My argument about releasing before perfection, is not about releasing things that don't work correctly. The most important thing to me, is that what is released gives the correct information in a timly manner. This is largely governed by the LLM that is being used. Not sure how much configuration goes into using those models, or is it just what you get is what you get? If the information being shared is correct, then get it out there. If we're waiting because there is still development on how the DH's hair moves when the head moves, or how the nostrils should flare with certain mouth movements, then this is the sort of stuff that I think we should not wait for. There is competition out there, and we don't want to miss the boat while striving for perfection of things, that while important, should not make or break a deal being signed...
