engineering finance systems that scale
toronto 📍
Joined June 2026
- Tweets693
- Following227
- Followers44
- Likes2.1K
dale retweeted
Dan Hooker acknowledges that fighting keeps him from going down a destructive path:
“I'm from a generation where I joined this because fighting is the most constructive thing I could possibly be doing with my time.
It saves me from doing very destructive things.
Anything that keeps me in the gym and keeps me fighting is the most productive thing I could possibly be doing.”
dale retweeted
It is the official policy of the United States government that there are only two genders: male & female.
dale retweeted
Only in Canada can a person who works their ass off 40 hours a week pay more in taxes than an “artist” who tells “stories”!
Work your ass off to paint a houses = pay FULL taxes
Paint a picture of a maple leaf = $45k tax break
dale retweeted
Think about it this way. A small model like a 2b or 4b parameter version may KNOW immediately the same amount of information that a specific topic may have on a single Wikipedia page.
A 1T parameter model has every Wikipedia article that has ever been written on that topic + every possible written piece ever published to the internet it could find or be trained on + every single possible abstraction of that same topic all at once, something say 250 to 500 times more vast.
Your amount of RAM hold the amount of vast information, the compute and power behind the RAM is what drives the speed.
last year i watched civil war and the premise felt like pure fiction. a third-term president. texas and california on the same side. the country split into blocs.
then texas goes blue on tuesday.
not wishing for anything close to that movie. but it's wild how a story that felt so far-fetched 18 months ago keeps getting a little less far-fetched.
end goal: enough ram to run local ai models that automate my life and home—all while letting me build whatever i want without subscriptions 😂
this kind of goes against my portfolio thesis though
