Merge branch 'v0.9.1-dev-wm' into 'v0.9.1-dev'

[fix]修复091 v0 engine mtp报错问题

See merge request dcutoolkit/deeplearing/vllm!155

vllm/spec_decode/spec_decode_worker.py

@@ -749,7 +749,7 @@ class SpecDecodeWorker(LoRANotSupportedWorkerBase):
                     seq_group_meta_with_hidden):
                 self.previous_hidden_states.update(hidden_states,
                                                    seq_group_meta_with_hidden)
-                self.previous_hidden_states.prune(seq_group_meta_with_hidden)
+                #self.previous_hidden_states.prune(seq_group_meta_with_hidden)
              
             # Store logits from target model execution.
             if self.tree_decoding: