Apply suggestions from code review

docs/en/tutorials/hook.md

@@ -123,7 +123,7 @@ The four features mentioned above are described below.
 
 - Make checkpoints for publish
 
-  If you want to automatically generate publishable checkpoints after training (remove unnecessary keys, such as optimizer state), you can set the `published_keys` parameter to choose which information to keep. Note: You need to set the `save_best` or `save_last` parameters accordingly, so that the releasable checkpoints will be generated. Setting `save_best` will generate the releasable weights of the optimal checkpoint, and setting `save_last` will generate the releasable final checkpoint. These two parameters can also be used set at the same time.
+  If you want to automatically generate publishable checkpoints after training (remove unnecessary keys, such as optimizer state), you can set the `published_keys` parameter to choose which information to keep. Note: You need to set the `save_best` or `save_last` parameters accordingly so that the releasable checkpoints will be generated. Setting `save_best` will generate the releasable weights of the optimal checkpoint, and setting `save_last` will generate the releasable final checkpoint. These two parameters can also be set at the same time.
 
   ```python
   default_hooks = dict(checkpoint=dict(type='CheckpointHook', interval=1, save_best='accuracy', rule='less', published_keys=['meta', 'state_dict']))

mmengine/hooks/checkpoint_hook.py

@@ -90,6 +90,7 @@ class CheckpointHook(Hook):
             or ``save_best`` is not ``None``, it will automatically
             publish model with keys in the list after training.
             Defaults to None.
+            `New in version 0.7.1.`
     Examples:
         >>> # Save best based on single metric
         >>> CheckpointHook(interval=2, by_epoch=True, save_best='acc',
@@ -370,10 +371,10 @@ def _publish_model(self, runner, ckpt_path: str) -> None:
                 logger='current')
         checkpoint_data = pickle.dumps(checkpoint)
         sha = hashlib.sha256(checkpoint_data).hexdigest()
-        final_path = osp.splitext(ckpt_path)[0] + f'-published-{sha[:8]}.pth'
+        final_path = osp.splitext(ckpt_path)[0] + f'-{sha[:8]}.pth'
         save_checkpoint(checkpoint, final_path)
         print_log(
-            f'The published model ({ckpt_path}) is saved at '
+            f'The checkpoint ({ckpt_path}) is published to '
             f'{final_path}.',
             logger='current')