고정된 트윗
MB
25.8K posts
After a thorough investigation I found out that the rumors of Drake paying this Lady 8 million dollars were false, and Drake only said he lost 8 million dollars on sports betting/gambling
English
Have you spent time inspecting reasoning traces of your kids? 🙂
English
On election night, no-name, no support Nithya Raman sobbed & conceded the race
She apologized to her family for failing
This made sense. She polled in the single digits & bombed in the debate. Now a historically insurmountable 3rd place
This was before “the machine” took over
English
This is the PERFECT example of why foreigners should not be allowed to run for office
Zul Mohammed just ran for Mayor of Carrollton, Texas. He’s from Pakistan
“No vet has made any sacrifice. I want to make that clear. I do not support the US military. No, I do not support the United States. I look down on both entities. I want to make that clear”
I can’t think of a better example of why only natural born citizens should be allowed to run for office
Also he is a Muslim, which further enforces the classic “I do not support the United States”
We need new election eligibility laws
English
The West Point Glee Club performed “Onward Christian Soldiers” on the Ed Sullivan Show in 1954
I’m glad they weren’t asked to sing “Onward Christian Soldiers, Excluding Latter-day Saints”
The Pentagon shouldn’t disparage the beliefs of Americans willing to die for their country
English
As eval is downstream of everything, it determines whether you will spend your time optimizing the right metrics.
The current gap between academia and industry AI labs is the attitude toward eval.
In academia, the eval set is very hard to change since a) you need to explain why your eval is better and b) you need to benchmark against your cited works with the new eval and show that your work is superior.
Doing both a and b at the same time invites risky rebuttal, even if you are doing a good job on a. It is far easier to benchmark against the eval set that everyone has agreed upon.
In contrast, in industry AI labs, customer feedback is your eval set and it keeps changing to cover the long tail that you could never think of during years of PhD programs.
If the loss functions are not a good proxy for customer feedback, then you change them until both are aligned.
Thus, academia might train students who are very good at hill climbing but inexperienced in building eval sets that capture hard real cases. To move the needle, building the right eval set matters the most.
English
"Tesla is to protect life on Earth,
SpaceX to extend life beyond."
— Elon Musk
English
