DeepLearning.AI TensorFlow Developer Professional Certificate - Course 3
- Week 1: Sentiment in text
- Tokenizer
- OOV (out-of-vocab)
- Padding
- Post (after)
- Truncating
- Text to Seq
- Tokenizer
- Week 2: Word Embeddings
- Embedding
- WE Visualization Dashboard: http://projector.tensorflow.org/
- Hyperparameter tuning
- Splitting words into sub-tokens: case sensitive, punctuation kept
- Not working well with ffnn
- Week 3: Sequence models
- Bi-LSTM
- Multiple Layer Bi-LSTM
- CNN
- Sarcasm with Bidirectional LSTM
- Sarcasm with 1D Convolutional Layer
- Overfitting: the model tends to overfit the training data since there might be OOV words in the validation data
- Week 4: Sequence models and literature
- Text generation
- (Optional): text generation on character level