内部错误：Blas GEMV启动失败：m = 4，n = 800 [[{{node PartitionedCall / MatMul}}]] [Op：__ inference_train_97]

Question

错误名称

  

Blas GEMV启动失败：m = 4，n = 800 [Op：MatMul]名称：MatMul /

刚刚安装

• tensorflow 2.0.0
• CUDA 10.1
• Cudnn 7.6.4

代码：

for i in range(100):    
     W, b = train(X_train, Y_train, W, b)
     print('Current Training Loss on iteration', i, ' -- ', loss(Y_train, 
  prediction(X_train, W, b)).numpy())`enter code here`

我无法解决上述内部错误

在蟒蛇提示下，我得到 F tensorflow / core / common_runtime / gpu / gpu_util.cc：342] CPU-> GPU Memcpy失败