Hi, I tried to run the following script by combining two metrics, but the error occurred:
Python 3.9.16 (main, Apr 5 2023, 13:58:15)
Type 'copyright', 'credits' or 'license' for more information
IPython 8.18.1 -- An enhanced Interactive Python. Type '?' for help.
In [1]: import evaluate
In [2]: evaluations = ("accuracy", "f1", "precision", "recall")
In [3]: clf_metrics = evaluate.combine(evaluations=evaluations)
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
Cell In[3], line 1
----> 1 clf_metrics = evaluate.combine(evaluations=evaluations)
File ~/.pyenv/versions/3.9.16/envs/my-dev/lib/python3.9/site-packages/evaluate/module.py:1034, in combine(evaluations, force_prefix)
1008 def combine(evaluations, force_prefix=False):
1009 """Combines several metrics, comparisons, or measurements into a single `CombinedEvaluations` object that
1010 can be used like a single evaluation module.
1011
(...)
1031 ```
1032 """
-> 1034 return CombinedEvaluations(evaluations, force_prefix=force_prefix)
File ~/.pyenv/versions/3.9.16/envs/my-dev/lib/python3.9/site-packages/evaluate/module.py:884, in CombinedEvaluations.__init__(self, evaluation_modules, force_prefix)
881 self.evaluation_module_names = list(evaluation_modules.keys())
882 loaded_modules = []
--> 884 for module in self.evaluation_modules:
885 if isinstance(module, str):
886 module = load(module)
AttributeError: 'CombinedEvaluations' object has no attribute 'evaluation_modules'
If we use a list format for evaluations, it works fine, but I think we should also support tuple format, which work in the same way as list. This can be easily achieved by modifying the following section.
Hi, I tried to run the following script by combining two metrics, but the error occurred:
If we use a list format for
evaluations
, it works fine, but I think we should also support tuple format, which work in the same way as list. This can be easily achieved by modifying the following section.https://github.com/huggingface/evaluate/blob/main/src/evaluate/module.py#L877
May I create a PR?