Generative Python Transformer p.1 - Acquiring Raw Data

In a quest to teach neural networks via transformers to write Python code. Project name: Generative Python Transformers!

[libberator]

I was like "your cursor is hovering over get_tags! It's right there!" But like my dad yelling at the football ref on the tv, you couldn't hear me either

[clarencereyes]

Cannot wait to see your next vids! You have made Python very interesting and so much fun to learn. Cannot wait to buy your NNFS book really soon.

[bantaibaman]

I think I'm more excited about the next video coming up than I am for any series on Netflix that's half released.

[wuda-io]

Thats awesome!! I am pumped to see more 🔥 looks good bruh

[bubbleboy]

You always make videos of stuff I've always wanted to try

[theunblunt]

I needed some hands-on experience for my startup! Thanks!

[TheDeadking]

I like watching you make 'silly' mistakes because it makes me feel better when I'm constantly making mistakes

[robinferizi]

You can do a date - time range with minutes aswell. Example The hours minuites seconds go after the T

[jonasls]

Been watching for many years now, I enjoy this a lot. Just wanted to say hey 😃

[LeManneCris]

Super cool topic man! 34:17 I've been working on generating poetry with transformers, what works best when training the first time is to use some consequent lines of text to predict the following one, this produced better syntax when generating text (Even though as you say, it doesn't really make such a big difference). I think this could work on python code too. Can't wait for the next episode!!! ✌

[kunalkushwaha]

Can you make a transformer from scratch in tensorflow ?

[srg]

I have one system with two second order odes. I want to solve them and plot results. I am getting one of the results but not getting the other one as desired. Can you help or suggest how can I do it?

[habeang]

HELP!!! I want to built a chatbot that can learn from user and return quite a good respond, But I have a low pc like GTX 1060 3GB and I'm not very experience. Any idea which model to use? I tried Seq2seq model right now and hopefully get a result back soon I also tried DialogGPT but the respond accuracy is not acceptable for my case. I'm more experience with pytorch so anything related to it would be a big help.

[HemangJoshi]

Can you please provide link to this code ?

[chandrashekarreddy]

I think get_tags, get_git_tags object was right there on the screen the whole time and you missed those 😂

[Ash-bcvw]

which environment pr IDE is he using to code ?

[mamazu]

It's funny how you print out all methods and properties of the object just to overlook the functions you actually want / need. :D If you are already know that but you can just grep for the word in the program output and it gets highlighted (assuming your terminal allows colors)

eg. python some_file | grep "tags"

[RandomBananaInk]

Would the forks of popular projects(lots of duplicated code) cause problems with this type of learning?

[OscarFonsecaQ]

I just thought to do the same thing a few months ago. Impacient to see your results. ;)

[Cieric]

Probably a little late considering it's been almost a month since you made the video but ranges will get you all results for both days, best to just do a single day like "created:2021-04-10" and next I don't know if it's it but there was a get_tags entry in that list which sounds like a function and an extra api call. So it sounds better to avoid it anyways, but that's up to you and how important you think those tags are.