Closed moh-yani closed 4 years ago
@moh-yani
It's normal, I spent more than 30 days to evaluate link prediction on FB15K-237, and WN18RR cost several days on a V100 GPU. The evaluation needs to replace every head/tail entity in every test triple with all entities in a knowledge graph and obtains the ranks. On ther other hand, BERT is a heavy model which needs a lot of computation.
Okay, thanks.
I am trying perform run_bert_link_prediction.py for FB15K data with GPU 8GB and 12 batch_size. After 7 days, it is still ongoing. The iteration process shows: Testing: 100% left: ??? Testing: 100% right: ??? mean rank until now: ??? hit@10 until now: ???
Is it normal? how long to perform that file in your experience?