| 30 | raise ValueError("Data not supported.") |
| 31 | |
| 32 | def load_hotpot_qa(self, cache_dir="data/hotpot_qa", sample_size=100, type="test"): |
| 33 | assert type in ["train", "validation", "test"] |
| 34 | data = datasets.load_dataset('hotpot_qa', 'fullwiki', cache_dir=cache_dir) |
| 35 | df = data[type].to_pandas() |
| 36 | sampled_df = df.sample(sample_size, random_state=self.seed)[["question", "answer"]].reset_index(drop=True) |
| 37 | return sampled_df |
| 38 | |
| 39 | def load_fever(self, cache_dir="data/fever", sample_size=100, type="test"): |
| 40 | assert type in ["train", "validation", "test"] |