So I have recently completed the course Generative AI with Large Language Models and in the labs for week 2 and week 3, the labs have used AWS’s pretrained PFET models. How can I train my own model instead of using an AWS checkpoint every time? Could somebody guide me through to a course I can take for that or something?
You could build on the course content to train using models in huggingface. The trend nowadays is to finetune a model from a checkpoint. Training from scratch is quite resource intensive.
1 Like
Hi @lukmanaj , thanks for the insights and yes, my plan is to use a model from huggingface. If you would refer to the screenshot I attached, you will find an exact clone of the course content. Now what I want to know is how I could do the training and apply rest of the tricks to get the job done without AWS’s own model. I want to do that part myself to be precise. If there are any courses or documentation I could refer to, kindly let me know. Cheers!
I think this will be helpful to you.
The short course shows how to explore the huggingface ecosystem.