Dominik Jain
Dominik Jain
@rajfly, I have added issue #1197 in order to avoid confusion about learning rate schedulers in the future. To summarize: * There was no performance issue in Tianshou. * The...
Yes, definitely.
@schroedk this could presumably be handled by specifying a torch device, right? sensai's device handling (using the combination of `cuda` flag and `gpu` index) dates back to torch 0.4 days....
We can only support pandas 2 if it's backwards compatible with pandas 1. If that's not the case, we'll have to add an upper version bound. We can think about...
Tests for regression models were added in commit be2c86050633d17e389ecc7994f17b1122b8a65c
You are assuming that a vectorised computation mechanism for the value generation exists. In many applications, this won't be the case, and the reason for "ColumnGeneratorCachedByIndex is encouraged" is that...
We can safely update dependencies only if backward compatibility is maintained (no code changes necessary), as otherwise it may break ongoing projects (some of which may not be able to...
Is this abandoned?
@MischaPanch ping
Please provide descriptions in English.