Cerebras Systems, every mention
41 scenes · ← back to Cerebras Systems
tap a year for its mentions
every year anyone Sarah Chieng 29Sean Lie 10Shawn Wang 9Matthew Berman 5Alessio Fanelli 4Drew Houston 3Dylan Patel 2Doug O'Laughlin 2Andrew Feldman 2Tri Dao 1
Verbatim, from the transcripts: the passages where Cerebras Systems comes up
The Inference Frontier: from 100 to 10,000 tokens per second — Sean Lie, Cerebras CTO
- ▶ 1:22 unnamed speaker Okay, we're here at Cerebris HQ with CTO Shawn Lee. 2 times in the scene
- ▶ 3:54 Sean Lie So, you know, we, we, we, we designed, uh, our next generation CS four architecture, um, with a new brand new system platform with the goal really, um, to make, 2 times in the scene
- ▶ 12:24 Sean Lie So right now, uh, we are basically, you know, sold out 4 times in the scene
- ▶ 15:56 unnamed speaker Decode cerebris? 2 times in the scene
- ▶ 24:15 Sean Lie And so I think that's definitely one of the things that was a topic of discussion, again, independent of, of Cerebris. 2 times in the scene
- ▶ 26:21 unnamed speaker Like, but it's good for Cerebus. 3 times in the scene
- ▶ 33:35 Sean Lie But I think that there's also a lot of value in trying to push, you know, the boundaries in other vectors as well, which is obviously, you know, what we're trying to do here at Cerebris.
- ▶ 36:13 Sean Lie I think for the entire industry, right, um, even if I take my Cerebrus hat off for a second, right, what excites me the most is
Next 100x in AI: Inference, Networking, & Self-Optimizing Models — Philip Kiely & Ali Taha, Baseten
- ▶ 1:11:03 Shawn Wang Compare this versus any Cerebrus diagram, right?
Podcast Crossover: AIE, AGI, frontier lab strategy with @matthew_berman and @swyxtv
- ▶ 3:30 Matthew Berman Ken to a Grok or a Cerebris? 4 times in the scene
- ▶ 6:36 Matthew Berman Maybe not, but like, you know, the Cerebrus, uh, sorry, not the Cerebrus, um,
The Stove Guy: Sam D'Amico Shows New AI Cooking Features on America's Most Powerful Stove at Impulse
- ▶ 18:29 Sam D'Amico I probably should run some like super fast cerebrus thing, but we'll get to that later.
Dylan Patel Explains the AI War While Cooking | In-Context Cooking
- ▶ 41:52 Dylan Patel Jensen's very paranoid, um, and that makes him, like, an amazing founder, um, and CEO, um, and so you, you have all these people freaking out, but it's like, the moment he sniffed wind of the OpenAI Cerebrus deal, he immediately went out… 2 times in the scene
Claude Code for Finance + The Global Memory Shortage: Doug O'Laughlin, SemiAnalysis
- ▶ 1:53:48 Doug O'Laughlin Until Cerebrus and Grok, honestly, they were all considered failures, and even then, we're like, what are they gonna do with Grok? 2 times in the scene
How Zyphra went all-in on AMD + Why Devs feel faster with AI but are slower — with Quentin Anthony
- ▶ 18:30 unnamed speaker Have you worked with any of, like, the GPU alternative hardware, like Grok, Cerebrus? 4 times in the scene
- ▶ 57:36 Quentin Anthony If I was AMD or Cerebris or whoever else, and I would go out and say,
⚡️Raising $1.1b to build the fastest LLM Chips on Earth — Andrew Feldman, Cerebras
- ▶ 0:04 unnamed speaker We're in the remote studio celebrating a big, big race with Andrew Feldman of Cerebris. 4 times in the scene
- ▶ 12:13 unnamed speaker I I'm getting a sense that like most of these are just trained on, on non-Cerebra silicon and then inference on silicon, on Cerebra silicon. 3 times in the scene
- ▶ 19:04 unnamed speaker Uh, now, now there's a dedicated effort on Cerebus code, which is doing very, very well. 2 times in the scene
- ▶ 25:58 unnamed speaker What is the Cerebrus grandmaster plan for like, you know, like the next 10 years?
- ▶ 28:55 Andrew Feldman Well, thank you and, and the Latent Space team for, for inviting Cerebrus onto your show. 2 times in the scene
⚡️Accelerators @ 3x NVIDIA H200 perf, Made in the USA - Thomas Sohmers + Mitesh Agrawal, Positron AI
- ▶ 7:46 Alessio Fanelli And maybe contrapose that both to, you know, NVIDIA and the kind of traditional ones, as well as Grok, obviously you work there, Cerebras, and then there's kind of like the long tail of all the other GPU alternatives things, but just give…
- ▶ 12:04 Shawn Wang Can I, can I also just spell out the, this also applies to Grok and the Cerebris and other sort of big chip companies, right? 2 times in the scene
- ▶ 39:13 Mitesh Agrawal You know, if you look at Cerebrus and Grok and others, they've, they've really tried to do this, their cloud kind of portal.
Claude Plays Pokémon Hackathon: Escape from Mt. Moon!
2024 Year in Review: The Big Scaling Debate, the Four Wars of AI, Top Themes and the Rise of Agents
- ▶ 14:38 Shawn Wang Cerebras will be happy to tell you that it gets their four or five B on their super large chips.
[Paper Club] Weight Streaming on Wafer-Scale Clusters (w/ Sarah Chieng of Cerebras)
- ▶ 0:06 Sarah Chieng I'm Sarah Chang and today I'm going to be presenting the weight streaming paper by Cerebrus. 6 times in the scene
- ▶ 1:30 Sarah Chieng Um, but basically this paper is called Training Giant Neural Networks Using Weight Streaming on Cerebrus Wafer Scale Clusters, and this paper covers weight streaming, which is a training technique, public, um, training execution flow by… 6 times in the scene
- ▶ 5:04 Sarah Chieng So, you know, this is, you know, a public facing chat that Cerebrus has. 3 times in the scene
- ▶ 5:58 Sarah Chieng So anyways, going back to the paper, though, um, so the paper is talking about how you combine each of these different, you know, so the Cerebris' main hardware is the WaferScale Engine three, and the WaferScale Engine three sits inside… 6 times in the scene
- ▶ 11:21 Sarah Chieng And so this system that Cerebrus introduces actually has three different components.
- ▶ 21:39 Sarah Chieng In this paper and, you know, in production, Cerebra's focus on data parallelism.
- ▶ 23:15 Sarah Chieng So here, so instead of CPU, GPU processing, we're looking at the Cerebris CSX system, which as I mentioned is what's, you know, the system that has that wafer scale engine inside. 2 times in the scene
- ▶ 36:10 Sarah Chieng Um, and so Cerebris leverages weight sparsity in two main ways during training. 3 times in the scene
- ▶ 40:39 Sarah Chieng And so this paper talks about, you know, two different ways that you can take all these activation tensors and actually lay it out onto the cerebrus wafer.
Building the Silicon Brain - Drew Houston of Dropbox
- ▶ 52:59 Drew Houston And then there's also, like, non-NVIDIA stacks, like the Grok, or Cerebris, or some of these custom silicon companies that are super interesting, and all, and outperformed the NVIDIA stack in terms of latency and things like that. 3 times in the scene
Building an open AI company - with Ce and Vipul of Together AI
- ▶ 12:16 Shawn Wang From Cerebras. 2 times in the scene
The State of Silicon and the GPU Poors - with Dylan Patel of SemiAnalysis
- ▶ 37:51 Alessio Fanelli So the TPU is obviously one, but there's Cerebras, there's GraphCore, there's Madaxx, Lemurian Labs, there's a lot of them. 3 times in the scene
FlashAttention-2: Making Transformers 800% faster AND exact
- ▶ 38:24 unnamed speaker Have you spent any time looking at some of the new kind of like AI chips companies, so to speak, like the cerebris of the world, like, 2 times in the scene
- ▶ 57:14 Tri Dao Um, we've seen some, you know, quite a few companies releasing this, you know, together, um, released, uh, Red Pajama, uh, Dataset, I think Cerebus, then worked on that and, you know, deduplicate and clean it up and release Slim Pajama and…
Ep 18: Petaflops to the People — with George Hotz of tinycorp
- ▶ 4:49 Shawn Wang Cerebris. 3 times in the scene