ValueError(Shape error) while running predict.py iam getting this error. Anyone can help me,Thanks in advance

Question

ValueError(Shape error) while running predict.py iam getting this error. Anyone can help me,Thanks in advance

naganikshith04 opened this issue 4 years ago · comments

C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:469: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint8 = np.dtype([("qint8", np.int8, 1)])
C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:470: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_quint8 = np.dtype([("quint8", np.uint8, 1)])
C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:471: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint16 = np.dtype([("qint16", np.int16, 1)])
C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:472: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_quint16 = np.dtype([("quint16", np.uint16, 1)])
C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:473: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
np_qint32 = np.dtype([("qint32", np.int32, 1)])
C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:476: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
np_resource = np.dtype([("resource", np.ubyte, 1)])
C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\h5py_init.py:36: FutureWarning: Conversion of the second argument of issubdtype from float to np.floating is deprecated. In future, it will be treated as np.float64 == np.dtype(float).type.
from ._conv import register_converters as _register_converters
Using TensorFlow backend.
Traceback (most recent call last):
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\common_shapes.py", line 686, in _call_cpp_shape_fn_impl
input_tensors_as_shapes, status)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\errors_impl.py", line 473, in exit
c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.InvalidArgumentError: Dimension 1 in both shapes must be equal, but are 358 and 279 for 'Assign_19' (op: 'Assign') with input shapes: [256,358], [256,279].

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "predict.py", line 148, in
generate()
File "predict.py", line 30, in generate
model = create_network(normalized_input, n_vocab)
File "predict.py", line 84, in create_network
model.load_weights('weights.hdf5')
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\keras_impl\keras\models.py", line 693, in load_weights
topology.load_weights_from_hdf5_group(f, layers)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\keras_impl\keras\engine\topology.py", line 1486, in load_weights_from_hdf5_group
K.batch_set_value(weight_value_tuples)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\keras_impl\keras\backend.py", line 2401, in batch_set_value
assign_op = x.assign(assign_placeholder)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\ops\variables.py", line 573, in assign
return state_ops.assign(self._variable, value, use_locking=use_locking)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\ops\state_ops.py", line 276, in assign
validate_shape=validate_shape)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\ops\gen_state_ops.py", line 56, in assign
use_locking=use_locking, name=name)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 787, in _apply_op_helper
op_def=op_def)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\ops.py", line 2958, in create_op
set_shapes_for_outputs(ret)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\ops.py", line 2209, in set_shapes_for_outputs
shapes = shape_func(op)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\ops.py", line 2159, in call_with_requiring
return call_cpp_shape_fn(op, require_shape_fn=True)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\common_shapes.py", line 627, in call_cpp_shape_fn
require_shape_fn)
File "C:\Users\nagan\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\common_shapes.py", line 691, in _call_cpp_shape_fn_impl
raise ValueError(err.message)
ValueError: Dimension 1 in both shapes must be equal, but are 358 and 279 for 'Assign_19' (op: 'Assign') with input shapes: [256,358], [256,279].

Zoe Sheill · Answer 1 · Wed Apr 22 2020 16:17:43 GMT+0800 (China Standard Time)

hey! i'm having the same problem - did you happen to find a solution?
thanks!

naganikshith04 · Answer 2 · Wed Apr 22 2020 16:20:42 GMT+0800 (China Standard Time)

@zsheill7 Sorry , but i was unable to get a solution.Sed:(

Sigurður Skúli Sigurgeirsson · Answer 3 · Wed Apr 22 2020 17:16:09 GMT+0800 (China Standard Time)

Hi,

This is due to inconsistencies between the model and the data you're using. If you make changes to the data then you have to train the model again so that you can generate songs.

naganikshith04 · Answer 4 · Wed Apr 22 2020 17:47:52 GMT+0800 (China Standard Time)

@Skuldur that means we need to train the model again?thank you.

Sigurður Skúli Sigurgeirsson · Answer 5 · Wed Apr 22 2020 17:51:54 GMT+0800 (China Standard Time)

Yes, you need to train the model again using your data and then use the weights file generated by training the model to generate sequences of notes. :)

The weights I have included in the repository only works with the data that's provided in the repository.

naganikshith04 · Answer 6 · Wed Apr 22 2020 17:54:11 GMT+0800 (China Standard Time)

@Skuldur i was usung your weights and data you have provided ...but this error has arised.

Sigurður Skúli Sigurgeirsson · Answer 7 · Wed Apr 22 2020 17:56:12 GMT+0800 (China Standard Time)

Okay, I'll look into this after work today. Finally have some time due to social isolation.

naganikshith04 · Answer 8 · Wed Apr 22 2020 17:57:15 GMT+0800 (China Standard Time)

@Skuldur Thankyou so much!

Sigurður Skúli Sigurgeirsson · Answer 9 · Sat Apr 25 2020 04:02:55 GMT+0800 (China Standard Time)

Hi,

I just pulled the repository into a clean directory and running predict.py worked perfectly. Are you certain that you didn't make any changes to any file after clone the repository?

naganikshith04 · Answer 10 · Sat Apr 25 2020 07:07:58 GMT+0800 (China Standard Time)

@Skuldur Thankyou very much. I will again clone the repo and try it again sorry man! :).

Sutirtha Chakraborty · Answer 11 · Thu Aug 06 2020 19:08:25 GMT+0800 (China Standard Time)

Still facing the issue, any clue? If I run a few epochs it works, any clue?

Sigurður Skúli Sigurgeirsson · Answer 12 · Thu Aug 06 2020 22:53:32 GMT+0800 (China Standard Time)

If you change the dataset and the data/notes file then you have to retrain the model because the amount of different notes and chords will differ from what the model is expecting.

shaswat-du · Answer 13 · Thu Aug 27 2020 12:02:53 GMT+0800 (China Standard Time)

Hi @Skuldur. The Weights doesn't fit the model used in the predict.py. It has nothing to do with the data set. Could you share the model you used for training? That'll be very helpful. Thanks.

Sigurður Skúli Sigurgeirsson · Answer 14 · Fri Aug 28 2020 01:53:07 GMT+0800 (China Standard Time)

Hi,

I just pulled the repository into a clean environment on my work computer and

Running predict.py directly worked perfectly.
Training a model using lstm.py and using the weights from that model in predict.py worked perfectly.

Are you certain that you didn't make any changes to any file after clone the repository? Predict.py uses data/notes to get information on dimensions. If you run lstm.py, that file gets overwritten with information corresponding to your current dataset.

trulee0913 · Answer 15 · Mon Nov 02 2020 12:12:07 GMT+0800 (China Standard Time)

I think it is not a problem of coding. It is the problem of communication.
For these who get the value error, simple open predict.py and make following changes should help.
After we train the model, we will get new weights file called weights improvement or something like that.
Simple replace the name of this file to weights.hdf5 should solve the issue. Have a nice music. This programe is awesome.