I think the philosophical concept of Open Source can’t really work in ML models unless the training data is open as well. As it stands, these “open source” models are still very much a black box. Nobody was really questioning the implementation of the GPT.
I think the philosophical concept of Open Source can’t really work in ML models unless the training data is open as well. As it stands, these “open source” models are still very much a black box. Nobody was really questioning the implementation of the GPT.
Yeah this would be like Google saying Google Search was “open source” because map-reduce was open, or something.