Ivan Habernal

Lecture 01 For some it was not clear enough (also the notation of z, y, etc. = better use w and x)

Feedback: > remembering is not really desired, but cf. deep double decent phenomenon https://arxiv.org/abs/1912.02292

Confusing: Examples of low-level NLP problems/tasks Example: Named Entity Recognition (NER)

Feedback: The loss function here makes little sense, will be discussed in detail in Lecture 2 anyway.

Feedback: Considered least relevant, heard too often, little relevance to NLP

Powerful for learning on graph-based data; not a mainstream in NLP, but a strong architecture.

would be nice to have them in the central repo

Lecture 2 was a bit detached from the introduction, not clear how minimizing a function is related at all. Add a big picture slide.

Currently we rely on Hadoop 2.6.0 which is present in AWS EMR 4.2.0. We should update to Hadoop 2.7.1 to keep up with the latest EMR version (4.4.0).

One missing package fixed it for me

1

I just had to install ```bash $ sudo apt-get install python-is-python3 ``` to fix `/usr/bin/env: "python": No such file or directory`. Maybe add it to the instructions.