End to End application for Custom Named Entity Recognition. Highlights:
- Powerd by GenAi
- Few shot Learning
- Training and inference pipelines
Auto_annotate
will take unlabeled text data and create labellied text data that can further be used for custom Named Entity Recognition (NER) Model training.
Chechout the Demo hosted at Link
run following command in terminal
pip install auto-ner
Run following command in terminal
git clone https://github.com/bokey007/auto_ner.git
cd auto_ner
python setup.py sdist bdist_wheel
pip install ./dist/auto_ner-0.1.2.tar.gz
auto_ner.run
- Above command will lauch the app on default port 8501.
- Open the browser and go to http://localhost:8501
- Select the image and then select the appropriate set of operations you want to perform on that perticular image.
- play with the parameters interatively untill you reach at optimal configuration.
auto_ner.run --port 8080
Above command can be used to specify the port on which you want to run the app.
- Create the baseline Spacy Model ([Transformer implementation on Hold])
- Meet the Expectations Training Bert ([ToDo])
- Exeed the expectations
- Few shot / Zero Shot NER
- Beyond mere NER : entyity linking ([ToDo])
Development tools:
- setuptools (https://pypi.org/project/setuptools/): Used to create a python package
- pipreqs (https://pypi.org/project/pipreqs/): Used to create requirements.txt file
- twine (https://pypi.org/project/twine/): Used to upload the package to pypi.org
- Github Actions (): Used to automate the process of uploading the package to pypi.org
- pytest (https://pypi.org/project/pytest/): Used to write unit tests
- wheel (https://pypi.org/project/wheel/): Used to create a wheel file