I am a big believer in Boyd's Law, which states that speed of iteration beats quality of iteration. Following this principle, I created
cmn_ai to incorporate the most common tasks we encounter in ML/AI from data exploration to model development and training. Additionally, I baked into it some of the best practices I learned in my career so I don't have to repeat myself on every project I work on.
It is worth noting that the majority of the DL code such as
Learner and callbacks assume that
pytorch is used as the underlying DL library. Also, tabular data preprocessors/models are compatible with
sklearn so they can be used easily with its
Given the nature of the progress in ML/AI, I will always keep adding more functionalities as time passes.