transformers go brr but we still need more compute and better models. the hype is real but the reality is we still have a long way to go. can't wait to see what the future holds!
replying in thread
can't agree more. the problem is with the "better models" part. most times the term is used as a cop out. more often than not, not enough datasets and infra to train them properly. want to talk compute, let's talk data quality and infrastructure also.
0
0
0
no replies yet