tfrecord 训练时报错 DataLossError corrupted record at 12
背景
图像分类,因为数据比较多所以转成了tfrecord进行训练,但是训练一个epoch之后总是报错。
报错信息
DataLossError (see above for traceback): corrupted record at 12
3/3 [==============================] - 73s 24s/step - loss: 0.6961 - accuracy: 0.5050
Traceback (most recent call last):
File "debug.py", line 148, in <module>
validation_data=eval_dataset)
File "/root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training.py", line 728, in fit
use_multiprocessing=use_multiprocessing)
File "/root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2.py", line 370, in fit
total_epochs=1)
File "/root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2.py", line 123, in run_one_epoch
batch_outs = execution_function(iterator)
File "/root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/keras/engine/training_v2_utils.py", line 86, in execution_function
distributed_function(input_fn))
File "/root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/eager/def_function.py", line 457, in __call__
result = self._call(*args, **kwds)
File "/root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/eager/def_function.py", line 526, in _call
return self._concrete_stateful_fn._filtered_call(canon_args, canon_kwds) # pylint: disable=protected-access
File "/root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/eager/function.py", line 1141, in _filtered_call
self.captured_inputs)
File "/root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/eager/function.py", line 1224, in _call_flat
ctx, args, cancellation_manager=cancellation_manager)
File "/root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/eager/function.py", line 511, in call
ctx=ctx)
File "/root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/eager/execute.py", line 67, in quick_execute
six.raise_from(core._status_to_exception(e.code, message), None)
File "<string>", line 3, in raise_from
tensorflow.python.framework.errors_impl.DataLossError: 2 root error(s) found.
(0) Data loss: corrupted record at 2146170
[[node IteratorGetNext (defined at /root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/framework/ops.py:1751) ]]
(1) Data loss: corrupted record at 2146170
[[node IteratorGetNext (defined at /root/anaconda3/envs/keke/lib/python3.6/site-packages/tensorflow_core/python/framework/ops.py:1751) ]]
[[IteratorGetNext/_2]]
0 successful operations.
0 derived errors ignored. [Op:__inference_distributed_function_89740]
Function call stack:
distributed_function -> distributed_function
解决思路
很明显是data generator的问题,要么数据有问题,要么迭代有问题。
我的思路是制作一个迷你tfreocrd数据集,run一下data generator,把数据print出来,查看数据有没有问题。
最终解决方法
为了求快我是多线程转的tfrecord,发现不用多线程就不会报错,所以还是老老实实单线程慢点转吧~