Hi Alasdair,
I notice that the repo does not support multi-GPU training due to it uses AllenNLP v0.9.0, which is mentioned in #6 (comment).
But what should I do if I want to upgrade the code to AllenNLP v1.0.0?
Is it enough to only change the "- allennlp==0.9.0" to" - allennlp==0.9.0" in environment.yml and just re-install the enviroment?
Thanks very much!