James Barry
James Barry
Hi @neubig, @Hyperparticle, I thought this might have something to do with some unexpected behaviour since the recent PR #13 but I cloned a version of udify c277adea9295b05772e3b508d05ce13aea8bde03 before this...
Thanks for looking into it @dfvalio. > Fixed my problem with the requirement specified by you. CONLLU must remain at 1.3.1 I tried with the `conllu` version that gets installed...
I will work on XLM (PyTorch)
Anyone have any success with this? I tried running their model on a conllu file but I am unable to "get" the representations like you would with the AllenNLP version....
`m` is the test set size, so as more labels are used for testing, the less there are for training. The training labels are used to generate synthetic data, so...
What is the best workaround for this? Should we modify the docker image to add a layer for installing/setting the path to Java?
@nreimers > ... in my implementation the ELMo embeddings are pre-computed for all sentences and then stored on disk (and hold in RAM while training). I am working on a...
Hi @nreimers, Thanks very much for the response. I will try the method you've used in your repository, thanks!