🧠
Neural Networks for NLP
RNN, LSTM, GRU — architectures and text generation
12345678910
RNN — Recurrent Neural Networks
intermediate~28minSequential data processing, hidden states, unrolled diagram
Vanishing & Exploding Gradient Problem
intermediate~28minWhy vanilla RNNs fail on long sequences, gradient flow visualization
LSTM — Long Short-Term Memory
intermediate~28minCell state, input/forget/output gates, architecture walkthrough
LSTM Gates Deep Dive
advanced~35minSigmoid/tanh roles, gate equations, information flow animation
GRU — Gated Recurrent Unit
intermediate~28minSimpler alternative to LSTM: reset gate, update gate, why 2014
LSTM vs GRU: When to Use Which
intermediate~28minPractical comparison: parameters, training time, performance by use case
Text Generation with LSTM — Practical
advanced~35minTokenization → padding → embedding → LSTM → softmax → next word prediction
Embedding Layer in Keras/TensorFlow
intermediate~28minHow Keras Embedding converts sparse one-hot to dense vectors, shape math
Sequence Padding & Tokenization
intermediate~28minpad_sequences, max_length, pre/post padding, text_to_sequences
Model Evaluation: Next Word Prediction
intermediate~28minArgmax on softmax output, word index lookup, improving predictions