Some implementations of popular learning models from scratch...
- Word2Vec: making a word embedding and then test it using skip-gram model with python and tensorflow.
- SenGen: predicting next word and generating sentences using dynamic lstm rnn and gensim embedding with python and tensorflow.
- OCR: converting image to text using convolutional and recurrent layers by optimizing a ctc loss with python and tensorflow.
- GAN: the popular generative adversarial networks for generating images based on a training data with python and tensorflow.