Loss decrease to 0 suddenly

I implement TransE with pytorch follow the algorithm in this paperhttps://papers.nips.cc/paper/5071-translating-embeddings-for-modeling-multi-relational-data.pdf. The Loss will always decrease to 0 after a few iterations although I have adjusted the learning rate to very small value. I am wondering if some people have any suggestions about this problem.
52 pm

You should explain more about your implementation…

1 Like