small fix

absa_evaluator.py

@@ -1,4 +1,4 @@
-from typing import Dict, List
+from typing import Dict, List, Set
 
 import evaluate
 from datasets import Features, Sequence, Value
@@ -172,14 +172,15 @@ class AbsaEvaluator(evaluate.Metric):
         }
         
 def adjust_predictions(
-    refs: List[List[Any]], preds: List[List[Any]], choices: List[Any]
+    refs: List[List[Any]], preds: List[List[Any]], choices: Set[Any]
 ) -> List[List[Any]]:
     """Adjust predictions to match the length of references with either a special token or random choice."""
+    choices_list = list(choices)
     adjusted_preds = []
     for ref, pred in zip(refs, preds):
         if len(pred) < len(ref):
             missing_count = len(ref) - len(pred)
-            pred.extend([choice(choices) for _ in range(missing_count)])
+            pred.extend([choice(choices_list) for _ in range(missing_count)])
         elif len(pred) > len(ref):
             pred = pred[:len(ref)]
         adjusted_preds.append(pred)
@@ -221,7 +222,7 @@ def absa_term_preprocess(
 
     # Define adjustment parameters
     special_token = "NONE"  # For missing aspect terms
-    sentiment_choices = unique_strings(flatten_list(truth_polarities))    
+    sentiment_choices = set(flatten_list(truth_polarities))    
 
     # Adjust the predictions to match the length of references
     adjusted_pred_terms = adjust_predictions(
@@ -241,22 +242,4 @@ def absa_term_preprocess(
 
 def flatten_list(nested_list):
     """Flatten a nested list into a single-level list."""
-    return list(chain.from_iterable(nested_list))
-
-def unique_strings(strings: List[str]) -> List[str]:
-    """
-    Convert a list of strings to a list of unique strings, preserving the original order.
-
-    Args:
-        strings (List[str]): The input list of strings.
-
-    Returns:
-        List[str]: A list of unique strings in the order of their first occurrence.
-    """
-    seen = set()
-    unique_list = []
-    for string in strings:
-        if string not in seen:
-            seen.add(string)
-            unique_list.append(string)
-    return unique_list
+    return list(chain.from_iterable(nested_list))