Ahmet Kaan Sever commited on
Commit
04b81d8
·
1 Parent(s): 8a3d32e

Getting 50 from each dataset

Browse files
Files changed (1) hide show
  1. src/deepeval/base_task.py +3 -4
src/deepeval/base_task.py CHANGED
@@ -194,10 +194,9 @@ class BaseTask(ABC):
194
  dataset= load_dataset(self.dataset_repo, token=HF_TOKEN, split="train")
195
  print("Dataset loaded.")
196
 
197
- # Load %25 of each dataset
198
- print("Original dataset size: ", len(dataset))
199
- dataset = dataset.shuffle(seed=42).select(range(int(len(dataset) * 0.25)))
200
- print("Reduced dataset size: ", len(dataset))
201
  end_time = datetime.now()
202
  print(f"Dataset loaded in {(end_time - start_time).seconds} seconds.")
203
  return dataset
 
194
  dataset= load_dataset(self.dataset_repo, token=HF_TOKEN, split="train")
195
  print("Dataset loaded.")
196
 
197
+ # Load 50 from each dataset
198
+ if len(dataset) > 50:
199
+ dataset = dataset.shuffle(seed=42).select(range(50))
 
200
  end_time = datetime.now()
201
  print(f"Dataset loaded in {(end_time - start_time).seconds} seconds.")
202
  return dataset