there is an entire small world inside the model.

don't forget this. ask chatgpt transparency and you shall recieve.

Heroic work!

metagenerations are so cool

holy cow

yes!!

this will be fixed

congrats, tom white

RL’s Razor: On-policy RL forgets less than SFT. Even at matched accuracy, RL shows less catastrophic forgetting Key factor: RL’s on-policy updates bias toward KL-minimal solutions Theory + LLM & toy experiments confirm RL stays closer to base model

amazing clarity

so that's what it looks like as a video

will i get more followers if I post my sora 2 access code here

we just reached number 3 on app store! i have had so many friends tell me sora 2 is the first time ai made them laugh, this is truly a new experience!

6921E0

i did not contribute much to sora 2 - but I had the pleasure of witnessing it's development. the level of care, attention and love put into a single model and a single product was amazing to witness. i've always believed deep learning rewards care, but I am now convinced the

We’ve developed a new way to train small AI models with internal mechanisms that are easier for humans to understand. Language models like the ones behind ChatGPT have complex, sometimes surprising structures, and we don’t yet fully understand how they work. This approach