Wouldn't opensourcing the weights and trying to commoditize LLMs be a smarter move? That seems to be the only chance to catch up with OpenAI.
Depends on what they're afraid of.
If they want to commoditize the space, then I think you're right.
If they think they can catch OpenAI, and they want to charge for AI services, then what they're doing makes sense.
Then they wouldn't release anything to the researchers. Reason LLaMA weights are spreading in the first place is because they let a big group of people get access, someone is bound to upload it as a torrent.
Seems they gave people with .edu emails access relatively quickly too, researcher or not.
I'm sure FB has shared enormous amounts of data with researchers in the past. And they've probably had very little trouble with said researchers uploading data to the internet.
I think it's a pretty bad take that FB sharing data with researchers means that they wanted it spread freely. Even if that data is LLM weights.
But I'm sure that Facebook when sharing data that is more precarious, like social connections, have been more selective with who they share the data with, compared to the LLaMA dataset which seems to be shared 100% with everyone with a .edu domain, and almost without review for others.
Current Microsoft strategy seems so push into this direction. Open Source certain technologies and acquire important pieces e.g. Github / Stake in OpenAI etc. to build a bigger picture that they can monetize later.
OpenAI is a real threat to Google & Facebook