Hello everyone - I have some YouTube resources I’m excited to share with you today.
Below you will find two videos on how to train/fine-tune commercially viable free and open-source AI/LLMs on your own dataset (Falcon & Llama-2).
I will soon be going through both of these workflows and sharing my results here.
AI Jason’s Falcon-7B Training & Fine-Tuning Tutorial
PromptEngineering’s Llama-2-7B Training & Fine-Tuning Tutorial
Most of the work is done in Google Colab - a cloud collaboration notebook that can run code. If you use the free tier, it may take longer to follow along - but the workflow is there.
Depending on your dataset and goals, it might be worth considering an upgrade to your plan when you’re ready to train a final version.
If you end up publishing a model, don’t be afraid to share them here!
Post your datasets too!
Update [9/23/23] This guide may be outdated! Check out the resources on the side bar or visit the HyperTech Workshop for an aggregated linkhub of resources you can explore (from training, tuning, deploying, and serving your on LLM at all stages see the 📺 YouTube section for those tutorials and amazing content creators!).
GL, HF, and happy devving!