@WhiteFang_Jr @Logan M on a slightly different topic. i'd like to tackle 'legal reasoning' using llama-index. prompting v fine-tuning v rag'ing. evaluate. basically, teach the model how to reason using legal logic and legal research principles. can any of you act as a second pair of eyes, provide feedback, may be co-author the paper, if there will be any. i 'll do the work. i'll just need someone to bounce the ideas off and review what i am doing from time to time.
ps. this was the impetus if you are curious what made me start thinking about it but there have been many more attempts, there is even now a benchmarking tool ('LegalBench') on legal stuff for various models. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=289789