Commit graph

58 commits

Author SHA1 Message Date
bcb6cb9da1 add xNES optimizer 2018-06-09 18:56:10 +02:00
fe9494b0d5 refactor ARS out of main (breaks a bunch of stuff) 2018-06-09 18:56:10 +02:00
d3e6441c40 reduce tile input to 5 per row using new layers 2018-06-09 16:20:20 +02:00
dd5ec3dbde make network linear 2018-06-09 16:20:07 +02:00
cbb094adc9 restore flagpole bonus, add missing overlay check 2018-06-09 04:35:09 +02:00
9fb98d3fe0 allow setting of world-level, plus random option 2018-06-09 04:34:21 +02:00
81d6b509d0 detect when mario is controllable 2018-06-09 01:43:22 +02:00
9b23327df4 add score multiplier 2018-06-08 23:59:43 +02:00
fec148fb79 don't turbo in playable mode, note overlay bug 2018-06-08 14:51:17 +02:00
c30f07f407 prevent reward gained from fireworks 2018-06-08 14:12:21 +02:00
c40e1f929d fix skipped inputs on lag frames 2018-06-08 13:48:59 +02:00
e24c3d31a4 use argsort 2018-06-08 13:46:38 +02:00
374fa4d876 cleanup 2018-06-08 02:45:07 +02:00
37d404e77d reduce embed layer to values actually used ingame 2018-06-07 22:40:31 +02:00
3030e83d00 refactor learn_from_epoch 2018-05-14 01:34:08 +02:00
15f0292485 remove defer_prints option (now always true) 2018-05-12 22:56:04 +02:00
0fb3b1780f remove some old comments 2018-05-12 22:55:04 +02:00
a836314b8b refactor game and utility functions 2018-05-12 22:44:53 +02:00
7db43038ac adjust range of timed inputs to stdev of roughly 1 2018-05-07 16:27:51 +02:00
7357c8ed62 move LayerNorm after Relu 2018-05-07 16:22:48 +02:00
946f05bd3e base timed inputs on start of trial time 2018-05-07 16:20:59 +02:00
ce64801368 fix some inputs 2018-05-07 09:19:24 +02:00
5201b75509 add Lipschitz heuristic/approximation 2018-05-07 05:57:52 +02:00
ee066154b2 add test trial logging 2018-05-07 05:57:18 +02:00
deb1ea7de0 add LayerNorm layer 2018-05-07 05:55:58 +02:00
feaf86dc6b allow weights/params file to be configured 2018-05-04 21:02:08 +02:00
90922a2bc3 add AMSgrad optimizer and logging 2018-05-03 16:48:12 +02:00
7831f534c9 tweaks 2018-05-03 16:48:12 +02:00
2bdd67b721 add playback_mode 2018-05-03 16:48:12 +02:00
b453438055 add graycode-like distribution option 2018-05-03 16:48:12 +02:00
6a01f609a9 split strictness to its own file 2018-05-03 16:48:12 +02:00
545618c70b refactor config vars to their own files 2018-05-03 16:48:12 +02:00
66bf689e04 reduce time waiting at world screen, tweak config 2018-05-03 16:48:12 +02:00
5636c7b2ed tweak config and network 2018-05-03 16:48:12 +02:00
d696bd8c21 reimplement softchoice and redo noise generation 2018-05-03 16:45:38 +02:00
5b98023073 experimental ARS stuff 2018-03-26 16:32:00 +02:00
57cead431d add TODOs to notice 2018-01-30 20:25:23 +01:00
27098141c3 not literally 2018-01-30 19:49:52 +01:00
c76ec6a87c more cleanup 2017-09-09 19:46:35 +00:00
01d7e5e230 cleanup 2017-09-09 19:37:01 +00:00
d384635000 select outputs from array instead of binary combinations 2017-09-08 10:43:32 +00:00
88dcd203a1 tweaks and fixes 2017-09-08 10:40:19 +00:00
5a8c0f6140 add graphviz printing and stuff 2017-09-07 23:06:43 +00:00
acc8378980 add and utilize Merge and Embed layers 2017-09-07 23:06:30 +00:00
6f2ffcdef7 fix frameskip stuff and give mario an extra life 2017-09-07 22:11:57 +00:00
3b4e195ae6 add frameskip 2017-09-07 21:20:53 +00:00
3d64df0574 preliminary batches and backwards passes
also adds negated noise trials because i forgot to commit that earlier
2017-09-07 19:18:11 +00:00
db603753f4 support multiple nodal inputs 2017-09-07 19:09:44 +00:00
16b38a7151 improve node traversal and export more stuff 2017-09-07 19:04:36 +00:00
ddc8cae4c6 add unperturbed trial with ranking 2017-09-07 19:00:09 +00:00