mnick / holographic-embeddings Goto Github PK

Hi Maximilian,

I was trying to run your project holographic-embeddings with mode=lp to get the AUC score. At first, I got the error message that

AttributeError: 'PairwiseStochasticTrainer' object has no attribute '_scores'

I only found the '_score' function is belong to the member 'model' of class 'PairwiseStochasticTrainer', so I changed the line 232 of base.py
from

scores = mdl._scores(self.ss, self.ps, self.os)

to

scores = mdl.model._scores(self.ss, self.ps, self.os)

It works now, but I still want to make sure whether it is the right way to do that!
Thank for your kindly help and looking forward to your reply.

Best,
Daye SHAO

Hello @mnick .

I'm am working in a project which uses https://github.com/mnick/scikit-kge library (and it's great!). That project is licensed under MIT terms, but this project, which seems useful to me, hasn't any license available.

Is it possible to add a LICENSE file, in order to know the terms in which this software can be used?

Thanks!

Hello,
I tried to execute the scripts but it gives me this error :

Traceback (most recent call last):
  File "kg/run_hole.py", line 61, in <module>
    ExpHolE().run()
  File "/home/thomas/dev/KGE/holographic-embeddings/kg/base.py", line 50, in run
    self.train()
  File "/home/thomas/dev/KGE/holographic-embeddings/kg/base.py", line 154, in train
    trn.fit(xs, ys)
  File "/home/thomas/bin/miniconda/lib/python3.6/site-packages/scikit_kge-0.1-py3.6.egg/skge/base.py", line 197, in fit
    scores_o = self.scores_o(mdl, s, p).flatten()
  File "/home/thomas/bin/miniconda/lib/python3.6/site-packages/scikit_kge-0.1-py3.6.egg/skge/base.py", line 142, in _optim
    sampler = sample.RandomModeSampler(self.args.ne, [0, 1], xs, sz)
  File "/home/thomas/bin/miniconda/lib/python3.6/site-packages/numpy/lib/shape_base.py", line 509, in split
    res = array_split(ary, indices_or_sections, axis)
  File "/home/thomas/bin/miniconda/lib/python3.6/site-packages/numpy/lib/shape_base.py", line 430, in array_split
    sub_arys.append(_nx.swapaxes(sary[st:end], axis, 0))
TypeError: slice indices must be integers or None or have an __index__ method

Thank you for your great work. When I learn your code, I am confused by the WN18.bin dataset, why entities in it are all numbers? what are they stand for ?

As explained in the main readme, countries dataset is supposed to be available in this repository but I can't find it.

How can one rebuild the same countries dataset used in HOLE experiments?

Thank you for your work. Besides WN18, you use FB15K for experimental evaluations. Can you release FB15K.bin and detail parameters to train the model for FB15K.

mldl@mldlUB1604:/ub16_prj/holographic-embeddings$ bash run_transe_wn18.sh
INFO:EX-KG:Fitting model TransE with trainer PairwiseStochasticTrainer and parameters Namespace(fin='data/wn18.bin', fout=None, init='nunif', lr=0.1, margin=2.0, me=500, mode='rank', nb=100, ncomp=50, ne=1, no_pairwise=False, sampler='random-mode', test_all=50)
Traceback (most recent call last):
File "kg/run_transe.py", line 41, in
ExpTransE().run()
File "/home/mldl/ub16_prj/holographic-embeddings/kg/base.py", line 50, in run
self.train()
File "/home/mldl/ub16_prj/holographic-embeddings/kg/base.py", line 154, in train
trn.fit(xs, ys)
File "/usr/local/lib/python3.5/dist-packages/scikit_kge-0.1-py3.5.egg/skge/base.py", line 197, in fit
scores_o = self.scores_o(mdl, s, p).flatten()
File "/usr/local/lib/python3.5/dist-packages/scikit_kge-0.1-py3.5.egg/skge/base.py", line 142, in _optim
sampler = sample.RandomModeSampler(self.args.ne, [0, 1], xs, sz)
File "/usr/local/lib/python3.5/dist-packages/numpy/lib/shape_base.py", line 509, in split
res = array_split(ary, indices_or_sections, axis)
File "/usr/local/lib/python3.5/dist-packages/numpy/lib/shape_base.py", line 430, in array_split
sub_arys.append(_nx.swapaxes(sary[st:end], axis, 0))
TypeError: slice indices must be integers or None or have an index method
mldl@mldlUB1604:/ub16_prj/holographic-embeddings$