may-/cnn-re-tf

How do you create the entities.pickle file?

Closed this issue · 4 comments

iymitchell commented

Where do you download the entities.pickle file to use for distant supervision? Need help with this

may- commented

Hi,

entities.cPickle is just a dump file created by distant_supervision.py (see line 371) Do you mean, you had an error during running distant_supervision.py script? If so, please let me know which error you've got. Sorry, perhaps I didn't understand your question properly?

iymitchell commented

I tried to run distant_supervision.py but it didn't create an entities.pkl file. I get this:

entities = util.load_from_dump(os.path.join(data_dir, "entities.cPickle"))
File "cnn-re-tf-master/util.py", line 285, in load_from_dump
with open(filename, 'rb') as infile:
IOError: [Errno 2] No such file or directory: 'data/entities.cPickle'

iymitchell commented

Just to clarify, I am using a different dataset from the ones you provided in the repository. I am just trying to recreate the process with my dataset. I am getting this error from the script when i try to run it.

Thanks!

may- commented

Ah, I see, now I know what you are talking about. The script returns that error because there are "old" tsv files still in the directory. Please try this:

$ cd cnn-re-tf
$ mv ./data ./data_old
$ mkdir data
$ mv ./data_old/gold_patterns.tsv ./data
$ python ./distant_supervision.py

How about now? (It takes quit long time... You can set small parameter values, if you just want to test.)