Brett Winton: 在此模型之上叠加人工反馈的强化学习可能成为价值链的一个独立部分。 实际上，这个开源模型的打包人工调优变成了它自己的业务。

原推：A reinforcement learning with human feedback overlay on top of this model may become a separate part of the value chain.

In effect packaged human tuning of this open source model becomes its own business.
https://twitter.com/wintonARK/status/1629178772464484352