Parameter-efficient fine-tuning with QLoRA and Hugging Face

In this sequel to my previous video ( • Summarizing legal docu... ), I delve into optimizing the fine-tuning of a Google FLAN-T5 model for legal text summarization. The focus is on employing QLoRA for parameter-efficient fine-tuning. All it takes is a few extra lines of simple code in your existing script.
This methodology allows us to train the model with remarkable cost efficiency, utilizing even modest GPU instances, which I demonstrate on AWS with Amazon SageMaker. Tune in for a detailed exploration of the technical nuances behind this process.
⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️
- Original model: huggingface.co/google/flan-t5...
- LoRA model: huggingface.co/juliensimon/fl...
- Dataset: huggingface.co/datasets/billsum
- Notebook: gitlab.com/juliensimon/huggin...
Follow me on Medium at / julsimon or Substack at julsimon.substack.com.

[@WhySoBroke]

Every minute of your videos is more valuable than a month in an expensive USA “public” university... very very grateful

[@juliensimonfr]

That's super nice, thank you. Happy to help!

[@AI_by_AI_007]

I remember one short year ago you did the "Analyze SEC filings with Transformers for fun and profit" video and I did that jumpstart as a demo for an internal audience at work --- it just blew people away... Then a month later Chat GPT came out. It feels a lifetime ago... So much has changed. Your work is terrific and your community building is legendary. I just want to say Thank you 🤘

[@juliensimonfr]

Thank you for the kind words. Keep rocking 😀