support hellaswag dataset

mlora/tasks/qa_tasks.py

@@ -144,11 +144,42 @@ def loading_data(self,
         return ret
 
 
+class HellaSwag(QuestionAnswerTask):
+    def __init__(self) -> None:
+        super().__init__(["A", "B", "C", "D"])
+
+    def loading_data(self,
+                     tokenizer: Tokenizer,
+                     is_train: bool = True) -> List[DataClass]:
+        data = hf_datasets.load_dataset(
+            "Rowan/hellaswag")["train" if is_train else "validation"]
+        logging.info("Preparing data for HellaSwag")
+        ret: List[DataClass] = []
+        for idx, data_point in enumerate(data):
+            prompt = "Please choose the correct ending to complete the given sentence: " + \
+                f"{data_point['activity_label']}. {data_point['ctx']}\n"
+            for label, text in enumerate(data_point["endings"]):
+                prompt += f" ({self.labels_[label]}) {text}"
+            prompt += "\nAnswer:"
+            if is_train:
+                prompt += " " + self.labels_[int(data_point["label"])]
+                labels = None
+            else:
+                labels = [int(data_point["label"])]
+            tokens = tokenizer.encode(data=prompt)
+            ret.append(DataClass(tokens_=tokens, labels_=labels))
+            if idx % 10000 == 0:
+                logging.info(f"Encode text data: {idx}/{len(data)}")
+
+        return ret
+
+
 def update_task_dict(task_dict):
     task_dict.update({
         "arc-e": ARC("ARC-Easy"),
         "arc-c": ARC("ARC-Challenge"),
         "boolq": Boolq(),
         "obqa": OpenBookQA(),
         "piqa": PIQA(),
+        "hellas": HellaSwag(),
     })