How to hack a LLM using PyReft (using your own data for Fine Tuning!)

🚀 Sign up to the newsletter

👨‍💻 Sign up for the Full Stack course and use YOUTUBE50 to get 50% off:

Hopefully you enjoyed this video.

Disclaimer: This has been developed for academic purposes. Nothing herein is financial advice, and NOT a recommendation to trade real money. Please use common sense and always first consult a professional before trading or investing.

Oh, and don't forget to connect with me!

Happy coding!
Nick

[aryamanarora]

Hey, I'm one of the authors of the ReFT paper and pyreft, thanks for making this awesome walkthrough! Let us know if you had any trouble setting anything up / you want any improvements to the documentation :)

[therealsirben]

Before I watch the video, just want to let you know that I was really waiting for you to do an LLM Finetuning video. Really excited for this one

[coreuped]

Slow tokenizers (use_fast=False) are those written in Python inside the Transformers library, while the fast versions (use_fast=True) are the ones provided by Tokenizers, which are written in Rust.

[aibchub]

Welcome Back ! The Nicholas's community need you to provide edge production topics like this

[TheGermanPlopis]

I love your Videos. I always learn new things. Thank you so much. You are such an awesome guy, please keep doing it, I always wait for new content from you.

[dhanushtharun]

Hi man i have one doubt you should clear that (book vs online tutorials/course) which is best
What you learn for programming.

[kenchang]

Excellent video that I need to try. Thank you for sharing.

[irshaadbahemia]

Hello Nicholas Renotte, just want to ask. Do you sometimes write machine learning algorithms from scratch or you only import the ml libraries such as Scikit-learn, Keras, Tensorflow?

[rishichowdhury]

Hey Nick I am getting this error while trying to train the pyreft model
TypeError: Object of type type is not JSON serializable

[jazzyfusion]

Great tutorial Nicholas, But I have a question, what happened if the Original Model changed, Llama3 -> Llama3.5, should we have to run all the process again?, I was looking at to create a DB embedding parallel to the Base Model, I'd like to hear your thought. Thanks

[souravbarua]

Please do make a video on fine tuning LLM with your own dataset using transformers trainer. Thank you.

[AdhilAseem]

How do we pass on an eval_dataset to the trainer?

[flynn]

Please can you do a TF project that utilises the dataset generators? They look quite complex and with datasets getting far larger now days, "streaming" it is the only option when testing locally. Thanks Nicholas!

[DrumAndSpaces]

I'm struggling to figure out the format for phi3 mini dataset. Any ideas? Can I still do it exactly like the video for my case?
Also great video perfect timing for just what I needed

[paulmiller]

Great video mate! How easy would this be to apply to the Llama 3 version of instruct 8b?

[flychuban]

Great video Nick! Do you know what params should I use when working on Macbook M3 Max. In the video you set_device('cuda').

[jameswhitaker]

Does this work with Mac M3 pro? Any parameters I’d need to change? Looks great!! Very excited about this

[yaqubnaqiyev]

How long have you been an ai engineer? I want to start i don't know where and what math and level of math needed. It would be very useful if you gives clear road map

[rafaeldesantis]

Hey Nicholas could you do a video about video classification ??? It’s been so hard to learn about it…

[devilgaming]

Bro can you do a project " speech to image translation using python"