Chintala: Time and data constrain Meta LLM releases more than GPUs
Soumith Chintala · Open Source AI is AI we can Trust — with Soumith Chintala of Meta AI · Mar 6, 2024 · at 46:46
Meta AI leader Soumith Chintala discusses resource allocation and why models stop training to move onto next-generation Llama iterations.
0:00 / 0:42exact quote · 42.4s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“So, I think the, it's all a matter of time. I think time is the biggest bottleneck. It's like, when do you stop training the previous one, and when do you start training the next one? And how do you make those decisions? The data, do you have net new data, better clean data for the next one, in a way that it's not worth, like, really focusing on the previous one? It's just a standard iterative product. You're like, when is the iPhone one? When, like, when do you start working on iPhone two? Where is the iPhone? Like, so on, right? So, mostly the considerations are time and generation rather than GPUs, in my opinion.”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Soumith Chintala
Chintala: George Hotz's TinyGrad requires major breakthroughs to match PyTorch
“There's no, like, I don't think, like, unless we have, like, great breakthroughs, like, George's vision is achievable, like, or, like, he should be thinking about a narrower problem, such as, I'm only gonna make this for, like, work for self-driving car con ne…”
Chintala: Apple's MLX will fail server-side due to lack of differentiation
“If they end up expanding onto the server side, and they'll probably build something like PyTorch as well, right? Like, eventually, that'll where it will land. And I think there, they will kind of fail on the, like, lack of differentiation. Like, it wouldn't be…”
Chintala: Inference Moats from Fast CUDA Kernels Last Only Months
“I think, like, Together and Fireworks and all these people are trying to build some faster CUDA kernels and faster, like, you know, hardware kernels in general. But those modes only last for a month or two. Like, these ideas quickly propagate.”
Chintala: Simple AI frameworks must accept long compile times
“You can write a very simple framework but then you also should be willing to eat the long compile times of, like, searching for that optimal performance at runtime.”
Chintala: Nvidia's primary competitive moat is NVLink interconnect, not GPU silicon
“The mode that Nvidia has right now, I feel like, is that they're, they have the interconnect that no one else has. Like, AMD GPUs are pretty good. I'm sure there's very silicon that is not bad at all, but, like, the interconnect like, NVLink is uniquely awesom…”
Chintala: LLM inference market will become a low-margin laundromat business
“My view of the LLM inference market in general is that it's like the laundromat model. Like you, the margins are going to drive down towards the bare minimum, like It's gonna be all kinds of arbitrage between how much you can get the hardware for, and then how…”