

Not sure if it’s actually religiously associated, but it’s kinda hard to let go of “the devil is a snake (and therefore all snakes are the devil)”

dnalounge, to random
@dnalounge@sfba.social avatar

♬ NU METAL WORLD at DNA Lounge starting now!
https://www.dnalounge.com/calendar/2023/05-05d.html

kellogh,
@kellogh@hachyderm.io avatar

@dnalounge Nu metal? More like ye old metal

quintessence, to random
@quintessence@hachyderm.io avatar

I have made such a habit of Close All Tabs Friday that when I miss a week it's pretty obvious.

Closing or bookmarking / storing all tabs weekly has really helped my "tab hygiene". Because research tab explosions are real. :blobfoxheartcute:

kellogh,
@kellogh@hachyderm.io avatar

@quintessence don’t you miss them?

lmorchard, to random

Trying out subscribing to a relay with my personal basement Mastodon instance. This may be a bad idea, but we'll see what happens!

kellogh,
@kellogh@hachyderm.io avatar

@lmorchard it would be cool to have a “crawler” that maps out how connected each instance is to the rest of the fediverse. Something visual

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • anitta
  • Durango
  • magazineikmin
  • InstantRegret
  • hgfsjryuu7
  • vwfavf
  • Youngstown
  • slotface
  • thenastyranch
  • ngwrru68w68
  • rosin
  • kavyap
  • PowerRangers
  • DreamBathrooms
  • cisconetworking
  • khanakhh
  • mdbf
  • tacticalgear
  • ethstaker
  • modclub
  • osvaldo12
  • everett
  • tester
  • cubers
  • GTA5RPClips
  • normalnudes
  • Leos
  • provamag3
  • All magazines
    •