How does kanachan vs mortal perform?
Pragmatism0220 opened this issue · 1 comments
Pragmatism0220 commented
Sorry, I can't find the discussion area, so I can only send an issue. I trained kanachan and mortal respectively. Perhaps because I didn't adjust the parameters carefully, in my test, the decisions of kanachan and mortal were basically consistent. But I haven't conducted a match test between them. So I'm curious, how did kanachan perform? Thank you!
Cryolite commented
kanachan is designed to provide a high degree of freedom in training models, so performance is likely to vary greatly depending on how the model is trained. Therefore, there is no representative performance of kanachan.
For a comparison with Mortal, I will ask the author of Mortal if it is possible (e.g., specification of simulations for comparison).