gbcolborne / hypernym_discovery Goto Github PK

Hi thank you very much for sharing this work!. I have the same issue when running get_corpus.sh. Is it still possible to download the corpus from that link?.

Hi , thank you for your contribution to hypernym discovery. I run your sorce code on my laboratory pc, and the result was nearly equal to UnSupervised model referred in http://aclweb.org/anthology/S18-1116.
Could you tell me how to achieve the main score of hybrid model in http://aclweb.org/anthology/S18-1116 ?

Shows this error while trying to implement your model on Google Colab. What needs to be done? Should I rather try it on PC?

Traceback (most recent call last):
File "predict.py", line 6, in
from Evaluator import Evaluator
File "/content/hypernym_discovery/Evaluator.py", line 7, in
from task9_scorer import average_precision
ModuleNotFoundError: No module named 'task9_scorer'

I can see description regarding task9_scorer in README.md but unable to implement it properly, it seems.

'''Evaluate predictions on the test set using scoring script of SemEval-2018 Task 9:

python2.7 path/to/SemEval-Task9/task9-scorer.py path/to/SemEval-Task9/test/gold/..test.gold.txt path/to/output/pred.txt
'''

Hello, I'm Jonathan and I'm an AI ENgineer Intern in a startup. I really interested in your work and I think your work is the best Hypernym Discovery model that I can find. There is not much stuff related to this Hypernym that has better documentation than yours. Thank you for that.

So, I was trying your code but I got this problem, the devloss seems to decrease until around epoch 40 and it continues to increase until it reaches the patient. The last devLoss is about 8 if I'm not mistaken, and the MAP can't reach more than 0.25. What do you think I missed? And by the way, I think there is some code that never actually executes, like make_train_dev_split.py, and it's not written in your README.md. Is it optional to run it? But I read that your prediction needs the test corpus that generated by the code I said earlier. Thank you so much.

hi, thanks for your excellent work. I found that the link for the corpus seems to be invalid during my reproduction. Is there any other way to get it?