carlb15 / Dinosaur_Island_Character_Lvl_Language_Model

Character Level RNN for generating dinosaur names.

NOTE: Implementation hidden due to Coursera Terms of Service.

Character Level RNN for Dinosaur Name Generation

Implementation of a multi-layer Recurrent Neural Network (RNN) for training a character level language model to generate dinosaur names using numpy. The model processes a corpus of dinosaur names, trains the character level language model, and generates text character by character that's similar to the original data (dinosaur names).

What was learned:

How to store text data for processing using an RNN
How to synthesize data, by sampling predictions at each time step and passing it to the next RNN-cell unit
How to build a character-level text generation recurrent neural network
Why clipping the gradients is important for avoiding exploding gradients
How to use numpy and keras for quick model implementation

Shakespeare Poem Generator

Shakespearian poem generator using Keras. Long Short Term Memory cells are used to ensure long-term dependencies within a sentence are captured when training the model.

Implementation

A model was trained for ~1000 epochs on a collection of Shakespearian poems called "The Sonnets". The LSTM is deeper, stacked LSTM model (2 layers). Uses Keras for quick model creation.

Credit:

Sequence Models Course for Deep Learning Specialization