-
What method would I go about if I wanted to create and train a new bot that is geared towards only beginner players, namely those in the ranks of 35 kyu to no higher than 10kyu. This would be a stron…
-
#I am seeking clarification about something. Currently, in leela-zero, when you do genmove b, then genmove w, with 1600 playouts, you get this :
```
Leela: genmove B
Thinking at most 50.0 seconds.…
remdu updated
6 years ago
-
Haven't checked all but I have looked into four different weight files.
It seems that in those files all channel biases are remained 0.0, i.e. all line number 4N-1 (0
-
I see you did random flip and rotation in the Player's expand_and_evaluate function.
I think this is need when add data for training, but don't necessary when evaluate for selecting action.
How abou…
-
Maybe a stupid and useless question but what's the point of the temperature parameter since we always choose the best move with the visit count ?
Moreover, if we decide the choose randomly the move w…
-
Splitting out a discussion for @gcp's comment in https://github.com/gcp/leela-zero/issues/591#issuecomment-357496732
> Bootstrapping 6x128 was (very!) unsuccessful so far. From eyeballing the results…
-
Mods feel free to close this since I know its not directly relevant. However I have the hunch that Leela Zero is already (network 128) beyond top pros (when on best hardware and professional match tim…
-
After we have default policy, why don't we think about reinforcement learning? Training begins with the default policy. At a basic level, this policy is simply played against itself iteratively, updat…
-
The original Find app appears to be significantly more accurate currently during real world test. I am wondering if it takes the a.i. a few days to figure out what to prioritize?
Is there any chanc…
-
In the paper, they say the following thing : "The rules of Go are invariant under rotation and reflection; this knowledge has been used in AlphaGo Zero both **_by augmenting the dataset during trainin…