Update lib versions and remove warnings from CI

docs/changelog.rst

@@ -7,6 +7,12 @@ Changelog
 Dev version
 -----------
 
+
+
+ *PR #476*
+
+* Update lib versions and remove warnings from tests/CI : https://github.com/rlberry-py/rlberry/issues/471
+
  *PR #474*
 
 * Create a new tool to load data from tensorboard logs : https://github.com/rlberry-py/rlberry/issues/472

pyproject.toml

@@ -26,7 +26,7 @@ pygame-ce = "*"
 matplotlib = "*"
 gymnasium = {version="^0.29.1",  extras=["atari", "accept-rom-license"]}
 dill = "*"
-docopt = "*"
+docopt-ng = "*"
 pyyaml = "*"
 tqdm = "*"
 adastop = "*"
@@ -36,7 +36,7 @@ optuna ={version="*", optional=true}
 ffmpeg-python = {version="*", optional=true}
 opencv-python = {version="*", optional=true}
 ale-py = {version="*", optional=true}
-stable-baselines3 = {version="*", optional=true}
+stable-baselines3 = {version=">=2.3", optional=true}
 tensorboard = {version="*", optional=true}
 torch = [
     { version = ">=2.0.0,<2.3", platform = "darwin", optional=true },

rlberry/manager/env_tools.py

@@ -23,7 +23,7 @@ def __func_to_script(func):
     name = fun_source.split("\n")[1]  # skip decorator
 
     m = re.search(
-        "(?<= )\w+", name
+        r"(?<= )\w+", name
     )  # isolate the name of function to use as script name
 
     source = "\n"

rlberry/manager/evaluation.py

@@ -123,7 +123,7 @@ def evaluate_agents(
     # plot
     if plot:
         plt.figure(fignum)
-        plt.boxplot(output.values, labels=output.columns)
+        plt.boxplot(output.values, tick_labels=output.columns)
         plt.xlabel("agent")
         plt.ylabel("evaluation output")
         if show:

rlberry/manager/experiment_manager.py

@@ -647,7 +647,9 @@ def eval_agents(
     def clear_output_dir(self):
         """Delete output_dir and all its data."""
         if self.optuna_study:
-            optuna.delete_study(self.optuna_study.study_name, self.optuna_storage_url)
+            optuna.delete_study(
+                study_name=self.optuna_study.study_name, storage=self.optuna_storage_url
+            )
         try:
             shutil.rmtree(self.output_dir_)
         except FileNotFoundError:

rlberry/manager/plotting.py

@@ -332,7 +332,9 @@
         # with cross validation bandwidth selection
         if not isinstance(smoothing_bandwidth, numbers.Number):
             if smoothing_bandwidth is None:
-                bandwidth = np.linspace(min_bandwidth_x, (max_x - min_x) / 100, 10)
+                bandwidth = np.linspace(
+                    min_bandwidth_x, max((max_x - min_x) / 100, min_bandwidth_x * 3), 10
+                )
             else:
                 bandwidth = smoothing_bandwidth
             nw = SmoothingParameterSearch(
@@ -358,7 +360,7 @@
             except:
                 raise ValueError("non-finite (or non float) data detected.")
             if not np.all(np.isfinite(X)):
-                logger.warn(
+                logger.warning(
                     "Some of the values are not finite. Not plotting the associated curves."
                 )
                 Xhat[f] = np.nan
@@ -419,7 +421,7 @@
 
             elif error_representation == "cb":
                 if n_tot_simu < 1 / (1 - level):
-                    logger.warn(
+                    logger.warning(
                         "Computing a cb that cannot achieve the level prescribed because there are not enough seeds."
                     )
 
@@ -445,7 +447,7 @@
                     )
                 else:
                     y_err = np.zeros(len(xplot))
-                    logger.warn(
+                    logger.warning(
                         "The variance of the curve was 0, the confidence bound is very biased"
                     )
 
@@ -546,7 +548,7 @@
                 float
             )
             if len(x_simu) != len(x_simu_0):
-                logger.warn("x axis is not the same for all the runs, truncating.")
+                logger.warning("x axis is not the same for all the runs, truncating.")
             x_simu_0 = np.intersect1d(x_simu_0, x_simu)
         df_name = df_name.loc[df_name[xlabel].apply(lambda x: x in x_simu_0)]
         assert (

rlberry/manager/tests/test_comparisons.py

@@ -6,12 +6,18 @@
 import pandas as pd
 
 
+it = 1
+
+
 class DummyAgent(AgentWithSimplePolicy):
     def __init__(self, env, eval_val=0, **kwargs):
+        global it
         AgentWithSimplePolicy.__init__(self, env, **kwargs)
         self.name = "DummyAgent"
         self.fitted = False
-        self.eval_val = eval_val
+        # self.eval_val = eval_val
+        self.eval_val = it
+        it += 1
 
         self.total_budget = 0.0
 
@@ -69,7 +75,7 @@ def test_compare(method, source):
         data_source = pd.DataFrame(
             {
                 "agent": (["Agent 1"] * 10) + (["Agent 2"] * 10),
-                "mean_eval": ([0] * 10) + ([10] * 10),
+                "mean_eval": range(0, 20),
             }
         )
 

rlberry/manager/tests/test_experiment_manager.py

@@ -534,7 +534,7 @@ def test_not_pickle(compress, agent):
 
 
 def test_fitbudget_exception():
-    msg = "\[ExperimentManager\] fit_budget missing in __init__\(\)\."  # /!\ regex : need to escape some char.
+    msg = r"\[ExperimentManager\] fit_budget missing in __init__\(\)\."  # /!\ regex : need to escape some char.
     with pytest.raises(ValueError, match=msg):
         # Define train and evaluation envs
         train_env = (GridWorld, {})

rlberry/utils/torch.py

@@ -78,7 +78,7 @@ def get_memory(pid=None):
         command, stdout=PIPE, stderr=PIPE, universal_newlines=True, shell=True
     ).stdout
     m = re.findall(
-        "\| *[0-9] *" + str(pid) + " *C *.*python.*? +([0-9]+).*\|",
+        r"\| *[0-9] *" + str(pid) + r" *C *.*python.*? +([0-9]+).*\|",
         result,
         re.MULTILINE,
     )