kellogh

kellogh, 4 months ago to python

update on #fossil: last night i pushed changes to make algorithms pluggable. it uses #python entry points for discovery, so you can develop your own plugin without cloning fossil. it gives you a few hooks:

• train: e.g. if you need to train an ML model, or just calculate aggregate statistics

• render: format toots into an object that’s passed to a jinja template. you can make the jinja template or just use one that already exists

• HTML forms: e.g. to capture hyper-parameters

kellogh, 2 months ago to random

why would you want a machine to be conscious? setting aside the debate of if it is, what would that let it do that an unconscious machine couldn't?

kellogh, 5 months ago to random

What are the PagerDuty alternatives these days?

kellogh, 10 months ago (edited 10 months ago) to random

How many first-name@company email addresses have you had? Over your whole career

kellogh, 7 months ago to random

this part made me sad. it must hurt to live like this

kellogh, 6 months ago to random

ngl i’m getting too old to be able to do the kind of mental gymnastics that lead to “just because it’s true doesn’t mean it’s a fact”

kellogh, 4 months ago to llm

a lot of takes today about how anything you can do with an #LLM would be better done with “ML”. yes, always from people who never did ML in their life.

see a claim. check their bio, they’ve never even heard of a binary classifier, yet they’re suddenly ML experts

kellogh, 20 days ago to random

with this new definition of “anti-semitic”, is it possible to discuss israeli politics without being anti-semitic?

kellogh, 1 year ago to random

i've noticed that there's a lot of fruitful #ai development around purely treating #LLMs as black boxes and focusing on prompt engineering + the ReAct pattern. Simply forcing the LLM to draw out it's thoughts over more text increases it's accuracy, and if you also interleave that with input from the user or calls to external services, e.g. Google, you can achieve very interesting results

kellogh, 7 months ago to LLMs

since everyone hates saying “open source”, what do i call these non-proprietary #LLMs?

kellogh, 6 months ago to random

protip: if you have trouble changing Git branches and frequently work multiple things at once, just checkout lots of copies of the repo

e.g.

• feature-dev/
• escalations/
• things-mike-makes-me-do/

kellogh, 6 months ago to random

when you use code reviews as a quality gate, it reduces bugs by reducing the throughput of the team. there’s not many cases where that’s actually what you want. typically you want to look for other quality controls

kellogh, 6 months ago to LLMs

Thought:

1. #LLMs aren't trained on data after 2021

2. Most writing becomes done by LLMs

Does that mean our language will start becoming frozen in time? Slower change?

kellogh, 5 months ago to random

my grandparents generation would use words like “smidgeon”, “dash”, “pinch”, “tad”, etc. but it wasn’t until my wife got these measuring spoons that i realized that they were actual imperial units

kellogh, 11 months ago to random

Java’s proposition of “compile once, run everywhere” seems very weak in 2023

kellogh, 2 months ago to random

in 2012 most tech people were worried about a bubble. Also in 2013-15, and 2016-20, and 2021-now. i thought for sure the crypto crash was going to trigger a larger bubble pop. yet somehow, it still hasn't happened.

not saying it won't happen, but there's a whole lot of history behind naysayers being wrong

kellogh, 7 months ago to random

fyi non-native english speakers: “fine” (as a one-word response) means the opposite of what you think https://m.youtube.com/watch?v=KfcrM7ukzCU

kellogh, 4 months ago to opensource

#polars is the ideal #opensource project, imo. it hits all the important things for me

• #rust
• #python
• #datascience
• replacing #pandas
• performance engineering
• integrates with a large open ecosystem instead of creating a walled garden
• pleasant to use

https://github.com/pola-rs/polars/releases/tag/rs-0.36.2

kellogh, 5 months ago to ChatGPT

this isn’t rocket science. if you put information in, it will come out. an exploit might not exist today, but it’s only a matter of time before it’s common. training data, prompts, RAG-injected info…all of it needs to follow basic security principles. https://not-just-memorization.github.io/extracting-training-data-from-chatgpt.html #chatgpt #ai #llms #llm #genai #GenerativeAI

kellogh, 3 months ago to random

dang... a nationwide AT&T outage. no wonder why things feel dead this morning https://www.foxbusiness.com/lifestyle/cell-phone-outage-hits-att-customers-nationwide-verizon-tmobile-users-also-affected

kellogh, 2 months ago to LLMs

Let’s be honest, if you’re a software engineer, you know where all this compute and power consumption is going. While it’s popular to blame #LLMs, y’all know how much is wasted on #docker, microservices, overscaled #kubernetes, spark/databricks and other unnecessary big data tech. It’s long past time we’re honest with the public about how much our practices are hurting the climate, and stop looking for scapegoats https://thereader.mitpress.mit.edu/the-staggering-ecological-impacts-of-computation-and-the-cloud/

kellogh, 1 month ago to hiking

anyone know of a plant identification app that actually works in the spring when the leaves aren’t full adult size? seek isn’t cutting it #hiking #botany #nature #camping

kellogh, 26 days ago to llm

i low key don't want to see a big jump in #LLM or #AI capabilities anytime soon. rn they're capable enough that my mom wants to use them, but bad enough that even she has an intuitive sense for when they're wrong

that's how you build "AIQ", the skill of using it. Lots of people toying with them, to feel out their capabilities and limitations

kellogh, 10 days ago to llm

there’s a big need for something stronger than #RAG, but more flexible and cheaper than a giant all-knowing #LLM.

A great part about RAG is that it’s just a database. You just issue INSERT/UPDATE and yeah, that’s how you maintain knowledge. No million dollar training process

kellogh, 18 hours ago to random

this has been bugging me a lot. like, yeah, there’s definitely AI scams out there. and yeah, a lot of people are using it from the wrong end, but it’s also clearly a substantial technology. time to realize that
https://mas.to/@carnage4life/112484753548884371