They have reportedly stopped testing on the open source Behemoth model, after it delivered «underwhelming» performance, TechCrunch reports.
«Unchanged» policy
They also caution that discussions are just discussions and don’t reflect official policy from the social media giant.
A spokesperson for Meta told TechCrunch that their policy remains unchanged, but Meta has also been known to use closed source models, like the ones powering the Meta AI assistant.
— We plan to continue releasing leading open source models, the spokesperson told the website. — We haven’t released everything we’ve developed historically, and we expect to continue training a mix of open and closed models going forward.
Open source can be dangerous
The idea of open sourcing models works well when you are playing catch-up, like Meta has been doing in the AI model/chatbot market, but as OpenAI recently found, it can be dangerous or irresponsible to release very advanced models with open weights.
Zuckerberg himself opined on this last summer, TechCrunch reports:
— If at some point, however, there’s some qualitative change in what the thing is capable of, and we feel like it’s not responsible to open source it, then we won’t. It’s all very difficult to predict.
So with a top notch team in place working on frontier models, the thinking at Meta may well be changing.
Read more: Original reporting by the New York Times, TechCrunch adds Meta spox detail.