Fix MLM datasets length

bigscience-workshop · Jun 27, 2022 · 9da7588 · 9da7588
1 parent c71b379
commit 9da7588
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/megatron/data/mlm_dataset.py b/megatron/data/mlm_dataset.py
@@ -335,7 +335,7 @@ def __init__(
             assert self.inputs_length + self.targets_length == args.seq_length
 
     def __len__(self):
-        return len(self.samples_mapping)
+        return len(self._gpt_dataset)
 
     def __getitem__(self, idx):
         if isinstance(idx, slice):