Jerry Ma

Hi Chris, the scipy function for KT won't really help for ``calc_swap_deltas`` since we'd still have to run it O(n^2) times for a total runtime of O(n^4). However, there is...

OK, this should be in a semi-decent state now. Please let me know if any changes are needed.

Training even 64x5 is going to be a tricky proposition with 1 GPU (no matter how beefy) ☹️ That said, the speed does seem abnormally slow. You could try running...

Hi @infinitycr, we recommend that you try the latest code in master (maybe use q_min_size 1), and dump the full server logs here if possible. Thanks!

Related to https://github.com/pytorch/ELF/issues/71, we are pushing a fix shortly. Apologies for the inconvenience!

@infinitycr , I believe #72 should fix the issue with master. Please let me know otherwise 😄

Hi @infinitycr , sorry for the issues. Please also try #76 (explanation: there is an additional requirement based on new PyTorch and parallelization of the initial convolution).

Thanks for your interest! We're familiar with the great work done by the LeelaChess folks. There are no immediate plans to adapt ELF to other games such as chess, but...

You are correct that these commands are not implemented. We will put it on our list, and in the meantime we welcome any contributions!

Hi @godmoves, thanks for your interest! The code as released does not support multi-GPU playing. The long launch time is unusual -- if you're still experiencing this and would like...