[GPU not working] W0309 21:57:26.429087 805030 analysis_predictor.cc:2166] Deprecated. Please use CreatePredictor instead.

When I use lac to do NER by gpu, all the result is 'a', it seems the lac is not working. The result by cpu is normal.
The input is `"地址：上海市长宁区古北路666号嘉麒大厦602室"`
and the result of cpu is : `[['地址', '：', '上海市', '长宁区', '古北路', '666号', '嘉麒大厦', '602', '室'], ['n', 'w', 'LOC', 'LOC', 'LOC', 'm', 'LOC', 'm', 'n']]`
the result of gpu is `W0309 21:54:09.091929 804304 analysis_predictor.cc:2166] Deprecated. Please use CreatePredictor instead.
[['地址', '：', '上海市', '长宁区', '古北', '路', '666', '号', '嘉', '麒', '大厦', '602', '室'], ['a', 'a', 'a', 'a', 'a', 'a', 'a', 'a', 'a', 'a', 'a', 'a', 'a']]`

My paddle version is "paddlepaddle-gpu:2.4.2" and the LAC version is 2.1.2.
My gpu env is torch10.1 cuda10.2 cudnn7.6.5_0
Why the gpu donot work and how to fix this problem?

No source file or test is named in the report. First reproduce GPU NER with the supplied Chinese input using PaddlePaddle GPU 2.4.2 and LAC 2.1.2, then compare it with the CPU result and inspect the reported predictor warning. Done means GPU inference returns the same valid tokens and entity tags as CPU, rather than all `a` tags.

Written by the indexing model from the issue text.