Open psychoslave opened 6 years ago
Following https://github.com/louismullie/stanford-core-nlp/issues/52 this is an other configuration issue on Fedora 27.
Even after modifying StanfordCoreNLP.jvm_args trying to run a full pipeline will raise a java heap out of memory error.
StanfordCoreNLP.jvm_args
[30] pry(main)> StanfordCoreNLP.jvm_args => ["-Xms512M", "-Xmx1024M"] [31] pry(main)> StanfordCoreNLP.jvm_args = ["-Xms2048M", "-Xmx1024M"] => ["-Xms2048M", "-Xmx1024M"] [34] pry(main)> pipeline = StanfordCoreNLP.load(:tokenize, :ssplit, :pos, :lemma, :parse, :ner, :dcoref) Adding annotator tokenize Adding annotator ssplit Adding annotator pos Loading default properties from tagger /home/psychoslave/.rbenv/versions/2.4.3/lib/ruby/gems/2.4.0/gems/stanford-core-nlp-0.5.3/bin/mod els/taggers/french.tagger Reading POS tagger model from /home/psychoslave/.rbenv/versions/2.4.3/lib/ruby/gems/2.4.0/gems/stanford-core-nlp-0.5.3/bin/models/tagge rs/french.tagger ... OutOfMemoryError: Java heap space from /home/psychoslave/.rbenv/versions/2.4.3/lib/ruby/gems/2.4.0/gems/stanford-core-nlp-0.5.3/lib/stanford-core-nlp.rb:187:in `method_missing'
Following https://github.com/louismullie/stanford-core-nlp/issues/52 this is an other configuration issue on Fedora 27.
Even after modifying
StanfordCoreNLP.jvm_args
trying to run a full pipeline will raise a java heap out of memory error.