From 36a1440d01cc15643849f5ba421f89ac403ccd82 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Fran=C3=A7ois=20Fleuret?= <francois@fleuret.org>
Date: Sat, 22 Jun 2024 19:24:50 +0200
Subject: [PATCH] Update.

---
 README.txt | 22 +++++++++++++---------
 main.py    |  4 ++--
 tasks.py   | 10 +++++-----
 world.py   |  2 +-
 4 files changed, 21 insertions(+), 17 deletions(-)

diff --git a/README.txt b/README.txt
index af96ee9..1bd8dbc 100644
--- a/README.txt
+++ b/README.txt
@@ -20,22 +20,25 @@ 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
+The "world" is a 7x9 grid with three "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.
+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.
+
+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]:
 
@@ -48,7 +51,8 @@ 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.)
diff --git a/main.py b/main.py
index 09ae823..b6f2783 100755
--- a/main.py
+++ b/main.py
@@ -360,7 +360,7 @@ def create_quizzes(
     task.store_new_quizzes(new_quizzes[nb_for_train:], for_train=False)
 
     task.save_image(
-        new_quizzes[:96],
+        new_quizzes[:72],
         args.result_dir,
         f"world_quiz_{n_epoch:04d}_{model.id:02d}.png",
         log_string,
@@ -404,7 +404,7 @@ if args.check:
     nb_new_quizzes_for_test = 10
 
 for n_epoch in range(args.nb_epochs):
-    a = [(model.id, model.main_test_accuracy) for model in models]
+    a = [(model.id, model.main_test_accuracy.item()) for model in models]
     a.sort(key=lambda p: p[0])
     log_string(f"current accuracies {a}")
 
diff --git a/tasks.py b/tasks.py
index 9a67127..ad95237 100755
--- a/tasks.py
+++ b/tasks.py
@@ -81,7 +81,7 @@ class World(Task):
     def save_image(self, input, result_dir, filename, logger):
         img = world.sample2img(input.to("cpu"), self.height, self.width)
         image_name = os.path.join(result_dir, filename)
-        torchvision.utils.save_image(img.float() / 255.0, image_name, nrow=8, padding=2)
+        torchvision.utils.save_image(img.float() / 255.0, image_name, nrow=6, padding=4)
         logger(f"wrote {image_name}")
 
     def make_ar_mask(self, input):
@@ -101,8 +101,8 @@ class World(Task):
 
         self.batch_size = batch_size
         self.device = device
-        self.height = 6
-        self.width = 8
+        self.height = 7
+        self.width = 9
 
         self.train_input = world.generate(
             nb_train_samples, height=self.height, width=self.width
@@ -126,7 +126,7 @@ class World(Task):
 
         if result_dir is not None:
             self.save_image(
-                self.train_input[:96], result_dir, f"world_train.png", logger
+                self.train_input[:72], result_dir, f"world_train.png", logger
             )
 
     def batches(self, split="train", desc=None):
@@ -222,7 +222,7 @@ class World(Task):
         )
 
         self.save_image(
-            result[:96],
+            result[:72],
             result_dir,
             f"world_prediction_{n_epoch:04d}_{model.id:02d}.png",
             logger,
diff --git a/world.py b/world.py
index 4055533..05d7505 100755
--- a/world.py
+++ b/world.py
@@ -59,7 +59,7 @@ def generate(
     nb,
     height,
     width,
-    nb_birds=2,
+    nb_birds=3,
     nb_iterations=2,
 ):
     pairs = []
-- 
2.39.5