Update.

[culture.git] / README.txt

diff --git a/README.txt b/README.txt
index 223565e..af96ee9 100644 (file)
--- a/README.txt
+++ b/README.txt
@@ -1,5 +1,54 @@
-For the stack experiment:
 
-./main.py --task=stack --stack_fraction_values_for_train=0.75 --stack_nb_stacks=3
+Trying to make GPTs build their own "culture".
 
-Takes ~1h10min on a 4090.
+* Motivation
+
+The original motivation of this experiment is the hypothesis that
+high-level cognition emerges from the competition among humans in the
+space of language and ideas.
+
+More precisely, communicating agents try to out-do competitors by
+creating stuff that is smart but doable, e.g. some other agents get
+it, but not all. Then, that smart thing is added to the "culture",
+they all learn and get to understand it, and it repeats.
+
+* Setup
+
+It starts with a "world model" that they got before they communicate,
+and from there, they try to "be smart" by proposing quizzes that can
+be solved but not by everybody.
+
+There are 5 competing GPTs.
+
+The "world" is a 6x8 grid with one or two "birds" moving in a straight
+line and bouncing on the world's borders. The colors correspond to a
+fixed "z-buffer order". It could be another "world", but this one has
+objectness, occlusion, and motion.
+
+Given a random world state, and the state after two iterations of
+birds moving, a "quiz" is to predict the second frame, given the
+first, or the opposite.
+
+My home-baked GPT-37M trained with 250k solves this with ~99% success.
+
+At every iteration, we select the GPT with the lowest test accuracy,
+and run one epoch. If its test accuracy got higher than 97.5%, it will
+create new quizzes. To do so, it generates a large number of pairs of
+frames, and checks which ones of these quizzes are hard but not too
+hard, which means
+
+[THIS IS THE IMPORTANT BIT]:
+
+it can be solved, in both time directions, by all the other GPTs **but
+one**
+
+The both time directions is to avoid a simple type of quizzes which is
+simply to deal with noise in the first frame.
+
+The GPT generates 1000 of such quizzes, that are added to the
+"culture", i.e. the training set.
+
+Then training resumes.
+
+The hope is that interesting concepts emerge (connectivity, symmetry,
+interior/exterior, shape vocabulary, etc.)