X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=README.txt;h=b34b5ea6617210c28a440fc49e414ff274ca7be1;hb=5c5668b0e52e2ae579d49ba8a44fafe2339ad8c0;hp=af96ee9d73173cc05be312e3e4d0d95882a8f8ea;hpb=f87a57354a1e575181e760fdaedbb2c2d5cf9fa0;p=culture.git

diff --git a/README.txt b/README.txt
index af96ee9..b34b5ea 100644
--- a/README.txt
+++ b/README.txt
@@ -1,6 +1,11 @@
 
+[This file may describe an older version than the current code]
+
 Trying to make GPTs build their own "culture".
 
+Francois Fleuret
+Jun 21st, 2024
+
 * Motivation
 
 The original motivation of this experiment is the hypothesis that
@@ -20,27 +25,31 @@ 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.
+The "world" is a 6x8 grid with three "birds" moving in a straight line
+and bouncing on the world's borders. It could be another "world", but
+this one has objectness and motion. There are ten colors and 4
+directions of motions, so roughly (6x8x4x10)**3 ~ 7e9 states.
 
 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.
+first, or the opposite. The starting and ending states are chosen, by
+rejection, so that there is no occlusion.
 
-My home-baked GPT-37M trained with 250k solves this with ~99% success.
+My home-baked GPT-37M trained with 250k solves this with ~99% success
+[to be verified with the new setup].
 
 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
+and run one epoch.
+
+* Creating new quizzes
 
-[THIS IS THE IMPORTANT BIT]:
+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**
+  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.
@@ -48,7 +57,8 @@ 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.
+We update the test accuracy of all the GPTs, and then we go to the
+next iteration.
 
 The hope is that interesting concepts emerge (connectivity, symmetry,
 interior/exterior, shape vocabulary, etc.)