Getting Started with NLP and spaCy Transcripts
Chapter: Part 4: NLP with huggingface and LLMs
Lecture: Text classification
Login or
purchase this course
to watch this video and the rest of the course contents.
0:00
Another way to explore Hugging Face models is to explore organizations.
0:06
So Explosion, which are the creators of Spacey, they have a bunch of models hosted here as well.
0:12
And as we can see, there's a couple of these models that do token classification, but there are also other models that do text classification instead.
0:22
As always, there's a button on the right hand corner over here that I can go ahead and click,
0:27
and this will give me the code that I need to run this model locally if I wanted to.
0:32
This particular model is a bit different than the models that we've seen so far in the sense that it actually performs a different task.
0:40
So far in this series of videos, we've been interested in taking texts like this, and then being able to highlight substrings.
0:46
So I'm able to say Vincent in this sentence is a person. This is a kind of task that you might have when you're doing NLP.
0:55
But sometimes it's not so much finding things inside of a sentence, sometimes it's saying well, I've got this bit of text, what kind of category is it?
1:04
So in that case, text might go in, and then you might have tags let's say. In this case, maybe there's anger in this text. Yes, no.
1:15
And there are some other emotions that we have over here. And this is of course different than named entity recognition.
1:21
This is more of a categorization kind of a task. Spacey can be trained for this, and there are projects that you can go ahead and copy
1:29
if that's something you're interested in. But there are also other projects that can do this for you as well.
1:35
And again, just to give an impression of how many models there are, Hugging Face at the time of making this recording has over 55,000 of them.
1:43
With that, I would also like to remind you that not every model here is going to be as relevant.
1:48
Partially, that's because maybe there are some labels that are being used in this model that don't fit your use case.
1:56
But moreover, you should also remember that anyone can upload a model as they see fit.
2:01
So it might also be the case that not every model that's listed here has the quality standards that you need from the get-go.