vllm部署baichuan2-7b awq w4a16量化后的模型报错。

DCU 加速卡驱动程序：rock-5.7.1-6.2.18-V1.1.2.aio.run
DCU 加速卡软件工具包：DTK-24.04.2

加速卡： K100_AI

vllm:0.5.0+das.opt1.dtk24042
torch: 2.1.0+das.opt1.dtk24042
其他软件根据此依赖。环境部署qwen1.5 qwen2，以及其awq量化后的模型都可以。部署baichuan2-7b基础模型也可以。但部署baichuan2-7b awq量化后的模型则不行，报错信息为：
File "/home/xxx/.conda/envs/vllm/lib/python3.10/site-packages/vllm/model_executor/models/baichuan.py", line 248, in forward
hidden_states = self.self_attn(
File "/home/xxx/.conda/envs/vllm/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/xxx/.conda/envs/vllm/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "/home/xxx/.conda/envs/vllm/lib/python3.10/site-packages/vllm/model_executor/models/baichuan.py", line 196, in forward
attn_output = self.attn(q, k, v, kv_cache, attn_metadata)
File "/home/xxx/.conda/envs/vllm/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/xxx/.conda/envs/vllm/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "/home/xxx/.conda/envs/vllm/lib/python3.10/site-packages/vllm/attention/layer.py", line 89, in forward
return self.impl.forward(query, key, value, kv_cache, attn_metadata,
File "/home/xxx/.conda/envs/vllm/lib/python3.10/site-packages/vllm/attention/backends/rocm_flash_attn.py", line 297, in forward
query = query.view(-1, self.num_heads, self.head_size)
RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.