🐛 add as a torch tensor?

Signed-off-by: Prashant Gupta <prashantgupta@us.ibm.com>

Author: prashantgupta24

vllm_spyre/v1/worker/spyre_model_runner.py

@@ -879,7 +879,9 @@ def _prepare_decode(
             slot = [start_slot + offset]
             slot_mapping.append(slot)
             output_token_ids = req_state.output_token_ids
-            generation_token = output_token_ids[-1]
+            generation_token = torch.tensor(
+                output_token_ids[-1], dtype=torch.long, device=self.device
+            )
             input_tokens.append([generation_token])
             seq_len = cached_request_data.num_computed_tokens[
                 cached_reqs_map[req_id]]