Group metrics generated by DeviceStatsMonitor for better visualization

CHANGELOG.md

@@ -175,6 +175,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Renamed `training_type_plugin` file to `strategy` ([#11239](https://github.com/PyTorchLightning/pytorch-lightning/pull/11239))
 
 
+- Changed `DeviceStatsMonitor` to group metrics based on the logger's `group_separator` ([]())
+
+
 ### Deprecated
 
 - Deprecated `ClusterEnvironment.master_{address,port}` in favor of `ClusterEnvironment.main_{address,port}` ([#10103](https://github.com/PyTorchLightning/pytorch-lightning/issues/10103))

pytorch_lightning/callbacks/device_stats_monitor.py

@@ -58,7 +58,8 @@ def on_train_batch_start(
             return
 
         device_stats = trainer.accelerator.get_device_stats(pl_module.device)
-        prefixed_device_stats = prefix_metrics_keys(device_stats, "on_train_batch_start")
+        separator = trainer.logger.group_separator
+        prefixed_device_stats = _prefix_metric_keys(device_stats, "on_train_batch_start", separator)
         assert trainer.logger is not None
         trainer.logger.log_metrics(prefixed_device_stats, step=trainer.global_step)
 
@@ -75,10 +76,11 @@ def on_train_batch_end(
             return
 
         device_stats = trainer.accelerator.get_device_stats(pl_module.device)
-        prefixed_device_stats = prefix_metrics_keys(device_stats, "on_train_batch_end")
+        separator = trainer.logger.group_separator
+        prefixed_device_stats = _prefix_metric_keys(device_stats, "on_train_batch_end", separator)
         assert trainer.logger is not None
         trainer.logger.log_metrics(prefixed_device_stats, step=trainer.global_step)
 
 
-def prefix_metrics_keys(metrics_dict: Dict[str, float], prefix: str) -> Dict[str, float]:
-    return {prefix + "." + k: v for k, v in metrics_dict.items()}
+def _prefix_metric_keys(metrics_dict: Dict[str, float], prefix: str, separator: str) -> Dict[str, float]:
+    return {prefix + separator + k: v for k, v in metrics_dict.items()}

tests/callbacks/test_device_stats_monitor.py

@@ -17,6 +17,7 @@
 
 from pytorch_lightning import Trainer
 from pytorch_lightning.callbacks import DeviceStatsMonitor
+from pytorch_lightning.callbacks.device_stats_monitor import _prefix_metric_keys
 from pytorch_lightning.loggers import CSVLogger
 from pytorch_lightning.utilities.distributed import rank_zero_only
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
@@ -128,3 +129,12 @@ def test_device_stats_monitor_no_logger(tmpdir):
 
     with pytest.raises(MisconfigurationException, match="Trainer that has no logger."):
         trainer.fit(model)
+
+
+def test_prefix_metric_keys(tmpdir):
+    """Test that metric key names are converted correctly."""
+    metrics = {"1": 1.0, "2": 2.0, "3": 3.0}
+    prefix = "foo"
+    separator = "."
+    converted_metrics = _prefix_metric_keys(metrics, prefix, separator)
+    assert converted_metrics == {"foo.1": 1.0, "foo.2": 2.0, "foo.3": 3.0}