tensorflow.python.framework.errors_impl.InvalidArgumentError: Feature: input_ids0 ，data type: int64

tensorflow.python.framework.errors_impl.InvalidArgumentError: Feature: input_ids0 (data type: int64) is required but could not be found. [[{{node ParseSingleExample/ParseSingleExample}}]] [[node IteratorGetNext (defined at /usr/local/lib/python3.5/dist-packages/tensorflow_estimator/python/estimator/util.py:110) ]]

错误原因 tfrecord写读不一致

原代码：

      features[f'input_ids{i}'] = tf.train.Feature(
        int64_list=tf.train.Int64List(value=list(input_ids.astype(np.int64))))

我错改为：

      features['input_ids{i}'] = tf.train.Feature(
        int64_list=tf.train.Int64List(value=list(input_ids.astype(np.int64))))

直接去掉 features[f'input_ids{i}']的‘f’，导致本来是循环的里面的index的‘i’成为str而不是变量了

报错修改后：

      features['input_ids%d' % (i)] = tf.train.Feature(
        int64_list=tf.train.Int64List(value=list(input_ids.astype(np.int64))))