i am 11

Does anyone have an H100 bare-metal node that I could use to debug kernels? I am not planning on using the compute on it. I just need to run Nsight Compute a few times a day.

great city of palo alto

Dinner not looking bad today!

LLMs don’t have feelings ofc

does anyone know where I can access a cloud 5090 with nsight compute permissions?

i ask gemini for help with jax sometimes, with hope that google might've done something to make the model know more jax every single time, the answers are dog shit. but i will keep trying. . . . . .

the bitter lesson of software is that OpenAI should just improve the coding agent instead of hand rolling it

how did peter thiel stand the one piece animations

life starts when you believe that things are within your ability and within your control

describing the memory layout of a single tensor in CuTe

three years ago, DiT replaced the legacy unet with a transformer-based denoising backbone. we knew the bulky VAEs would be the next to go -- we just waited until we could do it right. today, we introduce Representation Autoencoders (RAE). >> Retire VAEs. Use RAEs. 👇(1/n)

Moving back to Palo Alto ;—;

youtube.com/watch?v=Yf1o0T… some words to recite