search

txsun1997 / CoLAKE

COLING'2020: CoLAKE: Contextualized Language and Knowledge Embedding
https://aclanthology.org/2020.coling-main.327/
MIT License
114 stars 17 forks source link

Unable to carry out the steps under "Re-train CoLAKE" section of README.md #13

Closed calebjacksonhoward closed 3 years ago

calebjacksonhoward commented 3 years ago

Specifically, the step: python preprocess/gen_data.py 4

Fails with: Exception has occurred: KeyError 'ents' File "/home/caleb/Work/CareActr/projects/phase1/repo/CoLAKE/preprocess/gen_data.py", line 62, in load_data h, t = ins['ents'][0][0], ins['ents'][1][0] File "/home/caleb/Work/CareActr/projects/phase1/repo/CoLAKE/preprocess/gen_data.py", line 95, in d_ent, head_cluster = load_data()

txsun1997 commented 3 years ago

Sorry for the late response. It seems that the data '../ernie_data/fewrel/test.json' from https://github.com/thunlp/ERNIE has changed. Fortunately, as commented in the code, this part is to remove FewRel test set from our training data for a fair comparison. So if you do not aim to publish a paper, you can just discard this part of codes: line #54 - line #61.

calebjacksonhoward commented 3 years ago

Thank you!