Sam Altman

burning out is real btw. when i was 18 i thought it was something for dummies who didn't wanna work. it took me ~3 years to recover fully, and could physically not work for 5months after it happened

2 years ago diffusion lora training was literally side project I did in the final exam week because i was too lazy to study for the exam ...and now its industry of its own 🤷

research is hard.

Many people leaving everything and grind RL because it’s the trend. Others going all in on CUDA. Same people who will start grinding flow matching when someone releases a video gen model. Choose something you like and do it consistently. Grass is greener where you water it.

Training our most capable Gemini models relies heavily on our JAX software stack + Google's TPU hardware platforms. If you want to learn more, see this awesome book "How to Scale Your Model": jax-ml.github.io/scaling-book/ It was put together by my Google DeepMind colleagues

Cool. Now someone should make arxiv reels.

semianalysis.com/2023/05/04/goo…

The pharmacy course is a side project. Tbh if I had a choice I would leave it😅

o3-mini(tiny) Then it can be distilled to phone-level

all of my knowledge is self taught

companies just wanna make money, so all you need to do is show them how you can help them make money drop everything you have ever been told about getting a job and start over from that simple statement

I have EXCITING news: I've started a company! Introducing Sophont We’re building open multimodal foundation models for the future of healthcare. We need a DeepSeek for medical AI, and Sophont will be that company! Check out our website & blog post for more info (link

excited to share that I have joined Audiogen (as a research intern) to build the next generation of audio/music generation models 🚀. It's been extremely cool working with the team so far 🫡

having a real job really interferes with my ability to tweet 20 times a day

"Google has no moat" yeah sure dude

last presentable Jax thing I did was making a checkpointing library for Flax(nnx API) models. still needs more testing and improvement. I created it cus I wanted to make something simpler than the rest github.com/kelechi-c/nnx_…

officially an unc