Text-Generating Neural Network

Github repositories

• minimaxir/textgenrnn

  textgenrnn is a Python module built on top of Keras/TensorFlow that allows you to train your own text-generating neural network. It offers various features such as attention-weighting and skip-embedding to improve model quality and training speed. You can train the model on character-level or word-level data and configure the size and number of RNN layers. It supports training on large text files and can be used with GPUs for faster training. The pretrained model can be easily saved and loaded for generating text. Additionally, interactive mode enables step-by-step control over the output. The module is available on PyPI and requires a minimum TensorFlow version of 2.1.0.