Change StatsSummary to use properties

ml-agents/mlagents/trainers/stats.py

 
 
 class StatsSummary(NamedTuple):
-    mean: float
-    std: float
-    num: int
-    sum: float
-        return StatsSummary(0.0, 0.0, 0, 0.0, [0.0], StatsAggregationMethod.AVERAGE)
+        return StatsSummary([0.0], StatsAggregationMethod.AVERAGE)
 
     @property
     def aggregated_value(self):
             return self.mean
+
+    @property
+    def mean(self):
+        return np.mean(self.full_dist)
+
+    @property
+    def std(self):
+        return np.std(self.full_dist)
+
+    @property
+    def num(self):
+        return len(self.full_dist)
+
+    @property
+    def sum(self):
+        return np.sum(self.full_dist)
 
 
 class StatsPropertyType(Enum):
             return StatsSummary.empty()
 
         return StatsSummary(
-            mean=np.mean(stat_values),
-            std=np.std(stat_values),
-            num=len(stat_values),
-            sum=np.sum(stat_values),
             full_dist=stat_values,
             aggregation_method=StatsReporter.stats_aggregation[self.category][key],
         )

ml-agents/mlagents/trainers/tests/test_agent_processor.py

 
     expected_stats = {
         "averaged": StatsSummary(
-            mean=2.0,
-            std=mock.ANY,
-            num=2,
-            sum=4.0,
-            full_dist=mock.ANY,
-            aggregation_method=StatsAggregationMethod.AVERAGE,
+            full_dist=[1.0, 3.0], aggregation_method=StatsAggregationMethod.AVERAGE
-            mean=4.0,
-            std=0.0,
-            num=1,
-            sum=4.0,
-            full_dist=mock.ANY,
-            aggregation_method=StatsAggregationMethod.MOST_RECENT,
+            full_dist=[4.0], aggregation_method=StatsAggregationMethod.MOST_RECENT
-            mean=2.1,
-            std=mock.ANY,
-            num=2,
-            sum=4.2,
-            full_dist=mock.ANY,
-            aggregation_method=StatsAggregationMethod.SUM,
+            full_dist=[3.1, 1.1], aggregation_method=StatsAggregationMethod.SUM
         ),
     }
     stats_reporter.write_stats(123)

ml-agents/mlagents/trainers/tests/test_stats.py

     with tempfile.TemporaryDirectory(prefix="unittest-") as base_dir:
         tb_writer = TensorboardWriter(base_dir, clear_past_data=False)
         statssummary1 = StatsSummary(
-            mean=1.0,
-            std=1.0,
-            num=1,
-            sum=1.0,
-            full_dist=[0.0],
-            aggregation_method=StatsAggregationMethod.AVERAGE,
+            full_dist=[1.0], aggregation_method=StatsAggregationMethod.AVERAGE
         )
         tb_writer.write_stats("category1", {"key1": statssummary1}, 10)
 
 def test_tensorboard_writer_clear(tmp_path):
     tb_writer = TensorboardWriter(tmp_path, clear_past_data=False)
     statssummary1 = StatsSummary(
-        mean=1.0,
-        std=1.0,
-        num=1,
-        sum=1.0,
-        full_dist=[0.0],
-        aggregation_method=StatsAggregationMethod.AVERAGE,
+        full_dist=[1.0], aggregation_method=StatsAggregationMethod.AVERAGE
     )
     tb_writer.write_stats("category1", {"key1": statssummary1}, 10)
     # TB has some sort of timeout before making a new file
             category = "category1"
             console_writer = ConsoleWriter()
             statssummary1 = StatsSummary(
-                mean=1.0,
-                std=1.0,
-                num=1,
-                sum=1.0,
-                full_dist=[1.0],
-                aggregation_method=StatsAggregationMethod.AVERAGE,
+                full_dist=[1.0], aggregation_method=StatsAggregationMethod.AVERAGE
             )
             console_writer.write_stats(
                 category,
                 10,
             )
             statssummary2 = StatsSummary(
-                mean=0.0,
-                std=0.0,
-                num=1,
-                sum=0.0,
-                full_dist=[0.0],
-                aggregation_method=StatsAggregationMethod.AVERAGE,
+                full_dist=[0.0], aggregation_method=StatsAggregationMethod.AVERAGE
             )
             console_writer.write_stats(
                 category,
             )
 
         self.assertIn(
-            "Mean Reward: 1.000. Std of Reward: 1.000. Training.", cm.output[0]
+            "Mean Reward: 1.000. Std of Reward: 0.000. Training.", cm.output[0]
         )
         self.assertIn("Not Training.", cm.output[1])
 
             console_writer = ConsoleWriter()
             console_writer.add_property(category, StatsPropertyType.SELF_PLAY, True)
             statssummary1 = StatsSummary(
-                mean=1.0,
-                std=1.0,
-                num=1,
-                sum=1.0,
-                full_dist=[1.0],
-                aggregation_method=StatsAggregationMethod.AVERAGE,
+                full_dist=[1.0], aggregation_method=StatsAggregationMethod.AVERAGE
             )
             console_writer.write_stats(
                 category,
             )
 
         self.assertIn(
-            "Mean Reward: 1.000. Std of Reward: 1.000. Training.", cm.output[0]
+            "Mean Reward: 1.000. Std of Reward: 0.000. Training.", cm.output[0]
         )