Change FSDP to truncate dataset

Close #719

Author: mhuguesaws

3.test_cases/pytorch/FSDP/src/model_utils/concat_dataset.py

@@ -28,7 +28,7 @@ def __iter__(self) -> Iterable[Dict[str, bytes]]:
         mask_buffer = []
         for sample in self.hf_dataset:
             encoded = self.tokenizer(sample['text'],
-                                     truncation=False,
+                                     truncation=True,
                                      padding=False)
             iids = encoded['input_ids']
             mask = encoded['attention_mask']