ctak/Makefile, branch mcu An implementation of Tak and a computer opponent in C https://git.l-3.space/ctak/atom/Makefile?h=mcu 2021-01-17T00:37:18Z It works, but it's too slow 2021-01-17T00:37:18Z tslil tslil@posteo.de 2021-01-17T00:37:18Z urn:sha1:70bc5fb82a53f0646aa8d50d419c6956a2240745 Starting work on MCU port 2021-01-16T20:00:03Z tslil tslil@posteo.de 2021-01-16T19:52:56Z urn:sha1:bc9d62c4ea7de191471ca34de8f0694c049f5736 Quick sketch of non-determinism 2021-01-16T06:11:22Z tslil tslil@posteo.de 2021-01-16T06:11:22Z urn:sha1:40f8b0d0ec33ebd7ba6e3b730f15be43bda9a0a0 Dragon is not separate, fixed concession, allowing playing as b/w 2021-01-16T04:45:16Z tslil tslil@posteo.de 2021-01-16T04:45:16Z urn:sha1:e945328f7b06a62d76365ab90578871b073f4735 So after a small tweak, it actually plays well! 2021-01-14T18:00:18Z tslil tslil@posteo.de 2021-01-14T18:00:18Z urn:sha1:5a9157b120678d794e7c9c90d477284a16a90230 Trying alpha-beta pruning. There's a bug somewhere ... 2021-01-14T05:33:42Z tslil tslil@posteo.de 2021-01-14T05:33:42Z urn:sha1:bb3e2916ac2df096553da6a9c42f15ee3a6a02b8 Fixed minimax (!), fixed bugs in tak.c 2021-01-14T04:44:12Z tslil tslil@posteo.de 2021-01-14T04:41:32Z urn:sha1:5c0980452875f1201a85b9c4df0e965a8ce9da1c With minimax of depth 1 the evaluation function seems alright with the current method of training and generating weights Trying various things to teach ti 2021-01-13T05:18:30Z tslil tslil@posteo.de 2021-01-13T05:18:30Z urn:sha1:4b7f79c77097b86f1c837592388b77ab06f73e5c New weights 2021-01-12T21:19:31Z tslil tslil@posteo.de 2021-01-12T21:19:31Z urn:sha1:5b1a2acd18e9e8eb33d10ba3e219985c13c75f5f Train on data that is close to the end of the game only 2021-01-12T06:15:49Z tslil tslil@posteo.de 2021-01-12T06:15:49Z urn:sha1:43f572d1c3f8eb6e3dce7ac2b9c13774f4b283af