27 tensorflow-probability,
32 bsuite = buildPythonPackage rec {
37 disabled = pythonOlder "3.7";
40 inherit pname version;
41 hash = "sha256-ak9McvXl7Nz5toUaPaRaJek9lurxiQiIW209GnZEjX0=";
45 # Convert np.int -> np.int32 since np.int is deprecated, https://github.com/google-deepmind/bsuite/pull/48
47 url = "https://github.com/google-deepmind/bsuite/pull/48/commits/f8d81b2f1c27ef2c8c71ae286001ed879ea306ab.patch";
48 hash = "sha256-FXtvVS+U8brulq8Z27+yWIimB+kigGiUOIv1SHb1TA8=";
50 # Replace imp with importlib, https://github.com/google-deepmind/bsuite/pull/50
52 name = "replace-imp.patch";
53 url = "https://github.com/google-deepmind/bsuite/commit/d08b63655c7efa5b5bb0f35e825e17549d23e812.patch";
54 hash = "sha256-V5p/6edNXTpEckuSuxJ/mvfJng5yE/pfeMoYbvlNpEo=";
58 build-system = [ setuptools ];
84 tensorflow-probability
87 pythonImportsCheck = [ "bsuite" ];
90 # Tests require network connection
92 "test_longer_action_sequence"
94 "test_step_after_reset"
95 "test_step_on_fresh_environment"
96 "test_longer_action_sequence"
98 "test_step_after_reset"
99 "test_step_on_fresh_environment"
100 "test_longer_action_sequence"
102 "test_step_after_reset"
103 "test_step_on_fresh_environment"
105 "test_episode_truncation"
108 # Escape infinite recursion with rlax
112 check = bsuite.overridePythonAttrs (_: {
118 description = "Collection of experiments that investigate core capabilities of a reinforcement learning (RL) agent";
119 homepage = "https://github.com/deepmind/bsuite";
120 changelog = "https://github.com/google-deepmind/bsuite/releases/tag/${version}";
121 license = licenses.asl20;
122 maintainers = with maintainers; [ onny ];