tensorflow的tf.in_top_k()错误

最新推荐文章于 2023-11-01 15:11:26 发布

TCcoder

最新推荐文章于 2023-11-01 15:11:26 发布

阅读量1.2k

点赞数 1

CC 4.0 BY-SA版权

分类专栏： TensorFlow 文章标签： Python TensorFlow 图像分类神经网络

本文链接：https://blog.youkuaiyun.com/qq_36032990/article/details/78190544

TensorFlow 专栏收录该内容

1 篇文章

订阅专栏

本文探讨了在使用TensorFlow进行图像识别时遇到的一个常见错误：InTopK操作中的目标越界问题。通过分析错误来源及调整标签数据，解决了分类任务中的训练障碍。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

我在TensorFlow图像识别学习中，遇到的困难！！！

错误：

Traceback (most recent call last):
File "as1_train.py", line 199, in <module>
main()
File "as1_train.py", line 174, in main
run_training()
File "as1_train.py", line 68, in run_training
summary_str = sess.run(summary_op)
File "/home/mrzhang/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 895, in run
run_metadata_ptr)
File "/home/mrzhang/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1124, in _run
feed_dict_tensor, options, run_metadata)
File "/home/mrzhang/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1321, in _do_run
options, run_metadata)
File "/home/mrzhang/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1340, in _do_call
raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.InvalidArgumentError: targets[20] is out of range
[[Node: accuracy/InTopK = InTopK[T=DT_INT32, k=1, _device="/job:localhost/replica:0/task:0/cpu:0"](softmax_linear/softmax_linear_1/_5333, Reshape_1/_5335)]]

Caused by op u'accuracy/InTopK', defined at:
File "as1_train.py", line 199, in <module>
main()
File "as1_train.py", line 174, in main
run_training()
File "as1_train.py", line 40, in run_training
train_acc = as1_model.evaluation(train_logits,label_batch)
File "/home/mrzhang/xin/py3/test/as1_model.py", line 135, in evaluation
correct = tf.nn.in_top_k(logits, labels, 1)
File "/home/mrzhang/anaconda2/lib/python2.7/site-packages/tensorflow/python/ops/gen_nn_ops.py", line 1466, in in_top_k
targets=targets, k=k, name=name)
File "/home/mrzhang/anaconda2/lib/python2.7/site-packages/tensorflow/python/framework/op_def_library.py", line 767, in apply_op
op_def=op_def)
File "/home/mrzhang/anaconda2/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 2630, in create_op
original_op=self._default_original_op, op_def=op_def)
File "/home/mrzhang/anaconda2/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 1204, in __init__
self._traceback = self._graph._extract_stack() # pylint: disable=protected-access

InvalidArgumentError (see above for traceback): targets[20] is out of range
[[Node: accuracy/InTopK = InTopK[T=DT_INT32, k=1, _device="/job:localhost/replica:0/task:0/cpu:0"](softmax_linear/softmax_linear_1/_5333, Reshape_1/_5335)]]

找了好久，终于找到错误原因，不过并没有解决！！

可以参考：https://stackoverflow.com/questions/38881918/tensorflow-tf-nn-in-top-k-targets-out-of-range-error

上面大意是：比如我分类8类，如果我的标签labels是(1,2,3,4,6,7,8,9)，就会出现上面的错误；必须改成(1,2,3,4,5,6,7,8)；然而我遇到的问题是必须从0开始，即标签必须是(0,1,2,3,4,5,6,7)。然后可以训练了。

为什么会出现这个错误？还没有找到原因，原文中说：it`s an issue

有兴趣的可以去找找原因！！！