Does this work with the sent2vec unigram wiki model

[danielmorozoff]

Hi there,

Currently attempting to launch in docker container.

Been trying to try this out with the unigram model, but am getting silent failures when attempting to load the embedding distributor: embedding_distributor = launch.load_local_embedding_distributor('en')

It just closes out and crashes the docker container.

The command that launches the container:

docker run --memory=10g -v /Desktop/ai-research-keyphrase-extraction/deps/sent2vec/models/wiki_unigrams.bin:/sent2vec/pretrained_model.bin -it keyphrase-extraction

import launch
embedding_distributor = launch.load_local_embedding_distributor('en') <---FAILURE

Thanks in advance for your help. Dan

[kamilbs]

Hi Dan, It seems to be a memory issue , I tried with the bigram model and --memory=16g and it fails, whereas if I change the memory directly on the docker app everything works. Could you try to directly allow the memory from the app (Preferences > Advanced then apply and restart) Best, Kamil

[danielmorozoff]

memory prblm - closing issue