What If Your LLM Could Become an Expert on Anything You Want?

Video Description

Fine Tune your Large Language Model with PEFT, QLora and Transformers library over your data. Substack: https://www.mayaakim.com/ Medium: https://medium.com/@mayaakim X: https://twitter.com/Maya_Akim Discord: https://discord.gg/GGhr7pyTHD To rent a GPU from Massed Compute ⤵️ https://bit.ly/maya-akim Code for 50% discount: MayaAkim In this video I'll turn a regular Llama 2 with 7B parameters into an expert on Youtube titles by first quantizing it with bitsandbytes library and training it later. Code is here: https://colab.research.google.com/drive/1Ud2vVdjxs18qPCXG334E5rZfvo9b3nUN?usp=sharing Resources for deeper understanding: 1. What is PEFT - https://huggingface.co/blog/peft 2. What is quantized model - https://huggingface.co/blog/4bit-transformers-bitsandbytes 3. What is quantized model p2. 2 - https://towardsdatascience.com/introduction-to-weight-quantization-2494701b9c0c 4. Lora - https://arxiv.org/abs/2106.09685 5. QLora - https://www.cloudbooklet.com/qlora-efficient-finetuning-of-quantized-llms/ 6. QLora pt.2 - https://towardsdatascience.com/qlora-fine-tune-a-large-language-model-on-your-gpu-27bed5a03e2b 7. Fine Tuning in general - https://bdtechtalks.com/2023/07/10/llm-fine-tuning/ Chapters: 0:00 - problem with llms 0:49 - what is fine tuning? 1:05 - overview of the video 1:38 - dataset 1:51 - libraries 3:02 - what is quantized model? 3:31 - training Llama 2 5:54 - testing the fine tuned model 7:15 - save the fine tuned model #finetune #llama2 #llm #ai