[Text Generation][Fix] Assert that tokens have the same shape on ou…

…tput (#1143)

* initial implementation

* initial commit

* Update src/deepsparse/transformers/engines/nl_decoder_engine.py

* add split_engine_inputs

* reuse already existing softmax

* Revert "reuse already existing softmax"

This reverts commit 8234524.

* Reuse already existing softmax

* initial implementation

src/deepsparse/transformers/pipelines/text_generation.py

@@ -468,9 +468,8 @@ def has_cache(self) -> bool:
         """
         return self.multitoken_engine.kv_cache_enabled
 
-    @staticmethod
     def join_engine_outputs(
-        batch_outputs: List[List[numpy.ndarray]], orig_batch_size: int
+        self, batch_outputs: List[List[numpy.ndarray]], orig_batch_size: int
     ) -> List[numpy.ndarray]:
         """
         Takes a list of outputs (batches) from the engine
@@ -483,7 +482,22 @@ def join_engine_outputs(
         :return: A list of joined outputs
         """
         tokens, logits = zip(*batch_outputs)
+
+        # find the longest sequence in the batch of tokens
+        max_len = max([token.shape[1] for token in tokens])
+
+        # pad all tokens to the same length
+        tokens = [
+            pad_to_fixed_length(
+                array=prediction,
+                max_len=max_len,
+                value=self.tokenizer.pad_token_id,
+                axis=1,
+            )
+            for prediction in tokens
+        ]
         tokens = numpy.concatenate(tokens, axis=0)
+
         # find the longest sequence in the batch of logits
         max_len = max([logits.shape[1] for logits in logits])
         # pad all logits to the same length