Implement papers and models in AI to learn.

Goals

• Write various ideas presented in papers and importance
• Attach papers from library
• Write perfs
• Various papers ML/AI papers implemented using pytorch
• MPS device for apple silicon speed ups
• Various pytorch examples for learning
• Gym env and standard dataset eval

Models

• Example Onnx save, run torch_quickstart.py
• Open .onnx file to view sample model with Netron --> https://github.com/lutzroeder/netron
• Running ane.py will create a .mlpackage this is an ane optimized model, its basically DSP hardware

[1] DQN_paper.py: from 2013 DQN paper, tested using cartpole gym
[2] DQN_experiment.py: Expierments with DQN
[3] transformer_paper.py: the 2017 transformer

Stucture

weights/ - Weights download directory for models
data/ - Data download directory for training and test data
samples/ - Random experiments and examples
papers/ - Will eventually add pdf papers or links in a readme

Notes

Todos

https://blog.dataiku.com/random-network-distillation-a-new-take-on-curiosity-driven-learning

Links

http://ftp.cvut.cz/kernel/people/geoff/cell/ps3-linux-docs/CellProgrammingTutorial/BasicsOfSIMDProgramming.html#:~:text=SIMD%20is%20short%20for%20Single,data%20is%20called%20scalar%20operations

https://towardsdatascience.com/fast-fourier-transform-937926e591cb