Commit graph

36 commits

Author SHA1 Message Date
d49eb79475 tests: remove actor:stop - obsolete for restartable actors 2025-06-17 09:48:03 +02:00
b14c01f4e9 explicit become also in test: probe neuron 2025-06-16 10:54:27 +02:00
06aa41de77 actor now needs explicit become when changing behavior (state) 2025-06-16 10:40:36 +02:00
f86428a2c7 rename csys to asys (asynchronous actor systems 2025-06-11 13:40:10 +02:00
fbf735c0f2 work in progress: real decons/csys stuff 2025-06-09 23:31:59 +02:00
871c5ef5af prepare for additional functionality; step 1: simple numeric messages 2025-06-06 20:36:41 +02:00
b86ea21e3c provide an easy way to produce an updated message in a synapse via set-content 2025-06-05 15:30:17 +02:00
2fd468e109 forward messages via and additional neuron 2025-06-05 09:03:33 +02:00
be91b6a446 neuron: with additional 'env' argument; basic functionality for testing probe OK 2025-06-04 21:40:28 +02:00
0865fa9e65 start with 'testing' neuron 2025-06-04 21:05:59 +02:00
5cc56955a3 work in progress: csys - set up tests, basic functionality§ 2025-06-04 14:40:17 +02:00
7c557a2db4 rename test-decons package; start new csys stuff 2025-06-04 10:57:05 +02:00
a70a1487eb move functionality to new files / packages recurse and mlx 2025-05-31 11:48:52 +02:00
f08545db8e provide rreduce-1 that reduces innermost level only 2025-05-29 09:10:28 +02:00
0617177d72 start experimenting with quadratic functions 2025-05-28 11:34:12 +02:00
0768949897 gradient-descent working with line 2025-05-28 09:36:22 +02:00
4b20e4ef0c nabla-xp: improvements, + test 2025-05-28 08:35:14 +02:00
11e8ff59e4 gradient calculation basically working 2025-05-27 15:33:12 +02:00
7498dce57a work in progess: experimental nabla 2025-05-27 14:20:34 +02:00
8025938b81 trying to calculate the loss gradient when varying parameters 2025-05-26 16:11:54 +02:00
19ca8adbfc move lgx (debugging logger) to cl-scopes/util 2025-05-26 09:04:31 +02:00
fc83700642 clean-up: code improvements, remove obsolete stuff 2025-05-25 20:19:11 +02:00
cff6311103 provide a trial class (together with a list of trials) to collect loss values for varied parameter sets 2025-05-25 17:49:06 +02:00
52d1381aba move trivial exploration stuff (including tensor operations) to new file xplore.lisp 2025-05-25 15:55:09 +02:00
3c3539b34b improvements and refinements - provide separate functions for deviation and normalization 2025-05-25 09:37:49 +02:00
25197a0520 now with function l2-loss 2025-05-24 19:29:45 +02:00
0172a2fd46 combine lists (like vectors) and calculate sum of pairwise differences (= 'cost') 2025-05-24 17:14:03 +02:00
b4a337ee37 minor fixes; add rreduce: recursive reduce 2025-05-24 09:43:47 +02:00
fa683982b7 rapply stuff OK, with rcall and rcurry 2025-05-24 07:35:09 +02:00
8ae0ab9271 improvement, start fixing - should work with more than 2 params 2025-05-23 16:52:37 +02:00
c78fe13247 more general solution with rlambda 2025-05-23 13:48:35 +02:00
2ce2ca18f6 rapply basically working 2025-05-23 10:24:29 +02:00
d3417cf94a first try with recursive operation: traverse; start rapply + rcall 2025-05-23 09:33:14 +02:00
049c352ea8 provide tensor function, just creating arrays 2025-05-22 11:49:24 +02:00
580d9f34df more tests, start working with data sets and parameter sets 2025-05-22 10:28:28 +02:00
dcc401b6e9 initial import 2025-05-22 09:11:54 +02:00