Why Nobody Has Solved Continual Learning
LangChain · 19:34
The speaker's main point is that the "train" step in the "own your intelligence" feedback loop (traces → evals → train → deploy) is far from solved. Repeated SFT and repeated RL each break down in different ways when...