ctak/resources, branch statsAn implementation of Tak and a computer opponent in Chttps://git.l-3.space/ctak/atom/resources?h=stats2021-03-18T03:15:15ZCorrected generation of training data for 6s2021-03-18T03:15:15Ztslil clingmantslil@posteo.de2021-03-18T03:10:53Zurn:sha1:0010aa36411e7b0d0e58a1b3f52d0e2e246f68e8Don't generate header for training data + tweaks2021-02-12T02:04:42Ztsliltslil@posteo.de2021-02-12T02:04:42Zurn:sha1:dd25d66353121939a8f9b2bfe6a20f4100ff4976
For some reason it would seem that moving flats to a lower value and
increasing the proximity between caps and top flats improves
acquisition. Still not great, but every bit counts.
Change the training data generation a little2021-02-11T01:07:24Ztslil clingmantslil@posteo.de2021-02-11T01:07:24Zurn:sha1:52b2268a1c7b97809bdf0f02147f885ddfa6d397
Although it pains me to say it, ``label smoothing'' appears to be
actually work. I'm also currently experimenting with training simply
against _all_ games, instead of only bot matches. Once the training
finishes i'll pit cttei against itself with old and new weights,
hopefully there'll be a noticeable improvement.
TEI interface working!2021-02-05T05:12:18Ztslil clingmantslil@posteo.de2021-02-03T00:23:54Zurn:sha1:4f20d9fd63b6b3e30130c84c20f7cdbbbf2df76dTried some naive iterative deepening. Work on TEI interface next2021-02-02T02:46:24Ztslil clingmantslil@posteo.de2021-02-02T02:46:24Zurn:sha1:41b5dc810e25724e8111cf7f2ad1b4b52f9b1206
If TEI is implemented, then i could make use of Morten's
racetrack (https://github.com/MortenLohne/racetrack) and develop a
quantitative measure of the bot's performance. This is the current
priority.
Just some #weightgoals ;)2021-02-01T04:33:39Ztslil clingmantslil@posteo.de2021-02-01T04:33:39Zurn:sha1:a7d80d7346a7c6944d917df4903342527b17d71c
It turns out that while i was training on a 0/1 classification
problem, i was using 2*eval - 1. Training using this function instead,
and on bot-dominated game choices (chosen_player in extract.sh) seems
to have given a better evaluation function. At the least, Morten's
swindle doesn't work anymore.
Fairly important bug fixes to lcdlib, LCD now echoes input!2021-01-31T23:27:41Ztslil clingmantslil@posteo.de2021-01-31T20:33:30Zurn:sha1:948b4286d0b5c5a647d2ab3dd20cd12301f166be
Input polling without line-buffering is done using ncurses, so the
buildroot configuration had to change accordingly to include that
library.
The Makefile changed to accommodate stand-alone building of ct1986 and
to include -lcurses where appropriate. There were also some typos
about copying ct1986 and ctaklm to the correct directories.
Syntax errors, small tweak to training data generation2021-01-29T19:05:47Ztslil clingmantslil@posteo.de2021-01-29T19:05:06Zurn:sha1:e5fc8bbb6c127d8b08249bb4855d958e575eae01Cleaned up build system2021-01-29T18:06:49Ztslil clingmantslil@posteo.de2021-01-29T18:06:49Zurn:sha1:0ce9856d34c58f6cb9c38281e04d1c4cbd09804cBusybox is irritating2021-01-22T00:52:52Ztslil clingmantslil@posteo.de2021-01-22T00:52:52Zurn:sha1:426a0bfc107b354eb45b9c067e63cbe554c7e023