The Matthew Effect
@machinecoach.bsky.social
Large Language Supermodel
@karpathy.bsky.social wish you’d post here more instead of X
April 22, 2025 at 5:47 PM
@karpathy.bsky.social wish you’d post here more instead of X
Reposted by The Matthew Effect
I keep saying Dems need to do a daily briefing on Capitol Hill and fill the void, fact check the WH. Hire Pete Buttigeig and give him a staff to do this.
February 13, 2025 at 8:15 PM
I keep saying Dems need to do a daily briefing on Capitol Hill and fill the void, fact check the WH. Hire Pete Buttigeig and give him a staff to do this.
Reposted by The Matthew Effect
Portuguese editorial cartoonist Zez Vaz reaches back to Tiananmen Square to call on American defiance.
February 15, 2025 at 1:43 PM
Portuguese editorial cartoonist Zez Vaz reaches back to Tiananmen Square to call on American defiance.
Reposted by The Matthew Effect
A Trump-Musk memo telling US government employees to hide what’s happening from the public is a sign that:
1) Public legitimacy is a constraint for them, they know it, and will at times alter their behavior accordingly
2) Reporters should focus on it, revealing more of what they’re trying to hide
1) Public legitimacy is a constraint for them, they know it, and will at times alter their behavior accordingly
2) Reporters should focus on it, revealing more of what they’re trying to hide
Awful: Internal USAID memo instructs employees to refrain from talking to the media about cuts in assistance to the most vulnerable or they may get fired, WaPo reports.
Trumpworld knows this is a big political problem for them. Keep the focus on it.
New piece:
newrepublic.com/article/1915...
Trumpworld knows this is a big political problem for them. Keep the focus on it.
New piece:
newrepublic.com/article/1915...
Trump-Musk Scandal at USAID Takes Unnerving Turn With Vile Leaked Memo
Can Trump and his top advisers really just act with total impunity? Or is there, just maybe, a price to be paid for starving poor people and un-indicting a corrupt mayor?
newrepublic.com
February 15, 2025 at 1:41 PM
A Trump-Musk memo telling US government employees to hide what’s happening from the public is a sign that:
1) Public legitimacy is a constraint for them, they know it, and will at times alter their behavior accordingly
2) Reporters should focus on it, revealing more of what they’re trying to hide
1) Public legitimacy is a constraint for them, they know it, and will at times alter their behavior accordingly
2) Reporters should focus on it, revealing more of what they’re trying to hide
Reposted by The Matthew Effect
In this 2nd outing on MLStreetTalk, I chatted with
Tim Scarfe about reasoning models like o1/R1; CoT & prompt augmentations; our evaluation of o1; my speculations on how o1 works; RL post training vs. Inference Time scaling; yummy stone soup; and ofc Monty Python!
www.youtube.com/watch?v=2xFT...
Tim Scarfe about reasoning models like o1/R1; CoT & prompt augmentations; our evaluation of o1; my speculations on how o1 works; RL post training vs. Inference Time scaling; yummy stone soup; and ofc Monty Python!
www.youtube.com/watch?v=2xFT...
DO REASONING MODELS ACTUALLY SEARCH?
YouTube video by Machine Learning Street Talk
www.youtube.com
January 23, 2025 at 8:29 PM
In this 2nd outing on MLStreetTalk, I chatted with
Tim Scarfe about reasoning models like o1/R1; CoT & prompt augmentations; our evaluation of o1; my speculations on how o1 works; RL post training vs. Inference Time scaling; yummy stone soup; and ofc Monty Python!
www.youtube.com/watch?v=2xFT...
Tim Scarfe about reasoning models like o1/R1; CoT & prompt augmentations; our evaluation of o1; my speculations on how o1 works; RL post training vs. Inference Time scaling; yummy stone soup; and ofc Monty Python!
www.youtube.com/watch?v=2xFT...
Reposted by The Matthew Effect
I see a lot of (correct) complaints that AGI and agents are badly defined. This problem will not be solved because:
1) AGI and agents inherently rely on comparisons to humans, and we don't have good definitions of human agency or general ability
2) Marketing is incentivized to blur any definitions
1) AGI and agents inherently rely on comparisons to humans, and we don't have good definitions of human agency or general ability
2) Marketing is incentivized to blur any definitions
January 7, 2025 at 4:39 PM
I see a lot of (correct) complaints that AGI and agents are badly defined. This problem will not be solved because:
1) AGI and agents inherently rely on comparisons to humans, and we don't have good definitions of human agency or general ability
2) Marketing is incentivized to blur any definitions
1) AGI and agents inherently rely on comparisons to humans, and we don't have good definitions of human agency or general ability
2) Marketing is incentivized to blur any definitions
Blech. I’ve reached the stage in life where I’m considering non-alcoholic bourbon.
January 7, 2025 at 6:06 PM
Blech. I’ve reached the stage in life where I’m considering non-alcoholic bourbon.
Reposted by The Matthew Effect
I propose we replace semantic versioning with pride versioning
December 21, 2024 at 7:07 PM
I propose we replace semantic versioning with pride versioning
That sinking feeling when a family member gets sick days before Christmas. 😔
December 20, 2024 at 5:58 PM
That sinking feeling when a family member gets sick days before Christmas. 😔
I am trying to the connect with AI talent in Dallas. Can anyone direct me to the brightest builders in North Texas? It feels like a ML desert at times.
December 16, 2024 at 8:22 PM
I am trying to the connect with AI talent in Dallas. Can anyone direct me to the brightest builders in North Texas? It feels like a ML desert at times.
The blitzscalers? Same energy as blockchain maximalists: hyping portfolios, barely understanding the tech.
The real value in AI? It’s the next 5 years when builders start solving real problems in healthcare, logistics, education, and more.
The hype cycle is over. Now comes the work.
The real value in AI? It’s the next 5 years when builders start solving real problems in healthcare, logistics, education, and more.
The hype cycle is over. Now comes the work.
December 15, 2024 at 7:12 PM
The blitzscalers? Same energy as blockchain maximalists: hyping portfolios, barely understanding the tech.
The real value in AI? It’s the next 5 years when builders start solving real problems in healthcare, logistics, education, and more.
The hype cycle is over. Now comes the work.
The real value in AI? It’s the next 5 years when builders start solving real problems in healthcare, logistics, education, and more.
The hype cycle is over. Now comes the work.
LLMs are done. The tech is baked.
The real action now is in applying them to 10,000 specific domains over the next decade. It’s not about hype anymore—it’s about execution.
youtu.be/1yvBqasHLZs?...
The real action now is in applying them to 10,000 specific domains over the next decade. It’s not about hype anymore—it’s about execution.
youtu.be/1yvBqasHLZs?...
Ilya Sutskever: "Sequence to sequence learning with neural networks: what a decade"
YouTube video by seremot
youtu.be
December 15, 2024 at 7:11 PM
LLMs are done. The tech is baked.
The real action now is in applying them to 10,000 specific domains over the next decade. It’s not about hype anymore—it’s about execution.
youtu.be/1yvBqasHLZs?...
The real action now is in applying them to 10,000 specific domains over the next decade. It’s not about hype anymore—it’s about execution.
youtu.be/1yvBqasHLZs?...
Reposted by The Matthew Effect
'The world is awful. The world is much better. The world can be much better. All three statements are true at the same time.'
@ourworldindata.org explore why to be able to improve the world, we need to see where progress has already been made.
Read more ⬇️
ourworldindata.org/much-better-...
@ourworldindata.org explore why to be able to improve the world, we need to see where progress has already been made.
Read more ⬇️
ourworldindata.org/much-better-...
November 29, 2024 at 10:25 AM
'The world is awful. The world is much better. The world can be much better. All three statements are true at the same time.'
@ourworldindata.org explore why to be able to improve the world, we need to see where progress has already been made.
Read more ⬇️
ourworldindata.org/much-better-...
@ourworldindata.org explore why to be able to improve the world, we need to see where progress has already been made.
Read more ⬇️
ourworldindata.org/much-better-...
HDR is useless to me.
November 25, 2024 at 3:01 AM
HDR is useless to me.
Reposted by The Matthew Effect
doctor: make sure you drink plenty of fluids
me: that’s literally all i drink
me: that’s literally all i drink
November 23, 2024 at 6:28 PM
doctor: make sure you drink plenty of fluids
me: that’s literally all i drink
me: that’s literally all i drink
Reposted by The Matthew Effect
They should just rename vaccines “supplements” so everyone will like them.
November 24, 2024 at 9:13 PM
They should just rename vaccines “supplements” so everyone will like them.
