llm.c, every mention
11 scenes · ← back to llm.c
tap a year for its mentions
every year anyone Andrej Karpathy 27
Verbatim, from the transcripts: the passages where llm.c comes up
[Paper Club] Molmo + Pixmo + Whisper 3 Turbo - with Vibhu Sapra, Nathan Lambert, Amgadoz
- ▶ 1:09:22 unnamed speaker So one of them is like the GPU mode LLM.c taught by Karpathy.
llm.c's Origin and the Future of LLM Compilers - Andrej Karpathy at CUDA MODE
- ▶ 0:53 unnamed speaker To start hacking with others on LLM.C, which became one of the greatest and most active community projects in our server.
- ▶ 1:35 Andrej Karpathy Okay, so I'll tell you a bit about LLM.C.
- ▶ 6:58 Andrej Karpathy And one thing I'd really like to do in LN.c is I just want to keep things simple.
- ▶ 7:41 Andrej Karpathy In LLM.C, all of the allocation happens a single time at the beginning, so we pre-plan all of the memory that we're going to ever use, then it's fixed, and from then on it's just dynamics of just feeding data through it 4 times in the scene
- ▶ 12:13 Andrej Karpathy Uh, PMPP is actually quite good, uh, but also, I think, still kind of like, um, 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…
- ▶ 13:16 Andrej Karpathy A team of Avengers assembled from the internet when they saw LM.C and started contributing, so specifically Eric, Arun, Alexar, kind of like I would say core devs of LM.C and contributed a ton of work to LM.C and they, they started to like… 2 times in the scene
- ▶ 15:44 Andrej Karpathy Um, we implemented all kinds of data streams to overlap the part of the computation, and this ended up creating, like, a total disaster, um, and so that's why I scratched it down, because at one point of alan.c, as Arun would say, I… 4 times in the scene
- ▶ 18:17 Andrej Karpathy Uh, so, uh, you do need QDNN, which is the most heavy dependency, but QDNN is optional, so if you'd like to roll your own manual attention, that is possible in LLM.C, but QDNN is kind of like the hairiest dependency, but after that it's… 4 times in the scene
- ▶ 20:28 Andrej Karpathy We actually thought maybe we would have it done by today, but, uh, there's a few more, a few more, a little bit more work to do, but we will have Lama 3.1, um, training in Lama.c very, very soon. 3 times in the scene
- ▶ 21:26 Andrej Karpathy And that's that I think, um, I mean, what is LLMC? 7 times in the scene