@@ -45,7 +45,7 @@ def get_metric(self, prompt: str, ground_truth: str, model_prediction: str) -> U
4545 union_size = len (set_ground_truth .union (set_model_prediction ))
4646
4747 similarity = intersection_size / union_size if union_size != 0 else 0
48- return similarity
48+ return float ( similarity )
4949
5050
5151@QaTestRegistry .register ("dot_product" )
@@ -64,7 +64,7 @@ def get_metric(self, prompt: str, ground_truth: str, model_prediction: str) -> U
6464 embedding_ground_truth = self ._encode_sentence (ground_truth )
6565 embedding_model_prediction = self ._encode_sentence (model_prediction )
6666 dot_product_similarity = np .dot (embedding_ground_truth , embedding_model_prediction )
67- return dot_product_similarity
67+ return float ( dot_product_similarity )
6868
6969
7070@QaTestRegistry .register ("rouge_score" )
@@ -100,10 +100,9 @@ def get_metric(self, prompt: str, ground_truth: str, model_prediction: str) -> U
100100
101101 common_words = words_model_prediction .intersection (words_ground_truth )
102102 overlap_percentage = (len (common_words ) / len (words_ground_truth )) * 100
103- return overlap_percentage
103+ return float ( overlap_percentage )
104104
105105
106- @QaTestRegistry .register ("verb_percent" )
107106class PosCompositionTest (LLMQaTest ):
108107 def _get_pos_percent (self , text : str , pos_tags : List [str ]) -> float :
109108 words = word_tokenize (text )
0 commit comments