Are you doing a Lora or full weights?

> I wanted to train Meta's LLaMA model on this data, but considering their license, I'm not sure if that is the best way. Suggestions will be appreciated.

If we ignore OpenAI's licence, is it ok to perhaps ignore Meta's licence as well? Or is that going too far

> The trained model will be open source, under MIT License.

Is the dataset going to be open source as well? So that other people can use it to train other models.