nicklashansen/rnn_lstm_from_scratch
How to build RNNs and LSTMs from scratch with NumPy.
This educational project helps deep learning students and researchers understand how recurrent neural networks (RNNs) and Long Short-Term Memory (LSTM) networks process sequences. It takes structured sequences of text tokens as input and shows how these networks learn to predict the next token in the sequence. The output demonstrates the learning process and performance of different implementations, from basic NumPy to PyTorch, for those building or optimizing sequential models.
280 stars. No commits in the last 6 months.
Use this if you are a deep learning student or researcher who wants to understand the underlying mechanics of RNNs and LSTMs by building them from first principles.
Not ideal if you are looking for a plug-and-play solution to apply recurrent neural networks to your data, as this is an educational tool focused on implementation details.
Stars
280
Forks
72
Language
Jupyter Notebook
License
GPL-3.0
Category
Last pushed
Oct 04, 2020
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/nicklashansen/rnn_lstm_from_scratch"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
digantamisra98/Mish
Official Repository for "Mish: A Self Regularized Non-Monotonic Neural Activation Function" [BMVC 2020]
Sentdex/nnfs_book
Sample code from the Neural Networks from Scratch book.
itdxer/neupy
NeuPy is a Tensorflow based python library for prototyping and building neural networks
vzhou842/cnn-from-scratch
A Convolutional Neural Network implemented from scratch (using only numpy) in Python.
Synthaze/EpyNN
Educational python for Neural Networks.