Contents
- Attention Is All You Need (Paper)
- Neural Machine Translation of Rare Words with Subword Units (Paper): Proposed to use BPE in ML. Source code is also provided in GitHub.
- Byte Pair Encoding (blog): Covers step-by-step implementation with examples.
- Byte Pair Encoding (BPE) and Subword Tokenization (blog): Why subword tokenization is important is explained with step by step explanation and implementation.
CTC is used to train deep neural networks in speech recognition, handwriting recognition and other sequence problems.
- Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks. (Original paper)
- Sequence Modeling With CTC (blog): A good explanation with respect to speech recognition.
- An Intuitive Explanation of Connectionist Temporal Classification (blog): Discussed from the hand writing recognition view.
- Convolutional Neural Networks for Visual Recognition (blog): An in depth discussion on CNN and its operations. Includes both visual and mathematical concepts.
- Understanding LSTM Networks (blog): A good introduction to RNN and discussion of the problem. Then why LSTM is needed and block by block explanation.
- Weight Decay: basics with implementations (blog): Mathematical explanation with motivation behind weight decay and implementation form scratch.