- Tokenizer and Dense featurizer using pre-trained models supported by PaddleNLP.
pip install rasa-paddlenlp
In your config.yml, use the following configuration:
language: zh
pipeline:
- name: "rasa_paddlenlp.nlu.paddlenlp_tokenizer.PaddleNLPTokenizer"
model_name: bert
model_weights: bert-wwm-ext-chinese
# Flag to check whether to split intents
intent_tokenization_flag: false
# Symbol on which intent should be split
intent_split_symbol: "_"
- name: "rasa_paddlenlp.nlu.paddlenlp_featurizer.PaddleNLPFeaturizer"
model_name: bert
model_weights: bert-wwm-ext-chinese
# rest of your configurations
Currently there is code to support BERT pre-trained models, we just need to add the model definitions and default weights in order for other PaddleNLP-supported models.
This package took inspiration from the following projects:
This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage
project template.