Well, Gumbel Distribution is magical. Basically, given a sequence of K logits, i.e., "\log a_1, \log a_2, ..., \log a_K" and K independent gumbel random variables, i.e., "g_1, g_2, ..., g_K". We have

\argmax_i \log a_i + g_i ~ Categorical({a_i / sum(a)})

This gives you a very simple way to sampl

Our existing test suite has been written with the Python unittest framework. However, as the test suite has grown and opsdroid has become more complex we are running into issues with the tests. Mainly around setting up and tearing down tests.

The @opsdroid/maintainers team have decided that we want to migrate all tests to be written with the