Andrew Jackson

ok, what do you want me to do with these that i've got?

Ok, good to know! The community features on gokibitz.com are great, and a community-sgf-pastebin is super useful to be able to point new players at.

@kris-computer-go i'm not the author but i'm fairly certain the code used to compute the ratings is the same as the code used here: https://github.com/tensorflow/minigo/blob/master/ratings/ratings.py#L190 I scraped the .md files...

One possibility would be to change this line https://github.com/justinmayer/virtualfish/blob/6988e4aa09e058e07b10edc94524071f90d1fbf5/virtualfish/virtual.fish#L876, instead piping to something in `/tmp/`, so the caller could dump it out on a nonzero return code.

No, but i'm not the right guy to ask because i'm not an expert on the edge cases of virtualenvs, i'm just a guy trying to keep projects isolated and...

@jjoshua2 your theory is sound but you're misunderstanding how the pipeline works. For standard supervised learning (SL), looking at every data point makes sense and you can take the optimum...

(This "how many unique results are there for the value net" is why, in the AGZ paper, they describe training a supervised network on pro games by weighting the value...

for lack of a better place to put it -- what are the "bunches to do" -- anything i can help with? I'm currently adapting all the stuff in the...

Yes, it applies to all nodes, but doing 8x reads is (generally) better than averaging all of them. With the inference cache, since it's close to free to rebuild the...

What do you want to do with this issue, now that we're doing it? :)