Error when import seq2seq

[ay2456]

Hello,

I got this error when I tried to import seq2seq:

import seq2seq
Using TensorFlow backend.
Traceback (most recent call last):
File "", line 1, in
File "/usr/local/lib/python2.7/site-packages/seq2seq/init.py", line 1, in
from .cells import *
File "/usr/local/lib/python2.7/site-packages/seq2seq/cells.py", line 1, in
from recurrentshop import RNNCell, LSTMCell, weight
File "/usr/local/lib/python2.7/site-packages/recurrentshop/init.py", line 1, in
from .engine import RNNCell, RecurrentContainer, weight
File "/usr/local/lib/python2.7/site-packages/recurrentshop/engine.py", line 32, in
_backend = getattr(K, K.backend() + '_backend')
AttributeError: 'module' object has no attribute 'backend'

What should I do?

BTW,
I just installed the latest version of seq2seq and recurrentshop.
The following are the version of my python and keras.

python: 2.7.11
keras: 1.0.5

Thanks!

[farizrahman4u]

Update keras.

[ay2456]

Hi,

I updated my keras version and I can import seq2seq now.
However, my previous code is not working.

Previously, I have:
model = Seq2seq(batch_input_shape=(1, n_input_steps, n_input), hidden_dim=n_hidden, output_length=n_output_steps, output_dim=n_output, dropout=dropout, stateful=True, peek=True)

Since there is no Seq2seq now, I changed to Seq2Seq:
model = Seq2Seq(batch_input_shape=(1, n_input_steps, n_input), hidden_dim=n_hidden, output_length=n_output_steps, output_dim=n_output, dropout=dropout, stateful=True, peek=True)

But it returns an error:

File "/usr/local/lib/python2.7/site-packages/seq2seq/models.py", line 171, in Seq2Seq
encoded_seq = encoder(encoded_seq)
File "/usr/local/lib/python2.7/site-packages/keras/engine/topology.py", line 514, in call
self.add_inbound_node(inbound_layers, node_indices, tensor_indices)
File "/usr/local/lib/python2.7/site-packages/keras/engine/topology.py", line 572, in add_inbound_node
Node.create_node(self, inbound_layers, node_indices, tensor_indices)
File "/usr/local/lib/python2.7/site-packages/keras/engine/topology.py", line 149, in create_node
output_tensors = to_list(outbound_layer.call(input_tensors[0], mask=input_masks[0]))
File "/usr/local/lib/python2.7/site-packages/recurrentshop/engine.py", line 309, in call
last_output_0, outputs_0, states_0, updates = rnn(self.step, x, initial_states, go_backwards=self.go_backwards, mask=mask, unroll=unroll, input_length=input_shape[1])
File "/usr/local/lib/python2.7/site-packages/recurrentshop/engine.py", line 51, in
rnn = lambda _args, *_kwargs: list(K.rnn(_args, *_kwargs)) + [[]]
File "/usr/local/lib/python2.7/site-packages/keras/backend/tensorflow_backend.py", line 1206, in rnn
state = tf.concat(1, states)
File "/usr/local/lib/python2.7/site-packages/tensorflow/python/ops/array_ops.py", line 463, in concat
name=name)
File "/usr/local/lib/python2.7/site-packages/tensorflow/python/ops/gen_array_ops.py", line 315, in _concat
values=values, name=name)
File "/usr/local/lib/python2.7/site-packages/tensorflow/python/ops/op_def_library.py", line 704, in apply_op
op_def=op_def)
File "/usr/local/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 2262, in create_op
set_shapes_for_outputs(ret)
File "/usr/local/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 1702, in set_shapes_for_outputs
shapes = shape_func(op)
File "/usr/local/lib/python2.7/site-packages/tensorflow/python/ops/array_ops.py", line 505, in _ConcatShape
value_shape.ndims)
ValueError: concat_dim is out of range (values rank = 1)

What should I do?

Thanks!

[farizrahman4u]

Use unroll=True option with tf backend. This has to be fixed on Keras side.

[viswanath57]

Hi,

While importing seq2seq module, getting the following error

In [2]: import seq2seq

ImportError Traceback (most recent call last)
in ()
----> 1 import seq2seq

/home/viswanath/seq2seq/seq2seq/init.py in ()
----> 1 from .cells import *
2 from .models import *

/home/viswanath/seq2seq/seq2seq/cells.py in ()
----> 1 from recurrentshop import RNNCell, LSTMCell, weight
2 from keras import backend as K
3 import numpy as np
4
5

ImportError: cannot import name weight

[viswanath57]

It seems like most recent recurrentshop module does not have weight function

[PraveenKumarCSE]

Same issue for me also. The weight function removed from recurrentshop lib.

[dmitriydligach]

Same issue here. Is there a workaround?

[farizrahman4u]

fixed