Myungchul Shin

https://github.com/hanxiao/bert-as-service/blob/master/README.md i think this repo may help you.

good point you addressed. i’m looking for PR :) but i think the modification for ‘parser’ to accept data with batch dimension, a little bit mesh. so, i’m worried about...

i forked another repository which use bilstm-crf on the top of bert model. https://github.com/dsindex/BERT-BiLSTM-CRF-NER/blob/master/README.md this module yields 0.95~0.96 of f1 score on dev set. however, after aligning the predicted output(on...

i made scripts to merge the prediction result with original test data and evaluate by conlleval.pl. you can refer to https://github.com/dsindex/BERT-BiLSTM-CRF-NER/blob/master/predict.sh [update] - remove ext.py, merge.py - merge the prediction...

@YeopIn 1) you need to place a constituent parse tree corpus(`sejong_treebank.txt.v1`) to sejong directory. ``` $ ls align.py align_r.py c2d.py c2d.sh context.pbtxt_p env.sh eval.py log sejong_treebank.sample sejong_treebank.txt.v1 split.py split.sh tagged_input.sample...

@YeopIn 1) Is that true for Korean pos tagging using train_dragnn.sh? -> No, train_dragnn.sh stands for training dependency parser only. it is basically same as train_dragnn_sejong.sh. 2) data using UD_Korean(universal_dependencies-2.0-ud_treebans-v2.0tgz)?...

there was a similar discussion before https://github.com/dsindex/syntaxnet/issues/4#issuecomment-236389952 but, i couldn't find proper way to train Korean POS tagger. i thought ... it is worth that i use other Korean POS...

@LeDaVinci i have no idea for android. but i am curious why you try to build your system by using syntaxnet(such a heavy system in android). if you want to...

'parser_trainer.py' has '--pretrained_params, --pretrained_params_names' parameters. in documentation, they are used for global training. https://github.com/tensorflow/models/blob/master/research/syntaxnet/g3doc/syntaxnet-tutorial.md ``` bazel-bin/syntaxnet/parser_trainer \ --arg_prefix=brain_parser \ --batch_size=8 \ --decay_steps=100 \ --graph_builder=structured \ --hidden_layer_sizes=200,200 \ --learning_rate=0.02 \ --momentum=0.9...

i guess there is a dimension mismatch. ``` lhs shape= [1436,8] rhs shape= [1297,8] ``` what is the hidden layer size of the model you have? in 'train.sh', '64' is...