Llm.c

product on 1 show · 9 statements across 1 episodes · said 29 times in 2 episodes since 2024

Latent Space 29

Mentions by year, every show

tap a year for its mentions
001513022024episodesmentions
0122024episodes it came up in
007.511522024episodesmentions per episode

Latent Space 29

2024 29 mentions in 2 episodes 15 per episode

every mention on every show, scene by scene, with the transcript →

9 statements about Llm.c, every show

LATENT SPACE Prediction Not checkable as stated
Karpathy predicts LLMs will act as compilers generating bare-metal CUDA code
“If LLINs are about to become much better at coding over time, then I think you can expect that the LLIN could actually do this for any custom application over time. And so the LLINs could act as a kind of compiler What you're interested in, they're gonna do al…”
Andrej Karpathy Sep 21, 2024 ▶ 21:54 llm.c's Origin and the Future of LLM Compilers - Andrej Karpathy at CUDA MODE
LATENT SPACE Assertion Supported
Karpathy: llm.c was 20% faster and used 30% less memory than PyTorch
“At the time of that post, we were using, in LL and that's in 30% less memory, and we were 20% faster in training, just the truth.”
Andrej Karpathy Sep 21, 2024 ▶ 19:10 llm.c's Origin and the Future of LLM Compilers - Andrej Karpathy at CUDA MODE
LATENT SPACE Assertion Supported
Karpathy: llm.c achieves nearly 50% MFU on single-node GPT-2 training
“We have almost a 50% NFU here on one node, which is quite good.”
Andrej Karpathy Sep 21, 2024 ▶ 18:45 llm.c's Origin and the Future of LLM Compilers - Andrej Karpathy at CUDA MODE
LATENT SPACE Assertion Supported
Karpathy: llm.c trains GPT-2 on one H100 node in 24 hours for $600
“You can train it on a single node of H-one-hundreds in about 24 hours, and that costs roughly 600 dollars.”
Andrej Karpathy Sep 21, 2024 ▶ 18:02 llm.c's Origin and the Future of LLM Compilers - Andrej Karpathy at CUDA MODE
LATENT SPACE Disclosure
Karpathy: llm.c rejects performance PRs if added complexity harms developer onboarding
“A lot of lmc just kind of, like balancing the improvement and speed with the complexity of what you're actually introducing, and so I've actually rejected a lot of PRs because of that, because the code starts to get crazy, and I think that decreased the amount…”
Andrej Karpathy Sep 21, 2024 ▶ 17:25 llm.c's Origin and the Future of LLM Compilers - Andrej Karpathy at CUDA MODE
Karpathy: The popular PMPP textbook lacks advanced CUDA optimization techniques
“PMPP is actually quite good but also, I think, still kind of like mostly on the beginner level, because a lot of the CUDA code that we ended up developing in the lifetime of the LMC project, you would not find those things in, in this book, actually.”
Andrej Karpathy Sep 21, 2024 ▶ 12:13 llm.c's Origin and the Future of LLM Compilers - Andrej Karpathy at CUDA MODE
LATENT SPACE Disclosure
Karpathy: llm.c was created while jet-lagged on vacation in the Maldives
“All of this work that I described so far, I was on vacation and while I was jet-lagged in Maldives. So I, basically it's perfect because you wake up at one a.m., and there's nothing to do. So you write stuff like LLN.C.”
Andrej Karpathy Sep 21, 2024 ▶ 10:03 llm.c's Origin and the Future of LLM Compilers - Andrej Karpathy at CUDA MODE
LATENT SPACE Assertion Supported
Karpathy: llm.c avoids runtime crashes by pre-allocating memory statically without dependencies
“It's a single file C. There's no dependencies whatsoever. It compiles instantly. It runs instantly. All the memory is just allocated in a single blob. So if you start stepping, there's no way you're gonna boom later. It's all preplanned. It's fully determinist…”
Andrej Karpathy Sep 21, 2024 ▶ 9:17 llm.c's Origin and the Future of LLM Compilers - Andrej Karpathy at CUDA MODE
LATENT SPACE Disclosure
Karpathy: llm.c trains transformers in C with minimal C++
“We're training transformers in C at a pinch of C++.”
Andrej Karpathy Sep 21, 2024 ▶ 1:38 llm.c's Origin and the Future of LLM Compilers - Andrej Karpathy at CUDA MODE

← every entity, every show

Made with StarZero

Turn any episode into a week of clips.

This entire site, thousands of episodes across every show transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.