possibly wrong parameter order with tf.split

[xthan]

possibly wrong parameter order with tf.split

[xthan]

You are right, the order is wrong here because I did not test on TF >= 1.0 thoroughly. Will fix it. Thanks!

[zyyang]

I trained a 1.x model successfully, but could not run inference, neither.
Thinking about going with Keras.
Any suggestion?

[xthan]

What is wrong with your inference? Maybe I can help.

[zyyang]

Thanks, Xin Tong.
I trained the model with commit dd1e03e following your description.

But I failed to restoring parameters from model. (I trained on a TitanX server while using the model locally at my Mac.)
The loading goes like(taken from your latter committed code):

g = tf.Graph() with g.as_default(): model_config = configuration.ModelConfig() model_config.rnn_type = 'lstm' model = polyvore_model.PolyvoreModel(model_config, mode="inference") model.build() saver = tf.train.Saver() g.finalize() sess = tf.Session(graph=g) saver.restore(sess, 'model.ckpt-99543') # this is what I got after training.

The error ( ) was like:

DataLossError Traceback (most recent call last) <ipython-input-5-5b6a557fd54a> in <module>() ----> 1 saver.restore(sess, 'model.ckpt-99543') DataLossError: Unable to open table file ./model.ckpt-99543: Data loss: not an sstable (bad magic number): perhaps your file is in a different file format and you need to use a different restore operator? [[Node: save/RestoreV2_387 = RestoreV2[dtypes=[DT_FLOAT], _device="/job:localhost/replica:0/task:0/device:CPU:0"](_arg_save/Const_0_0, save/RestoreV2_387/tensor_names, save/RestoreV2_387/shape_and_slices)]] .... DataLossError (see above for traceback): Unable to open table file ./model.ckpt-99543: Data loss: not an sstable (bad magic number): perhaps your file is in a different file format and you need to use a different restore operator? [[Node: save/RestoreV2_387 = RestoreV2[dtypes=[DT_FLOAT], _device="/job:localhost/replica:0/task:0/device:CPU:0"](_arg_save/Const_0_0, save/RestoreV2_387/tensor_names, save/RestoreV2_387/shape_and_slices)]]

I have no idea about this error. it is hard to debug, so I am planning to re impl your paper in Keras...