ifis-tu-bs / pyke Goto Github PK

During the execution of TransR, the following error occurs:

Traceback (most recent call last):
  File "/Users/milost/anaconda/envs/pyKE/bin/pyke", line 11, in <module>
    load_entry_point('pyKE', 'console_scripts', 'pyke')()
  File "/Users/milost/anaconda/envs/pyKE/lib/python3.6/site-packages/click/core.py", line 764, in __call__
    return self.main(*args, **kwargs)
  File "/Users/milost/anaconda/envs/pyKE/lib/python3.6/site-packages/click/core.py", line 717, in main
    rv = self.invoke(ctx)
  File "/Users/milost/anaconda/envs/pyKE/lib/python3.6/site-packages/click/core.py", line 1137, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "/Users/milost/anaconda/envs/pyKE/lib/python3.6/site-packages/click/core.py", line 956, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/Users/milost/anaconda/envs/pyKE/lib/python3.6/site-packages/click/core.py", line 555, in invoke
    return callback(*args, **kwargs)
  File "/Users/milost/Code/Python/pyKE/pyke/bin/pyke.py", line 68, in transr
    margin=margin,
  File "/Users/milost/Code/Python/pyKE/pyke/embedding.py", line 52, in __init__
    self.__init_config()
  File "/Users/milost/Code/Python/pyKE/pyke/embedding.py", line 78, in __init_config
    con.set_model(self.model_class)
  File "/Users/milost/Code/Python/pyKE/pyke/openke/Config.py", line 257, in set_model
    self.trainModel = self.model(config=self)
  File "/Users/milost/Code/Python/pyKE/pyke/models/Model.py", line 87, in __init__
    self.loss_def()
  File "/Users/milost/Code/Python/pyKE/pyke/models/TransR.py", line 51, in loss_def
    p_h = tf.reshape(self._transfer(pos_matrix, pos_h_e), [-1, config.rel_size])
  File "/Users/milost/Code/Python/pyKE/pyke/models/TransR.py", line 10, in _transfer
    return tf.batch_matmul(transfer_matrix, embeddings)
AttributeError: module 'tensorflow' has no attribute 'batch_matmul'

According to this issue the tf.batch_matmul() op was removed in 3a88ec0. So we need to replace to the call to tf.batch_matmul with another call, probably something like tf.linalg.matmul

Add Sphinx documentation and submit to Read The Docs.

The model is currently only saved after the whole training process.

• The user should be able to save the embedding to a JSON file.
• The application should save the model during the training process. This is archived with a parameter "autosave" which is the time period after which the model is saved.
• The application should read an existing model and continue training.

Add unit tests to provide faster testing

It should be possible to get the embedding of a specific entity. To this end, we need to add the corresponding query functionality to the Embedding class.

The user must provide benchmark files in order to train a model. These files are composed of an entity, a relation, a train and a validation file.
The user should be able to provide a N-Triples-file to train a model (https://en.wikipedia.org/wiki/N-Triples).

The training of TransE is incorrect. The loss goes down too fast.

Parameters:
dim: 50
neg_ent: 5
folds: 20
epochs: 20

After the fourth iteration the loss reaches 0.0. The original project takes a lot longer to decrease.

Tensorflow allocated the complete amount of gpu memory. This behaviour should be changed to a variable amount of memory to allow multiple users to use one GPU.

See: https://stackoverflow.com/questions/34199233/how-to-prevent-tensorflow-from-allocating-the-totality-of-a-gpu-memory

During the training process, only two numbers are displayed on the console. Intuitively it is certainly clear what is meant by the two values, but it would be nice if they were better readable.

The project should be restructured to provide a python package which can be installed via
python setup.py install.

It would be nice to have a console script that allows you to start the training process for embeddings directly from the console.

The C++ part of the application and the shared library should be removed.
The contents of the library should be available in Python.