tensorflow: list index out of range in freeze_graph.py

I have the same problem when using keras advanced layers.

same here.

I believe this is still not working with the latest version of tensorflow 2.1.0. The error becomes: self.handle_op = var.op.inputs[0] IndexError: tuple index out of range @ymodak Any updates on this?

The full stack error is:

 File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 491, in <module>
    run_main()
  File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 487, in run_main
    app.run(main=my_main, argv=[sys.argv[0]] + unparsed)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/platform/app.py", line 40, in run
    _run(main=main, argv=argv, flags_parser=_parse_flags_tolerate_undef)
  File "/path/to/python3.6/site-packages/absl/app.py", line 300, in run
    _run_main(main, args)
  File "/path/to/python3.6/site-packages/absl/app.py", line 251, in _run_main
    sys.exit(main(argv))
  File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 486, in <lambda>
    my_main = lambda unused_args: main(unused_args, flags)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 378, in main
    flags.saved_model_tags, checkpoint_version)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 361, in freeze_graph
    checkpoint_version=checkpoint_version)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 190, in freeze_graph_with_def_protos
    var_list=var_list, write_version=checkpoint_version)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saver.py", line 828, in __init__
    self.build()
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saver.py", line 840, in build
    self._build(self._filename, build_save=True, build_restore=True)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saver.py", line 878, in _build
    build_restore=build_restore)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saver.py", line 482, in _build_internal
    names_to_saveables)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saving/saveable_object_util.py", line 349, in validate_and_slice_inputs
    for converted_saveable_object in saveable_objects_for_op(op, name):
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saving/saveable_object_util.py", line 210, in saveable_objects_for_op
    variable, "", name)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saving/saveable_object_util.py", line 84, in __init__
    self.handle_op = var.op.inputs[0]
IndexError: tuple index out of range

@jvishnuvardhan Thank you for checking on this. It links to couple other issues #22029 #5387 #4363

Here I provides a minimal code using tf hub:

import tensorflow as tf
import tensorflow_hub as hub

module = hub.Module("https://tfhub.dev/google/imagenet/resnet_v2_50/classification/1")
saver = tf.train.Saver()
with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    tf.train.write_graph(sess.graph, "resnet", "model.pb", as_text=False)
    saver.save(sess,'resnet/model')

Then I encountered the IndexError as I run

python -m tensorflow.python.tools.freeze_graph --input_graph=resnet/model.pb --input_checkpoint=resnet/model --output_graph=resnet/frozen_graph.pb --output_node_names=module/resnet_v2_50/predictions/Reshape_1 --input_binary

At the same time, i have successfully frozen a plain-vanilla single layer CNN. Could you point some directions to check? Thanks!

I believe this is still not working with the latest version of tensorflow 2.1.0. The error becomes: self.handle_op = var.op.inputs[0] IndexError: tuple index out of range @ymodak Any updates on this?

The full stack error is:

 File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 491, in <module>
    run_main()
  File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 487, in run_main
    app.run(main=my_main, argv=[sys.argv[0]] + unparsed)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/platform/app.py", line 40, in run
    _run(main=main, argv=argv, flags_parser=_parse_flags_tolerate_undef)
  File "/path/to/python3.6/site-packages/absl/app.py", line 300, in run
    _run_main(main, args)
  File "/path/to/python3.6/site-packages/absl/app.py", line 251, in _run_main
    sys.exit(main(argv))
  File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 486, in <lambda>
    my_main = lambda unused_args: main(unused_args, flags)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 378, in main
    flags.saved_model_tags, checkpoint_version)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 361, in freeze_graph
    checkpoint_version=checkpoint_version)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/tools/freeze_graph.py", line 190, in freeze_graph_with_def_protos
    var_list=var_list, write_version=checkpoint_version)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saver.py", line 828, in __init__
    self.build()
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saver.py", line 840, in build
    self._build(self._filename, build_save=True, build_restore=True)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saver.py", line 878, in _build
    build_restore=build_restore)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saver.py", line 482, in _build_internal
    names_to_saveables)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saving/saveable_object_util.py", line 349, in validate_and_slice_inputs
    for converted_saveable_object in saveable_objects_for_op(op, name):
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saving/saveable_object_util.py", line 210, in saveable_objects_for_op
    variable, "", name)
  File "/path/to/python3.6/site-packages/tensorflow_core/python/training/saving/saveable_object_util.py", line 84, in __init__
    self.handle_op = var.op.inputs[0]
IndexError: tuple index out of range

Do you have any progress？I have the same problem.

I get the same issue when I try to freeze the bidirectional LSTM network from tensor2tensor (https://github.com/tensorflow/tensor2tensor/blob/v.1.12.0/tensor2tensor/models/lstm.py).

I got same issue. Here is the actual script help me to freeze the model. Note that you need to modify the output node name manually. And I successfully loaded the frozen model onto android.

@jiayiliu - yes, I am using the internal tensorflow.keras.

My workaround has been to use tf.saved_model.simple_save instead of a Saver and have freeze_graph use input_saved_model_dir.

It gets me past this problem but I don’t consider the issue resolved. If freeze_graph is not going to support working on normal graphs it should deprecate that api and update the docs, or at least put a note that for keras models you have to used saved_model.

I’m getting the same error on a keras implementation of yolo.

The error occurs in __init__ of a ResourceVariableSaveable when trying to access var.op.inputs[0] on a tensor:

Tensor("batch_normalization_1/beta:0", shape=(), dtype=resource)

op is a “VarHandleOp” and var.op.inputs is a tensorflow.python.framework.ops.Operation._InputList object.

The input list in this case is empty. Does the resource variable have to have an input? What should handle_op be in this case?

Any suggestions for how to proceed?