FIX: to_dict with nn models #949
Conversation
|
🚀 Deployed on https://deploy-preview-949--etna-docs.netlify.app |
Codecov Report
@@ Coverage Diff @@
## master #949 +/- ##
==========================================
+ Coverage 85.31% 85.38% +0.07%
==========================================
Files 147 148 +1
Lines 7890 7936 +46
==========================================
+ Hits 6731 6776 +45
- Misses 1159 1160 +1
📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
etna/core/mixins.py
Outdated
| @@ -80,6 +93,22 @@ def to_dict(self): | |||
| params["_target_"] = BaseMixin._get_target_from_class(self) | |||
| return params | |||
|
|
|||
| @staticmethod | |||
| def _unsafe_to_dict(value: Any) -> Dict[str, Any]: | |||
| """Collect all information about etna object in dict.""" | |||
There was a problem hiding this comment.
It isn't only about etna object, we use it for callbacks from pytorch_lightning.
|
|
||
| self._init_params = {} | ||
| args_dict = dict( | ||
| zip([arg for arg, param in init_args.items() if param.kind == param.POSITIONAL_OR_KEYWORD], deepcopy_args) |
There was a problem hiding this comment.
Can we meet problems with other kinds? E.g. kwargs parameters: docs.
There was a problem hiding this comment.
They will be in kwargs
There was a problem hiding this comment.
Will it work with POSITIONAL_ONLY somehow?
There was a problem hiding this comment.
I think we should write in documentation about the limitations of this approach.
There was a problem hiding this comment.
I've added comment
Before submitting (must do checklist)
Proposed Changes
Closing issues