Sebastian Raschka

speaker

1,024 appearances 1 recordings 1 series first heard Feb 2026 last heard 1 Feb

Sebastian Raschka’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 1 in all, peaking in Feb 2026 with 1.

Appearances

newest first · ▶ plays the moment

Lex Fridman Podcast · #490 – State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI · 1 Feb 2026 podcast

So we don't have infinite compute.

So you have to decide, do I want to spend that compute more on making the model larger?

But, you know, it's like a trade-off.

It's like in an ideal world, you want to do all of them.

And I think in that sense, scaling is still pretty much alive.

You would still get a better model.

But like we saw with GPD 4.5, it's just not worth it.

I mean, it's like, because you can...

let's say you can unlock more performance with other techniques at that current moment.

Especially if you look at inference scaling, that's one of the biggest gains this year with O1, where it took a smaller model further than pre-training a larger model like GBD 4.5.

So it's like, I wouldn't say pre-training scaling is dead.

It's just like there are other more attractive ways to scale right now at the moment.

But

At some point, you know, you will still want to make some progress on the pre-training.

The thing is also to consider where you want to spend your money.

If you spend it more on the pre-training, it's like a fixed cost.

You train the model and then it has this capability forever.

You can always use it and so forth.

With inference scaling, you don't spend money during training.

You spend money later per query.

Showing 241–260 of 1,024 · page 13 of 52