about the early stopping

[llw0727]

hi Tim，sorry to bother you！ i noticed you used early stopping in the paper，but i didnt found the code could u show the code about the early stopping ：）

[TimDettmers]

This is something that was not really done well by me. There is no early stopping code. Instead what I do is I go through the logs and look at the best validation mean reciprocal rank and then I noted the test results for that epoch. It would have been better to implement that rather than doing it manually. It would be better to add it in retrospect since this can be quite confusing. I might have some time to do this next week.

My colleague was working with his own codebase to produce results for DistMult and ComplEx and he might have used early stopping. His codebase is this one: https://github.com/uclnlp/inferbeddings