[Bugfix] Fix PP for Multi-Step (#8887)

vllm/engine/output_processor/multi_step.py

@@ -97,6 +97,9 @@ class MultiStepOutputProcessor(SequenceGroupOutputProcessor):
         assert len(seqs) == 1, (
             "Beam search not supported in multi-step decoding.")
         seq = seqs[0]
+        seq_id = seq.seq_id
+        assert all(
+            [seq_id == output.samples[0].parent_seq_id for output in outputs])
 
         if is_async:
             # Async case: We process tokens one by one. Here, we know the token