search

wzhouad / ATLOP

Source code for paper "Document-Level Relation Extraction with Adaptive Thresholding and Localized Context Pooling", AAAI 2021
197 stars 39 forks source link

Using a fixed seed doesn't seed to work #5

Closed mjeensung closed 3 years ago

mjeensung commented 3 years ago

The random seed is fixed to 66 in the code, but the result slightly changes even in the same setting.

Is it possible to see the same results every run?

wzhouad commented 3 years ago

I don't know how to make the code fully deterministic. When I run the code with the same seed, the loss differs after a few steps and accumulates in training, leading to different results. I think it's a common problem in Pytorch.

mjeensung commented 3 years ago

It's interesting. Thank you for your answer.