So far I used mainy training mode target weekneses, but now I tried also other twoo modes and it doesn't work as good as I expected.
Random mode choose moves with same probability, right? I have small side line in the first moves. So I get this small side line in 50% cases. But it is not balanced, because the main line has much more moves(positions). I would prefer to change the probability depending on how many moves(positions) folows.
And also full coverage mode works a little bit strange. For example I have 100 positions. And I want to train first 50 positions in one day and another 50 positions in next day. But in next day I get almost same 50 positions as first day. So there probably in no "memory" and I have to train all 100 position at once, if I want to train them all.