“can train”
10,000
clips found
4
videos
Examples in video
YouTubeLex Fridman758K views · 2026-01-31
State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI | Lex Fridman Podcast #490
“can train on those best answers, and that's
synthetic data. It's very different than”
also the fact that ChatGPT now
gives wonderful answers, and you … early ChatGPT with lots
of hallucination data.
Play at 69:39
YouTubeLex Fridman4.5M views · 2024-08-02
Elon Musk: Neuralink and the Future of Humanity | Lex Fridman Podcast #438
“they can train off that data specifically”
so like if I play Webgrid for 10 minutes, … in order to get me a better model.
Play at 476:10
YouTubeLex Fridman758K views · 2026-01-31
State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI | Lex Fridman Podcast #490
“so that others can train these models.
There's the Institute for Foundation”
we have actual competition for people
that are trying to release everything … Models/LM360, which has had their K2
Play at 29:56
YouTubeLex Fridman758K views · 2026-01-31
State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI | Lex Fridman Podcast #490
“how much you can train or still get benefit,
where RLHF, because it's a preference”
- So RLVR is more, let's say, unlimited in … tuning, you reach a certain point where it
doesn't really make sense to spend more
Play at 113:11
YouTubeLex Fridman758K views · 2026-01-31
State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI | Lex Fridman Podcast #490
“because you can train it faster and you're
not hitting these compute walls as soon.”
The reason we get the Claude
4.5 Sonnet model first is … They can just try a lot more things and get
the model faster, even though the bigger
Play at 170:12
YouTubeJay Shetty Podcast2.1M views · 2024-05-27
F1 Driver Charles Leclerc Unveils The Real Behind The Scenes of Formula 1 Racing
“sports in the world you can train as”
to others uh football or whatever other … much as you want in form one you can't I
Play at 34:59
YouTubeTheo Von1.3M views · 2025-06-05
Arnold Schwarzenegger | This Past Weekend w/ Theo Von #587
“And I don't know if you can train a”
all of this stuff is also personality. … personality or not. I mean I don't know
Play at 49:44
YouTubeLex Fridman758K views · 2026-01-31
State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI | Lex Fridman Podcast #490
“out which problems to give the model and how long
you can train it for and how much inference you can”
out which problems to
give the model and how … enable the model to use when solving
these verifiable problems. So as
Play at 108:34
YouTubeLex Fridman758K views · 2026-01-31
State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI | Lex Fridman Podcast #490
“frontier models are all pushing into these
harder domains when they can train on more”
generates so many tokens or much
harder software problems. So the … problems and the model will learn
more skills at once. The RLHF
Play at 109:23
YouTubeLex Fridman758K views · 2026-01-31
State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI | Lex Fridman Podcast #490
“unlock to have these models because you can
customize them, you can train them, you”
model locally how they would use ChatGPT.
But also for companies I think it's a huge … can add post-training,
add more data. Like,
Play at 35:51