Technically Speaking
What DeepSeek figured out, and why you shouldn't copy it (w/ Julien Launay & Sesh Nalla)
1:17:42
For years the recipe for better AI was simple: more data, bigger models.
Then DeepSeek's R1 showed models could learn to reason by practicing on problems with checkable answers, and Reinforcement Learning became the way forward.
But what actually is RL? And is it something every startup should be doing, or just a practice for big AI labs?
Julien Launay, cofounder of Adaptive ML, and Sesh…