T5 prompt learning fixes missing from r.11.0 merge (NVIDIA#5075) (NVI…

…DIA#5101)

* Fix special tokens

Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca>

* Fix

Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca>

* Empty

Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca>

Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca>
Co-authored-by: David <amosalla@asu.edu>

Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca>
Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca>
Co-authored-by: David <amosalla@asu.edu>
Co-authored-by: Eric Harper <complex451@gmail.com>
Signed-off-by: 1-800-bad-code <shane.carroll@utsa.edu>

nemo/collections/nlp/models/language_modeling/megatron_t5_prompt_learning_model.py

@@ -511,11 +511,15 @@ def predict_step(self, batch: Any, batch_idx: int, dataloader_idx: int = 0) -> A
                 idx = pred.index(self.tokenizer.eos_id)
                 pred = pred[:idx]
 
+            special_token_ids = (
+                self.tokenizer.special_token_to_id.values()
+                if hasattr(self.tokenizer, 'special_token_to_id')
+                else self.tokenizer.tokenizer.additional_special_tokens_ids
+            )
             pred = [
                 id
                 for id in pred
-                if id not in self.tokenizer.tokenizer.additional_special_tokens_ids
-                and id not in self.tokenizer.text_to_ids(T5Sentinel.FIRST.value)
+                if id not in special_token_ids and id not in self.tokenizer.text_to_ids(T5Sentinel.FIRST.value)
             ]  # delete the sentinel token at the beginning of prediction
 
             pred = self.tokenizer.ids_to_text(pred)
@@ -532,8 +536,7 @@ def predict_step(self, batch: Any, batch_idx: int, dataloader_idx: int = 0) -> A
                 label = [
                     id
                     for id in label
-                    if id not in self.tokenizer.tokenizer.additional_special_tokens_ids
-                    and id not in self.tokenizer.text_to_ids(T5Sentinel.FIRST.value)
+                    if id not in special_token_ids and id not in self.tokenizer.text_to_ids(T5Sentinel.FIRST.value)
                 ]  # delete the sentinel token at the beginning of label
 
                 label = self.tokenizer.ids_to_text(label)