fix: Fix GPU power utilization logging and error handling

src/neptune/common/hardware/gauges/gpu.py

@@ -66,7 +66,10 @@ def name(self):
         return str(self.card_index)
 
     def value(self):
-        return self.__gpu_monitor.get_card_power_usage(self.card_index) // MILLIWATTS_IN_ONE_WATT
+        power_usage = self.__gpu_monitor.get_card_power_usage(self.card_index)
+        if power_usage is None:
+            return None
+        return self.__gpu_monitor.get_card_power_usage(self.card_index) / MILLIWATTS_IN_ONE_WATT
 
     def __eq__(self, other):
         return self.__class__ == other.__class__ and self.card_index == other.card_index

src/neptune/common/hardware/gpu/gpu_monitor.py

@@ -78,10 +78,10 @@ def __nvml_get_or_else(self, getter, default=None):
         try:
             nvmlInit()
             return getter()
-        except NVMLError:
+        except NVMLError as e:
             if not GPUMonitor.nvml_error_printed:
                 warning = (
-                    "Info (NVML): %s. GPU usage metrics may not be reported. For more information, "
+                    f"Info (NVML): {e}. GPU usage metrics may not be reported. For more information, "
                     "see https://docs.neptune.ai/help/nvml_error/"
                 )
                 warn_once(message=warning, exception=NeptuneWarning)