Learning and sharing my process with QLoRA (quantized low rank adapters) finetuning. In this case, I use a custommade reddit dataset, but you can use anything you want.
I referenced a LOT of stuff in this video, I will do my best to link everything, but let me know if I forget anything.
Resources:
WSBGPT7B Model: https://huggingface.co/Sentdex/WSBGP...
WSBGPT13B Model: https://huggingface.co/Sentdex/WSBGP...
WSB Training data: https://huggingface.co/datasets/Sentd...
Code:
QLoRA Repo: https://github.com/artidoro/qlora
qlora.py: https://github.com/artidoro/qlora/blo...
Simple qlora training notebook: https://colab.research.google.com/dri...
qlora merging/dequantizing code: https://gist.github.com/ChrisHayduk/1...
Referenced Research Papers:
Intrinsic Dimensionality Explains the Effectiveness of Language Model FineTuning: https://arxiv.org/abs/2012.13255
LoRA: LowRank Adaptation of Large Language Models: https://arxiv.org/abs/2106.09685
QLoRA: Efficient Finetuning of Quantized LLMs: https://arxiv.org/abs/2305.14314
Yannic's GPT4chan model: https://huggingface.co/ykilcher/gpt4...
Condemnation letter: https://docs.google.com/forms/d/e/1FA...
• GPT4chan: This is the worst AI ever
Contents:
0:00 Why QLoRA?
0:55 LoRA/QLoRA Research
4:13 Finetuning dataset
11:10 QLoRA Training Process
15:02 QLoRA Adapters
17:10 Merging, Dequantizing, and Sharing
19:34 WSB QLoRA finetuned model examples
Neural Networks from Scratch book: https://nnfs.io
Channel membership: / @sentdex
Discord: / discord
Reddit: / sentdex
Support the content: https://pythonprogramming.net/support...
Twitter: / sentdex
Instagram: / sentdex
Facebook: / pythonprogramming.net
Twitch: / sentdex