Text Generation with LSTM and Spell with Nabil Hassein
Nabil Hassein demonstrates how to train an "LSTM" neural network to generate text in the style of a particular author using Spell and ml5.js.
This stream is sponsored by Spell.
Sign up here: https://spell.run/codingtrain
"As creators of machine learning projects for art or otherwise, we have to take responsibility for what our programs produce and the impact that output has on people who interact with our creations. Given how common bias and oppression is in the world generally, many if not most datasets (including song lyrics) reflect that reality, and without countermeasures we as programmers are very likely to reproduce those harms. It is also worth explicitly noting that authorship and context matter, and identical words (or images, etc.) can assume completely different significance depending on who says them and when. I encourage everyone to take seriously the ethical aspects of the ml5.js documentation along with the technical material, and to consider your responsibility as a technologist to acknowledge and address the harm that the field of computing has too often caused for marginalized groups"
Nabil Hassein is a freelance technologist and educator based in Brooklyn, NY. He has previously worked as an infrastructure engineer at Khan Academy and a couple of startups, taught math and programming in both public schools and private settings, and occasionally writes and speaks. His website is https://nabilhassein.github.io.
🎥 Workflow: Python and Virtualenv: https://youtu.be/nnhjvHYRsmM
🎥 Introduction to Spell: https://youtu.be/ggBOAPtFjYU
🔗 ml5.js: https://ml5js.org
🔗 Generative-DOOM: https://nabilhassein.github.io/generative-DOOM/
🔗 The Unreasonable Effectiveness of Recurrent Neural Networks: http://karpathy.github.io/2015/05/21/rnn-effectiveness/
🔗 Understanding LSTM Networks: http://colah.github.io/posts/2015-08-Understanding-LSTMs/
🔗 Project Gutenberg: https://www.gutenberg.org/
🔗 Training a LSTM network: https://github.com/ml5js/training-lstm
🔗 ml5.js examples: https://github.com/ml5js/ml5-examples
🔗 p5.js: https://p5js.org
📄 Code of Conduct: https://github.com/CodingTrain/Code-of-Conduct
Nabil Hassein demonstrates how to train an “LSTM” neural network to generate text in the style of a particular author using Spell and ml5.js.
This stream is sponsored by Spell.
Sign up here: https://spell.run/codingtrain
“As creators of machine learning projects for art or otherwise, we have to take responsibility for what our programs produce and the impact that output has on people who interact with our creations. Given how common bias and oppression is in the world generally, many if not most datasets (including song lyrics) reflect that reality, and without countermeasures we as programmers are very likely to reproduce those harms. It is also worth explicitly noting that authorship and context matter, and identical words (or images, etc.) can assume completely different significance depending on who says them and when. I encourage everyone to take seriously the ethical aspects of the ml5.js documentation along with the technical material, and to consider your responsibility as a technologist to acknowledge and address the harm that the field of computing has too often caused for marginalized groups”
Nabil Hassein is a freelance technologist and educator based in Brooklyn, NY. He has previously worked as an infrastructure engineer at Khan Academy and a couple of startups, taught math and programming in both public schools and private settings, and occasionally writes and speaks. His website is https://nabilhassein.github.io.
🎥 Workflow: Python and Virtualenv: https://youtu.be/nnhjvHYRsmM
🎥 Introduction to Spell: https://youtu.be/ggBOAPtFjYU
🔗 ml5.js: https://ml5js.org
🔗 Generative-DOOM: https://nabilhassein.github.io/generative-DOOM/
🔗 The Unreasonable Effectiveness of Recurrent Neural Networks: http://karpathy.github.io/2015/05/21/rnn-effectiveness/
🔗 Understanding LSTM Networks: http://colah.github.io/posts/2015-08-Understanding-LSTMs/
🔗 Project Gutenberg: https://www.gutenberg.org/
🔗 Training a LSTM network: https://github.com/ml5js/training-lstm
🔗 ml5.js examples: https://github.com/ml5js/ml5-examples
🔗 p5.js: https://p5js.org
📄 Code of Conduct: https://github.com/CodingTrain/Code-of-Conduct
