Properly migrating the evaluate client

fl4health/clients/evaluate_client.py

@@ -14,6 +14,7 @@
 from fl4health.parameter_exchange.full_exchanger import FullParameterExchanger
 from fl4health.parameter_exchange.parameter_exchanger_base import ParameterExchanger
 from fl4health.reporting.base_reporter import BaseReporter
+from fl4health.reporting.report_manager import ReportsManager
 from fl4health.utils.losses import EvaluationLosses, LossMeter, LossMeterType
 from fl4health.utils.metrics import Metric, MetricManager
 from fl4health.utils.random import generate_hash
@@ -46,10 +47,8 @@ def __init__(
         self.initialized = False
 
         # Initialize reporters with client information.
-        self.reporters = [] if reporters is None else list(reporters)
-
-        for r in self.reporters:
-            r.initialize(id=self.client_name)
+        self.reports_manager = ReportsManager(reporters)
+        self.reports_manager.initialize(id=self.client_name)
 
         # This data loader should be instantiated as the one on which to run evaluation
         self.global_loss_meter = LossMeter[EvaluationLosses](loss_meter_type, EvaluationLosses)
@@ -88,8 +87,7 @@ def setup_client(self, config: Config) -> None:
         self.criterion = self.get_criterion(config)
         self.parameter_exchanger = self.get_parameter_exchanger(config)
 
-        for r in self.reporters:
-            r.report({"host_type": "client", "initialized": str(datetime.datetime.now())})
+        self.reports_manager.report({"host_type": "client", "initialized": str(datetime.datetime.now())})
 
         self.initialized = True
 
@@ -117,17 +115,19 @@ def evaluate(self, parameters: NDArrays, config: Config) -> Tuple[float, int, Di
         assert self.local_model or self.global_model
 
         loss, metric_values = self.validate()
-        elapsed = datetime.datetime.now() - start_time
-
-        for r in self.reporters:
-            r.report(
-                {
-                    "eval_metrics": metric_values,
-                    "eval_loss": loss,
-                    "eval_start": str(start_time),
-                    "eval_time_elapsed": str(elapsed),
-                },
-            )
+        end_time = datetime.datetime.now()
+        elapsed = end_time - start_time
+
+        self.reports_manager.report(
+            {
+                "eval_metrics": metric_values,
+                "eval_loss": loss,
+                "eval_start": str(start_time),
+                "eval_time_elapsed": str(elapsed),
+                "eval_end": str(end_time),
+            },
+            0,
+        )
 
         # EvaluateRes should return the loss, number of examples on client, and a dictionary holding metrics
         # calculation results.