Page 1 of 2

Lc0 Evaluation Explanation

Posted: Thu Aug 29, 2019 5:30 pm
by Hamster
I have read the explanation of the Lc0 output on https://github.com/LeelaChessZero/lc0/w ... Chess-Zero
but I still do not understand it or I do not know how to get the most from it. As an example of the challenge I am facing see the below output from Arena. Why is Be2 the move that Lc0 selects? It is neither the one with the highest P nor Q nor Q+U nor V value.

Image

Re: Lc0 Evaluation Explanation

Posted: Thu Aug 29, 2019 6:08 pm
by cma6
" highest P nor Q nor Q+U nor V value."
Hamster, please excuse this lc0 newbie question: What is the meaning of "P", "Q", "Q+U", & "V"?

Re: Lc0 Evaluation Explanation

Posted: Thu Aug 29, 2019 6:10 pm
by zullil
cma6 wrote: Thu Aug 29, 2019 6:08 pm " highest P nor Q nor Q+U nor V value."
Hamster, please excuse this lc0 newbie question: What is the meaning of "P", "Q", "Q+U", & "V"?
Follow link in his post.

Re: Lc0 Evaluation Explanation

Posted: Thu Aug 29, 2019 6:34 pm
by grahamj
LC0 chooses the move with the most visits. Searching the explanation for 'visits' may help.

Re: Lc0 Evaluation Explanation

Posted: Thu Aug 29, 2019 6:55 pm
by zullil
grahamj wrote: Thu Aug 29, 2019 6:34 pm LC0 chooses the move with the most visits. Searching the explanation for 'visits' may help.
So, in the image posted above, Be2 is chosen based on N = 152?

Re: Lc0 Evaluation Explanation

Posted: Thu Aug 29, 2019 7:03 pm
by Paloma
zullil wrote: Thu Aug 29, 2019 6:10 pm
cma6 wrote: Thu Aug 29, 2019 6:08 pm " highest P nor Q nor Q+U nor V value."
Hamster, please excuse this lc0 newbie question: What is the meaning of "P", "Q", "Q+U", & "V"?
Follow link in his post.
Which post?

Re: Lc0 Evaluation Explanation

Posted: Thu Aug 29, 2019 7:09 pm
by zullil
Paloma wrote: Thu Aug 29, 2019 7:03 pm
zullil wrote: Thu Aug 29, 2019 6:10 pm
cma6 wrote: Thu Aug 29, 2019 6:08 pm " highest P nor Q nor Q+U nor V value."
Hamster, please excuse this lc0 newbie question: What is the meaning of "P", "Q", "Q+U", & "V"?
Follow link in his post.
Which post?
Well, there's only one embedded link in this entire thread.

Re: Lc0 Evaluation Explanation

Posted: Fri Aug 30, 2019 2:17 am
by jp
Hamster wrote: Thu Aug 29, 2019 5:30 pm Why is Be2 the move that Lc0 selects? It is neither the one with the highest P nor Q nor Q+U nor V value.
Do you mean it played Be2 after the screenshot?

It looks to me that it does have the highest Q+U, as well as the highest N, but the only one the list of moves is in order for is N (visits).

I don't know what the "(+7)" in the top line means or the "(712)".

Re: Lc0 Evaluation Explanation

Posted: Fri Aug 30, 2019 10:23 pm
by Hamster
jp wrote: Fri Aug 30, 2019 2:17 am
Hamster wrote: Thu Aug 29, 2019 5:30 pm Why is Be2 the move that Lc0 selects? It is neither the one with the highest P nor Q nor Q+U nor V value.
Do you mean it played Be2 after the screenshot?

It looks to me that it does have the highest Q+U, as well as the highest N, but the only one the list of moves is in order for is N (visits).

I don't know what the "(+7)" in the top line means or the "(712)".
You are right, Q+U is actually the highest.

If Lc0 chooses the move with the most visits what would be a useful interpretation of the other variables?
E.g. why not choose the average expected value of all playouts for a move (Q)?

Re: Lc0 Evaluation Explanation

Posted: Fri Aug 30, 2019 11:13 pm
by cma6
Zulli:
Do you know the meaning of these terms: "P", "Q", "Q+U", & "V"?
I am guessing that V = visits. But, as pointed out, Be2 does not have the highest V value of all the moves shown. Or does the lc0 site have an explanation of the parameters, since when I searched here for "visits", it gave me no new information.