I am updated syntaxnet/examples/dragnn/trainer_tutorial.ipynb to train english using official tensorflow/syntaxnet docker image.
docker run -it -p 8888:8888 -v ~/ud-treebanks-conll2017/UD_English:/UD_Eng tensorflow/syntaxnet
open "http://localhost:8888/?token=xxxx" in the browser
open dragnn/trainer_tutorial.ipynb and change it to train english, for example:
then change the text to english to validate
Visualize the output of our mini-trained model on a test sentence.
text = 'go to USA'
then run the script in the notebook web application. The training is done, but parsing result is incorrect.
So question:can I can use the script trainer_tutorial.ipynb for training? The script seems much simpler than this github. How to modify it to run training sucessfully?
Hi,
I am updated syntaxnet/examples/dragnn/trainer_tutorial.ipynb to train english using official tensorflow/syntaxnet docker image.
docker run -it -p 8888:8888 -v ~/ud-treebanks-conll2017/UD_English:/UD_Eng tensorflow/syntaxnet open "http://localhost:8888/?token=xxxx" in the browser open dragnn/trainer_tutorial.ipynb and change it to train english, for example:
DATA_DIR = '/UD_Eng' TENSORBOARD_DIR = '/notebooks/tensorboard' CHECKPOINT_FILENAME = '{}/eng-checkpoint'.format(DATA_DIR) TRAINING_CORPUS_PATH = '{}/en-ud-train.conllu'.format(DATA_DIR) DEV_CORPUS_PATH = '{}/en-ud-dev.conllu'.format(DATA_DIR)
then change the text to english to validate Visualize the output of our mini-trained model on a test sentence.
text = 'go to USA'
then run the script in the notebook web application. The training is done, but parsing result is incorrect.
So question:can I can use the script trainer_tutorial.ipynb for training? The script seems much simpler than this github. How to modify it to run training sucessfully?
Thanks.