Let me quote from the research paper since it can involve many legacy works:
"By using a combined policy and value network architecture, and by using a low weight on the value component, it was possible to avoid overfitting to the values (a problem described in prior work 12)."
Just a side note for other readers. In some early articles, I find myself need to be more careful about the use of "we". For established technologies, the word "we" refers to the practitioners and the readers on how to implement these technologies. I hope that it does not give people the impression that I was involved with the original research paper. That is not my intention. The article is written as a demonstration/illustration but not in the form of a research paper. Sometimes, the paper researchers are so well known that I thought there is no confusion. But if any confusion to any readers, I am sorry.
I will review this article again to fix that and hopefully, for more popular articles, I can change them gradually if it is a problem and if time is allowed.