运行TensorFlow keras报错:Function call stack: keras_scratch_graph

1 模型已经编译好了，就是进入训练阶段报错，具体信息如下：

train starting ...
Epoch 1/5
Traceback (most recent call last):
  File "simbert_training_test.py", line 353, in <module>
    callbacks=[evaluator],
  File "/usr/local/lib/python3.6/dist-packages/keras/legacy/interfaces.py", line 91, in wrapper
    return func(*args, **kwargs)
  File "/usr/local/lib/python3.6/dist-packages/keras/engine/training.py", line 1732, in fit_generator
    initial_epoch=initial_epoch)
  File "/usr/local/lib/python3.6/dist-packages/keras/engine/training_generator.py", line 220, in fit_generator
    reset_metrics=False)
  File "/usr/local/lib/python3.6/dist-packages/keras/engine/training.py", line 1514, in train_on_batch
    outputs = self.train_function(ins)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/backend.py", line 3792, in __call__
    outputs = self._graph_fn(*converted_inputs)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py", line 1605, in __call__
    return self._call_impl(args, kwargs)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py", line 1645, in _call_impl
    return self._call_flat(args, self.captured_inputs, cancellation_manager)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py", line 1746, in _call_flat
    ctx, args, cancellation_manager=cancellation_manager))
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py", line 598, in call
    ctx=ctx)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/execute.py", line 60, in quick_execute
    inputs, attrs, num_outputs)
tensorflow.python.framework.errors_impl.InvalidArgumentError:  Incompatible shapes: [30,52,768] vs. [20,52,768]
     [[node Embedding-Token-Segment/add (defined at /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3009) ]] [Op:__inference_keras_scratch_graph_41707]

Function call stack:
keras_scratch_graph
2 在网上找各种方法试了一下，都不行；又回头去重新检查数据，发现输入数据的批次不一致导致的，并且报错信息已经明确指明了tensorflow.python.framework.errors_impl.InvalidArgumentError:  Incompatible shapes: [30,52,768] vs. [20,52,768]，我输入的是两个input,第一个input批次为30，第二个为20，重新调整批次，已经解决问题。