search

microth / PathLSTM

Neural SRL model
71 stars 14 forks source link

Request for added detail in README.md #1

Closed aminorex closed 7 years ago

aminorex commented 7 years ago

If you please, could you provide a downloadable reference for the other model files referenced in scripts.parse.sh? Specifically, LEMMA_MODEL, POS_MODEL, and PARSER_MODEL.

aminorex commented 7 years ago

If these are best represented by https://storage.googleapis.com/google-code-archive-downloads/v2/code.google.com/mate-tools/CoNLL2009-ST-English-ALL.anna-3.3.lemmatizer.model, https://storage.googleapis.com/google-code-archive-downloads/v2/code.google.com/mate-tools/CoNLL2009-ST-English-ALL.anna-3.3.postagger.model, and https://storage.googleapis.com/google-code-archive-downloads/v2/code.google.com/mate-tools/CoNLL2009-ST-English-ALL.anna-3.3.parser.model, respectively, an explicit comment to that effect would be appreciated, as a point of clarification.

microth commented 7 years ago

This is already mentioned in the README, 2nd line in section "Dependencies".

aminorex commented 7 years ago

Oops, thanks.

Also you might mention the dependency on opennlp tools, and which version(s) are acceptable. I am using 1.5.3 with reasonably good results so far.

After compiling, using those model files and anna-3.3.jar (or anna-3.61.jar) I observe what appears to be an artifact resulting from an inconsistency between the models and the library:

./scripts/parse.sh 2009/data/trial.txt 33.50.191 is2.data.ParametersFloat 142:read -> read parameters 134217727 not zero 296071 33.50.195 is2.data.Cluster 111: -> Read cluster with 0 words java.io.EOFException at java.io.DataInputStream.readBoolean(DataInputStream.java:244) at is2.lemmatizer.Lemmatizer.readModel(Lemmatizer.java:181) at is2.lemmatizer.Lemmatizer.(Lemmatizer.java:64) at se.lth.cs.srl.util.BohnetHelper.getLemmatizer(BohnetHelper.java:19) at se.lth.cs.srl.languages.Language.getLemmatizer(Language.java:146) at se.lth.cs.srl.languages.Language.getPreprocessor(Language.java:111) at se.lth.cs.srl.CompletePipeline.getCompletePipeline(CompletePipeline.java:55) at se.lth.cs.srl.CompletePipeline.main(CompletePipeline.java:133) 33.53.753 is2.data.ParametersFloat 142:read -> read parameters 134217727 not zero 1613201 33.53.753 is2.data.Cluster 111: -> Read cluster with 0 words 33.53.754 is2.tag.Lexicon 103: -> Read lexicon with 0 words 33.53.755 is2.tag.Tagger 141:readModel -> Loading data finished. 33.53.763 is2.parser.Parser 188:readModel -> Reading data started 33.53.784 is2.data.Cluster 111: -> Read cluster with 0 words 33.58.38 is2.parser.ParametersFloat 101:read -> read parameters 134217727 not zero 19957525 33.58.38 is2.parser.Parser 201:readModel -> parsing -- li size 134217727 Exception in thread "main" java.lang.NoSuchFieldError: featureCreation at is2.parser.Parser.readModel(Parser.java:208) at is2.parser.Parser.(Parser.java:71) at is2.parser.Parser.(Parser.java:83) at se.lth.cs.srl.util.BohnetHelper.getParser(BohnetHelper.java:43) at se.lth.cs.srl.languages.Language.getPreprocessor(Language.java:117) at se.lth.cs.srl.CompletePipeline.getCompletePipeline(CompletePipeline.java:55) at se.lth.cs.srl.CompletePipeline.main(CompletePipeline.java:133)

Any suggestions or comments on this error would be much appreciated.

microth commented 7 years ago

Make sure you only have anna-3.3 in your classpath. Remove 3.61 or any other version, if necessary. If problem persists, contact the authors of anna-3.3. OpenNLP might be imported in some class file, but it is not used in any PathLSTM components by default.