以下是发生错误的代码
history = model.fit(x=X_train, y=y_train, batch_size = 100, epochs=400,
verbose=2, validation_data=(X_test, y_test),
shuffle=True, initial_epoch=0)
错误日志如下：
InvalidArgumentError Traceback (most recent call last)
e:\Git_repository\bear_fault_diagnosis\cnn_lstm_model.ipynb 单元格 11 line 1
----> 1 history = model.fit(x=X_train, y=y_train, batch_size = 100, epochs=400,
2 verbose=2, validation_data=(X_test, y_test),
3 shuffle=True, initial_epoch=0)
File d:\viewer\Conda\envs\failuredetection\lib\site-packages\keras\src\utils\traceback_utils.py:70, in filter_traceback..error_handler(*args, **kwargs)
67 filtered_tb = _process_traceback_frames(e.traceback)
68 # To get the full stack trace, call:
69 # tf.debugging.disable_traceback_filtering()
---> 70 raise e.with_traceback(filtered_tb) from None
71 finally:
72 del filtered_tb
File d:\viewer\Conda\envs\failuredetection\lib\site-packages\tensorflow\python\eager\execute.py:53, in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
51 try:
52 ctx.ensure_initialized()
---> 53 tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
54 inputs, attrs, num_outputs)
55 except core._NotOkStatusException as e:
56 if name is not None:
InvalidArgumentError: Graph execution error:
...
File "d:\viewer\Conda\envs\failuredetection\lib\site-packages\keras\src\optimizers\optimizer.py", line 276, in compute_gradients
grads = tape.gradient(loss, var_list)
Node: 'gradient_tape/model/conv1d_1/Conv1D/Conv2DBackpropInput'
Conv2DCustomBackpropInputOp only supports NHWC.
[[{{node gradient_tape/model/conv1d_1/Conv1D/Conv2DBackpropInput}}]] [Op:__inference_train_function_7795]
以下是发生错误的代码
history = model.fit(x=X_train, y=y_train, batch_size = 100, epochs=400,
verbose=2, validation_data=(X_test, y_test),
shuffle=True, initial_epoch=0)
错误日志如下：
InvalidArgumentError Traceback (most recent call last)
e:\Git_repository\bear_fault_diagnosis\cnn_lstm_model.ipynb 单元格 11 line 1
----> 1 history = model.fit(x=X_train, y=y_train, batch_size = 100, epochs=400,
2 verbose=2, validation_data=(X_test, y_test),
3 shuffle=True, initial_epoch=0)
File d:\viewer\Conda\envs\failuredetection\lib\site-packages\keras\src\utils\traceback_utils.py:70, in filter_traceback..error_handler(*args, **kwargs)
67 filtered_tb = _process_traceback_frames(e.traceback)
68 # To get the full stack trace, call:
69 #
tf.debugging.disable_traceback_filtering()---> 70 raise e.with_traceback(filtered_tb) from None
71 finally:
72 del filtered_tb
File d:\viewer\Conda\envs\failuredetection\lib\site-packages\tensorflow\python\eager\execute.py:53, in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
51 try:
52 ctx.ensure_initialized()
---> 53 tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
54 inputs, attrs, num_outputs)
55 except core._NotOkStatusException as e:
56 if name is not None:
InvalidArgumentError: Graph execution error:
...
File "d:\viewer\Conda\envs\failuredetection\lib\site-packages\keras\src\optimizers\optimizer.py", line 276, in compute_gradients
grads = tape.gradient(loss, var_list)
Node: 'gradient_tape/model/conv1d_1/Conv1D/Conv2DBackpropInput'
Conv2DCustomBackpropInputOp only supports NHWC.
[[{{node gradient_tape/model/conv1d_1/Conv1D/Conv2DBackpropInput}}]] [Op:__inference_train_function_7795]