Fix start_token is not correct

src/guidellm/dataset/synthetic.py

@@ -194,19 +194,21 @@ def __iter__(
             }
 
     def _create_prompt(
-        self, prompt_tokens: int, start_index: int, unique_prefix: Optional[int] = None
+        self,
+        prompt_tokens: int,
+        start_index: int,
+        unique_prefix: Optional[int] = None,  # noqa: ARG002
     ) -> list[int]:
         if prompt_tokens <= 0:
             return []
 
         left = start_index
         right = start_index + 4 * prompt_tokens
-        start_tokens = [unique_prefix] if unique_prefix else []
 
         while left < right:
             mid = (left + right) // 2
             test_prompt = self.text_creator.create_text(start_index, mid - start_index)
-            test_tokens = start_tokens + self.processor.encode(test_prompt)
+            test_tokens = self.processor.encode(test_prompt)
 
             if len(test_tokens) == prompt_tokens:
                 return test_tokens
@@ -216,7 +218,7 @@ def _create_prompt(
                 right = mid
 
         final_text = self.text_creator.create_text(start_index, left - start_index)
-        return start_tokens + self.processor.encode(final_text)
+        return self.processor.encode(final_text)
 
 
 class SyntheticDatasetCreator(DatasetCreator):

tests/unit/dataset/test_synthetic.py

@@ -415,7 +415,7 @@ def test_create_prompt_method(
 
         # Test normal case
         result = generator._create_prompt(5, 0, 42)
-        assert result == [42, 1, 2, 3]
+        assert result == [1, 2, 3]
 
         # Test zero tokens
         result = generator._create_prompt(0, 0, 42)