tensorflow gpu出现如下错误怎么解决？

Question

文件“train.py”，第 50 行，在 model.fit(x=[H_input, H, SNR], y=H, batch_size=256, epochs=50000, verbose=2, validation_split=0.1, callbacks=[reduce_lr, checkpoint]) 文件“/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training.py”，第 66 行，_method_wrapper 返回方法（self，*args，**kwargs）文件“/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training.py”，第 848 行，适合 tmp_logs = train_function(iterator) 文件“/usr/local/lib/ python3.6/dist-packages/tensorflow/python/eager/def_function.py”，第 580 行，调用结果 = self._call(*args, **kwds) 文件“/usr/local/lib/python3.6/ dist-packages/tensorflow/python/eager/def_function.py”，第 644 行，在 _call return self._stateless_fn(*args, **kwds) 文件“/usr/local/lib/python3.6/dist-packages/tensorflow /python/eager/function.py"，第 2420 行，在调用中返回 graph_function._filtered_call(args, kwargs) # pylint: disable=protected-access File "/usr/local/lib/python3.6/dist -packages/tensorflow/python/eager/function.py”，第 1665 行，在 _filtered_call self.captured_inputs) 文件“/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py” ，第 1746 行，在 _call_flat ctx，args，cancellation_manager=cancellation_manager)) 文件“/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py”，第 598 行，在调用 ctx= ctx) 文件“/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/execute.py”，第 60 行，在 quick_execute 输入、attrs、num_outputs) tensorflow.python.guArmentError.errors ：In[0] 和 In[1] 必须具有兼容的批次尺寸：[9,1,64] vs. [64,9,1] [[node model/lambda_1/MatMul（定义在 /content/drive/My Drive /BF-design-with-DL-master/BF-design-with-DL-master/utils.py:28) ]] [Op:__inference_train_function_1697]

Function 调用堆栈：train_function

Answer 1

看起来您在某处以错误的顺序创建列表must have compatible batch dimensions: [9,1,64] vs. [64,9,1]