Reasons why ChatGPT might fail but expert-mini-ChatGPTs will win
I wrote an entire thread about this idea - https://twitter.com/MJayatilake/status/1628807363644821506
Takeaway - The most impact and value will come from domain-specific, narrowed-down AI agents, rather than the general-purpose ones.
These will soon become integrated everywhere in our workflows.
Hey thoughts on how to get started fine tuning a model?
If you're looking at language models, I recommend checking out HuggingFace. They have good documentation on this and how to use the models.
But tbh, I haven't seen many good detailed tutorials on fine-tuning. So I plan to write one myself soon.
Sounds good, I'll be looking out for that if I haven't already figured it out by then
In case this helps - https://huggingface.co/docs/transformers/training
thx