The following repo contains code for training a word-level language model to generate text based on a sequence of previous text. This type of model, while more advanced than the quickstart models offered out-of-the-box by Skafos, can easily be converted to Core ML and saved to iOS applications leveraging our framework. Follow along with this blog post, or forge ahead on your own.
If you want to try out this example, first fork this repo to your own github account, then clone (or pull) it into a Skafos Jupyter Lab instance:
# Clone into your JLab using HTTPS auth
$ git clone https://github.com/<your-account>/WordLanguageModel.git
# OR pull into your JLab using HTTPS auth
$ git remote add origin https://github.com/<your account>/WordLanguageModel.git
The components of this repo are:
word_language_model.ipynb
- a Python notebook that contains code to train and export a word language model suing a recurrent neural network.requirements.txt
- a file describing all required Python dependencies.
Please contact us with questions or feedback! Here are two ways: