Fix logging of nan parameters (#9364)

* Fix logging of nan parameters

CHANGELOG.md

@@ -323,6 +323,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed incorrect main progress bar indicator when resuming training mid-epoch ([#9310](https://github.com/PyTorchLightning/pytorch-lightning/pull/9310))
 
 
+- Fixed logging of nan parameters ([#9364](https://github.com/PyTorchLightning/pytorch-lightning/pull/9364))
+
+
 - Fixed `replace_sampler` missing the batch size under specific conditions ([#9367](https://github.com/PyTorchLightning/pytorch-lightning/pull/9367))
 
 

pytorch_lightning/utilities/finite_checks.py

@@ -25,7 +25,7 @@ def print_nan_gradients(model: nn.Module) -> None:
     """Iterates over model parameters and prints out parameter + gradient information if NaN."""
     for param in model.parameters():
         if (param.grad is not None) and torch.isnan(param.grad.float()).any():
-            log.info(param, param.grad)
+            log.info(f"{param}, {param.grad}")
 
 
 def detect_nan_parameters(model: nn.Module) -> None: