Hey everyone! So, I’m diving into this ArabicNLP research project. The plan is to fine-tune a bunch of sota models and then evaluate them against each other etc. But here’s the kicker – we’re kinda tight on budget. Any hacks to figure out the cost? like, is there some formula that given the model size and datasets size (assuming the fietuning will be on an A100 80gb) can estimate the cost for us?