Getting Started with NLP and spaCy Transcripts
Chapter: Part 4: NLP with huggingface and LLMs
Lecture: ChatGPT for NLP

Login or purchase this course to watch this video and the rest of the course contents.
0:00 If you're interested in doing NLP these days though, there's also this other tool that by this time you probably will have heard of, called chat-gpt.
0:10 This represents a large language model, which gives you the option of communicating in natural
0:15 language in order to get knowledge that you're interested in. And that means that I am able to do stuff here that can also help me in my NLP project.
0:24 And just to give an example of something, suppose that I'm actually interested in finding Python packages that are very popular.
0:31 Well, one thing I could ask chat-gpt is to give me a list of those packages. I can give it this prompt, give me a list of popular packages, give 100.
0:42 Usually it's a good idea to add that to the prompt if you want to actually get 100 of them.
0:46 And if I were to now run this, then chat-gpt will happily generate me a long list of popular Python projects.
0:57 Now a small grain of salt needs to be applied here, because you could say that Jupyter Notebook
1:03 technically, you know, it's a Python tool so it's fair game in a way. But Jupyter Notebook is the name of a tool, not the name of the Python package.
1:12 And similarly when I scroll up here, there is a site called readthedocs which has lots of documentation on it, and it's a cool resource.
1:21 But again, it's not a Python package as far as I know. Although this list is plenty useful, because a list like this is something I can use to
1:29 generate me a pattern file that I can use inside of spaCy, it is good to know that if
1:34 you were using chat-gpt for something like this, that the list itself is probably not going to be 100% perfect.
1:40 But still, stuff like this can be immensely helpful. A tool like chat-gpt can be used to give me a long name list, and then from that point
1:49 on I can use this name list locally without communicating to a third-party vendor.
1:54 And stuff like that does make these LLMs pretty useful to have around when you're doing an NLP project.
2:00 But if you're willing to experiment a little, we can actually turn this up a notch as well.


Talk Python's Mastodon Michael Kennedy's Mastodon