Toggle navigation
JUPYTER
FAQ
View as Slides
View as Code
Python 3 Kernel
View on GitHub
Execute on Binder
Download Notebook
stat-nlp-book
chapters
Notebook
Assignments
¶
Assignments
¶
Require programming, math, and analysis
3 Assignments, with 30/40/30 weights, where the last assignment is a group assignment:
Generative Modelling with Smoothing (due Nov 1)
Discriminative Modelling with Feature Engineering (due Nov 29)
Representation Learning (due Jan 10)
Usual rules of plagiarism apply
Will use automatic evaluation of your scores on NLP tasks
Late Penalties
¶
Details
here
Up to 2
working days
late: 10% points deduction, but no lower than 50%
2-5 working days late: Mark capped at 50%
More than 5 working days: 0%
Plagiarism
¶
Don't do it
Don't enable it
Check
rules
if unclear
Hints
¶
For good "soft scores": Engage with the data (don't just try approaches X, Y, Z until get better results)
the difference between >80 and >90 can boil down to this
Try to quickly produce a passable solution first
Think about how robust your results (e.g. do they depend on the specific train/dev set split?)
Write-up: be concise, structure your text, motivate decisions
You are allowed to use both development and training set as "larger training set" but understand the potential drawbacks.
Assignment 1
¶
Language Modelling
On
Moodle