You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A string that is constructed using "" has a token-count that depends on its size. I would assume that this is done to avoid freely storing lots of data.
This token cost can be avioded by putting your data in the name of a function, and then returning nameof() within that funtion.
The text was updated successfully, but these errors were encountered:
Sience a name is a single token, could You in theory make the stockfish neural network wheights and biases into a single string and load that for the first round and use that as your evaluation?
Dammit, I was thinking of abusing this with an emulator (not that it was going to fit, I'm surprised how much I fit in though). BTW, this isn't unbounded, it is merely a more than intended.
A string that is constructed using "" has a token-count that depends on its size. I would assume that this is done to avoid freely storing lots of data.
This token cost can be avioded by putting your data in the name of a function, and then returning nameof() within that funtion.
The text was updated successfully, but these errors were encountered: