Changelog entry, references to the GH PR

CHANGES.md

@@ -13,6 +13,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Changed
 ### Fixed
 
+- Fixed a couple of issues when saving and loading parameters while using accelerate (via `AccelerateMixin`) in a multi-GPU setting (#1008)
+
 ## [0.14.0] - 2023-06-24
 
 ### Added

examples/accelerate-multigpu/README.md

@@ -43,7 +43,7 @@ use_cpu: false
 
 ## `run-save-load.py`
 
-The context of this script is that there were issues with saving and loading when using `AccelerateMixin`. The provided script is to ensure that everything works as expected. Same as the first one, for a proper test, this script needs to run in a multi-GPU setting.
+The context of this script is that there were issues with saving and loading when using `AccelerateMixin`. The provided script is to ensure that everything works as expected. Same as the first one, for a proper test, this script needs to run in a multi-GPU setting. For more information, check PR #1008.
 
 Run the scripts like this:
 

skorch/tests/test_hf.py

@@ -688,6 +688,8 @@ def test_save_load_params(
         # Note that we cannot really test this with unit tests because it
         # requires a multi-GPU setup. To run a proper test, please run the
         # `run-save-load.py` script in skorch/examples/accelerate-multi-gpu/.
+
+        # More context in PR #1008
         accelerator = accelerator_cls()
 
         def get_accelerate_net():