Getting error 'str' object has no attribute 'dtype' when exporting textsum model for TensorFlow Serving

Question

I am currently trying to get a TF textsum model exported using the PREDICT SIGNATURE. I have _Decode returning a result from a passed in test article string and then I pass that to buildTensorInfo. This is in-fact a string being returned.

Now when I run the textsum_export.py logic to export the model, it gets to the point where is is building out the TensorInfo object however errors out with the below trace. I know that the PREDICT signature is usually used with images. Is this the problem? Can I not use this for the Textsum model because I am working with strings?

Error is:

Traceback (most recent call last):
  File "export_textsum.py", line 129, in Export
    tensor_info_outputs = tf.saved_model.utils.build_tensor_info(res)
  File "/usr/local/lib/python2.7/site-packages/tensorflow/python/saved_model/utils_impl.py", line 37, in build_tensor_info
    dtype_enum = dtypes.as_dtype(tensor.dtype).as_datatype_enum
AttributeError: 'str' object has no attribute 'dtype'

The TF session where the model is exported is below:

with tf.Session(config = config) as sess:

                # Restore variables from training checkpoints.
                ckpt = tf.train.get_checkpoint_state(FLAGS.checkpoint_dir)
                if ckpt and ckpt.model_checkpoint_path:
                    saver.restore(sess, ckpt.model_checkpoint_path)
                    global_step = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1]
                    print('Successfully loaded model from %s at step=%s.' %
                        (ckpt.model_checkpoint_path, global_step))
                    res = decoder._Decode(saver, sess)

                    print("Decoder value {}".format(type(res)))
                else:
                    print('No checkpoint file found at %s' % FLAGS.checkpoint_dir)
                    return

                # Export model
                export_path = os.path.join(FLAGS.export_dir,str(FLAGS.export_version))
                print('Exporting trained model to %s' % export_path)


                #-------------------------------------------

                tensor_info_inputs = tf.saved_model.utils.build_tensor_info(serialized_tf_example)
                tensor_info_outputs = tf.saved_model.utils.build_tensor_info(res)

                prediction_signature = (
                    tf.saved_model.signature_def_utils.build_signature_def(
                        inputs={ tf.saved_model.signature_constants.PREDICT_INPUTS: tensor_info_inputs},
                        outputs={tf.saved_model.signature_constants.PREDICT_OUTPUTS:tensor_info_outputs},
                        method_name=tf.saved_model.signature_constants.PREDICT_METHOD_NAME
                        ))

                #----------------------------------

                legacy_init_op = tf.group(tf.tables_initializer(), name='legacy_init_op')
                builder = saved_model_builder.SavedModelBuilder(export_path)

                builder.add_meta_graph_and_variables(
                    sess=sess, 
                    tags=[tf.saved_model.tag_constants.SERVING],
                    signature_def_map={
                        'predict':prediction_signature,
                    },
                    legacy_init_op=legacy_init_op)
                builder.save()

                print('Successfully exported model to %s' % export_path)

PREDICT signature work with tensors, res_tensor = tf.convert_to_tensor(res) — Gaurav Pawar
– Gaurav Pawar, Commented Oct 21, 2017 at 13:43
Gaurav you are awesome! That worked perfectly. I can't seem to set this comment as the answer but you should be the one to get the credit. If you can provide your comment above as an answer I will accept it. Thanks again! — xtr33me
– xtr33me, Commented Oct 21, 2017 at 22:57

Gaurav Pawar · Accepted Answer · 2017-10-22 05:54:14Z

4

PREDICT signature work with tensors, if res is 'str' type python variable, then res_tensor will be of dtype tf.string

res_tensor = tf.convert_to_tensor(res)

answered Oct 22, 2017 at 5:54

Gaurav Pawar

4482 silver badges8 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Getting error 'str' object has no attribute 'dtype' when exporting textsum model for TensorFlow Serving

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related