LIME is 'locally interpetable model-agnostic explanations'. Just reproducing it, to learn about it.
The paper is at http://arxiv.org/pdf/1602.04938v1.pdf , by Ribeiro, Singh, and Guestrin. Ribeiro has a blog post about it at https://homes.cs.washington.edu/~marcotcr/blog/lime/ . There is code provided by Ribeiro at https://github.com/marcotcr/lime
There is thus already ample documentation and code about LIME, and this repo is for self-study purposes primarily.
There is a jupyter notebook at LIME.ipynb, which goes through what is LIME, and shows some explanations for one training example, using an rbf svm model.