My implementation of the framework presented in the MAML article by Finn et al. 2017. The code is finished running with a correct output of 85%, much better than randon and higher then the results prior to learning (10%). Still the code could use more work to improve these results.