Updates to add TF now that 2.16.0rc0 supports Python 3.12

google-deepmind · Feb 24, 2024 · 62f0baa · 62f0baa
1 parent e2b6eab
commit 62f0baa
Show file tree

Hide file tree

Showing 4 changed files with 12 additions and 13 deletions.
diff --git a/.github/workflows/actions.yml b/.github/workflows/actions.yml
@@ -20,7 +20,8 @@ jobs:
           OPEN_SPIEL_BUILD_WITH_ORTOOLS: "OFF"
           OPEN_SPIEL_BUILD_WITH_ORTOOLS_DOWNLOAD_URL: ""
           # Disable due to not yet available for Python 3.12
-          # Missing deps: tensorflow, qdldl (needed by cvxpy and osqp) 
+          # Still must change versions everything in python_extra_deps.sh
+          # Missing deps: qdldl (needed by cvxpy and osqp) 
           # Note also some tests in python/CMakeLists.txt disabled for this reason.
           OPEN_SPIEL_BUILD_WITH_TENSORFLOW: "OFF"
           OPEN_SPIEL_BUILD_WITH_MISC: "OFF"

diff --git a/open_spiel/python/CMakeLists.txt b/open_spiel/python/CMakeLists.txt
@@ -275,8 +275,7 @@ set(PYTHON_TESTS ${PYTHON_TESTS}
 # Add Jax tests if it is enabled.
 if (OPEN_SPIEL_ENABLE_JAX)
   set (PYTHON_TESTS ${PYTHON_TESTS}
-    # Temporarily disabled: depends on TF and TF doesn't run on Python 3.12.
-    # jax/deep_cfr_jax_test.py
+    jax/deep_cfr_jax_test.py
     jax/dqn_jax_test.py
     jax/nfsp_jax_test.py
     jax/opponent_shaping_jax_test.py
@@ -295,8 +294,7 @@ if (OPEN_SPIEL_ENABLE_PYTORCH)
       pytorch/deep_cfr_pytorch_test.py
       pytorch/eva_pytorch_test.py
       pytorch/losses/rl_losses_pytorch_test.py
-      # Temporarily disabled: depends on TF and TF doesn't run on Python 3.12.
-      # pytorch/policy_gradient_pytorch_test.py
+      pytorch/policy_gradient_pytorch_test.py
       pytorch/ppo_pytorch_test.py
       pytorch/neurd_pytorch_test.py
   )

diff --git a/open_spiel/python/algorithms/rcfr_test.py b/open_spiel/python/algorithms/rcfr_test.py
@@ -20,14 +20,15 @@
 import numpy as np
 # Note: this import needs to come before Tensorflow to fix a malloc error.
 import pyspiel  # pylint: disable=g-bad-import-order
-import tensorflow.compat.v1 as tf
+# import tensorflow.compat.v1 as tf
+import tensorflow as tf
 
 from open_spiel.python.algorithms import rcfr
 
 # Temporarily disable TF2 behavior while the code is not updated.
-tf.disable_v2_behavior()
+#tf.disable_v2_behavior()
 
-tf.enable_eager_execution()
+#tf.enable_eager_execution()
 
 _GAME = pyspiel.load_game('kuhn_poker')
 _BOOLEANS = [False, True]
@@ -46,7 +47,7 @@ class RcfrTest(parameterized.TestCase, tf.test.TestCase):
 
   def setUp(self):
     super(RcfrTest, self).setUp()
-    tf.random.set_random_seed(42)
+    tf.random.set_seed(42)
 
   def test_with_one_hot_action_features_single_state_vector(self):
     information_state_features = [1., 2., 3.]
@@ -476,7 +477,7 @@ def test_rcfr_functions(self):
       data = data.batch(12)
       data = data.repeat(num_epochs)
 
-      optimizer = tf.keras.optimizers.Adam(lr=0.005, amsgrad=True)
+      optimizer = tf.optimizers.Adam(learning_rate=0.005, amsgrad=True)
 
       for x, y in data:
         optimizer.minimize(
@@ -504,7 +505,7 @@ def _train(model, data):
       data = data.batch(12)
       data = data.repeat(num_epochs)
 
-      optimizer = tf.keras.optimizers.Adam(lr=0.005, amsgrad=True)
+      optimizer = tf.optimizers.Adam(learning_rate=0.005, amsgrad=True)
 
       for x, y in data:
         optimizer.minimize(
@@ -565,7 +566,7 @@ def _train(model, data):
       data = data.batch(12)
       data = data.repeat(num_epochs)
 
-      optimizer = tf.keras.optimizers.Adam(lr=0.005, amsgrad=True)
+      optimizer = tf.optimizers.Adam(learning_rate=0.005, amsgrad=True)
 
       for x, y in data:
         optimizer.minimize(

diff --git a/open_spiel/scripts/find_tensorflow.sh b/open_spiel/scripts/find_tensorflow.sh
@@ -16,7 +16,6 @@
 
 read -r -d '' TESTSCRIPT << EOT
 import tensorflow as tf
-import tensorflow_probability
 print(tf.__version__)
 EOT