Merge branch 'v0.7.2-dev-wm' into 'v0.7.2-dev'

[fix]修复开启并行解码后，在极端测试情况下服务无响应问题

See merge request dcutoolkit/deeplearing/vllm!97

vllm/spec_decode/spec_decode_worker.py

@@ -690,6 +690,8 @@ class SpecDecodeWorker(LoraNotSupportedWorkerBase):
                 hidden_states = hidden_states[
                     torch.where(sampler_output.sampled_token_ids -
                                 VLLM_INVALID_TOKEN_ID)[0]]
+            
+            if not skip_proposer:
                 if self.previous_hidden_states is None and len(
                         seq_group_meta_with_hidden):
                     self.previous_hidden_states = HiddenStates(