[W AmpForeachNonFiniteCheckAndUnscaleKernelNpu.cpp:46] Warning: Non finite check and unscale on NPU device! (function operator()) 

Traceback (most recent call last): 

  File "/home/ma-user/work/test.py", line 68, in <module> 

    scaler.step(optimizer)    # 更新参数（自动unscaling） 

  File "/home/ma-user/anaconda3/lib/python3.9/site-packages/torch_npu/npu/amp/grad_scaler.py", line 402, in step 

    retval = self._maybe_opt_step(optimizer, optimizer_state, *args, **kwargs) 

  File "/home/ma-user/anaconda3/lib/python3.9/site-packages/torch_npu/npu/amp/grad_scaler.py", line 345, in _maybe_opt_step 

    retval = optimizer.step(*args, **kwargs) 

  File "/home/ma-user/anaconda3/lib/python3.9/site-packages/torch/optim/optimizer.py", line 280, in wrapper 

    out = func(*args, **kwargs) 

  File "/home/ma-user/anaconda3/lib/python3.9/site-packages/torch/optim/optimizer.py", line 33, in _use_grad 

    ret = func(self, *args, **kwargs) 

  File "/home/ma-user/anaconda3/lib/python3.9/site-packages/torch/optim/sgd.py", line 76, in step 

    sgd(params_with_grad, 

  File "/home/ma-user/anaconda3/lib/python3.9/site-packages/torch/optim/sgd.py", line 222, in sgd 

    func(params, 

  File "/home/ma-user/anaconda3/lib/python3.9/site-packages/torch/optim/sgd.py", line 325, in _multi_tensor_sgd 

    torch._foreach_add_(device_params, device_grads, alpha=-lr) 

RuntimeError: call aclnnForeachAddList failed, detail:EZ1001: Get path and read config failed. 

        TraceBack (most recent call last): 

        Check NnopbaseCollecterWork(binCollecter.get()) failed 

        Assert ((NnopbaseInit()) == 0) failed 

        Check NnopbaseCreateExecutorSpace(&executorSpace) failed