Looking At Your Data
Every data set should be accompanied by one of these
What is a data sheet?
You need me before you can even pick a spaCy model for a random text
What is language identification?
The two-stage machine learning loop
What is train/test?
I look here first for data sets and models
What is huggingface?
I am a student at Colby who has worked with NLP at a startup
Who is David Istrati?
If I want to know how a dataset was labeled, I refer to these
What are the annotation guidelines?
People think I find words
What is tokenization?
When we use annotated data for machine learning
A next AI-related course I could take at Colby
What is CS 251/2?
Who is Jon Godbout?
A common issue with data extracted from structured documents
What is lost format? (or lost context)
I'm there if you need word classes
What is part of speech tagging?
When we take a model trained to read language, and train it more for our task
What is finetuning?
A next NLP-related course I could take for fun
What is the huggingface course? (or what is fast.ai?)
I am a linguist who systematically takes apart the field of NLP
Who is Emily Bender?
Data, split by label
What is stratified sampling?
I connect names of things to the real world
What is entity linking?
A performance metric that is the harmonized mean of two others
What is F1?
I look here first for benchmark results and paper links for a dataset
What is papers with code?
After a PhD in remote and participatory sensing and computer vision, I now work at Colby
Who is Tahiya Chowdhury?
Splitting data repeatedly to assess significance
What is cross validation?
Not supervised, I find basic statements in a sentence
What is open information extraction?
A useful way to analyze how well a model is performing class by class
What is a confusion matrix?
I look here first for core NLP papers
What is the ACL anthology?
Starting my career in computer vision, I was fired for my paper about large language models
Who is Timnit Gebru?