You Jiacheng
@YouJiacheng
加入 August 2015
2.4K 正在關注 12.6K 粉絲
it's more like RL with expert init?
the hard thing is how to restrict the transition inside the constrained set C. this work uses a privileged prompted model to act as the transition.
but this might be over-restrictive and break the on-policy-ness.
顯示更多
