Closed PikaCat-OuO closed 2 years ago
I haven't looked much into running it on CPU yet, so I can't really help much here. If someone creates a PR for it I am open to adding it, but I myself likely won't work on it.
Ok, besides, If I have two bin files generated using the same configuration, how can I merge them into one bin file?
On unix simply concat them cat a.bin b.bin > ab.bin
.
I get it. Binary concatenation. Simply copy one file to the tail of the other file.
I found out that https://github.com/glinscott/nnue-pytorch/issues/87 managed to run the trainer on cpu only mode. But as I notice that at some point https://github.com/glinscott/nnue-pytorch/commit/0764091ec544446ae5852e3e7296f252a2046dd1 a custom kernel has been applied to the feature transformer. So It means more changes need to be applied to make it work on cpu side.
But the changes of the custom kernel is so large that it's very hard to make it run on cpu? i.e. a whole bunch of changes need to be applied?