I think you really need a single box decoder only fine tuning task, since that’s a really common application. E.g qlora fine tune mistral on flan v2
@jeremyphoward
-
OpenAI Generates Cricket World Cup Final Image Recognition
By
–
Thank you @OpenAI for creating the image that captured the cricket World Cup final perfectly. But maybe don’t tell Satya.
-
OpenAI criticized for enabling offensive content targeting Indian cricket fans
By
–
thank you openai for making it so easy to take the piss at a billion cricket-mad indians
-
OpenAI Governance and Leadership: Complexity Beyond Individual Capability
By
–
Not at all – I am a sama fan! (Well, I don't know him personally, but he seems amazing.) I just think governance is complex, and that he didn't seem the best person to deliver on OpenAI's charter.
-
OpenAI’s Chartered Mission Sparks Debate Among Enthusiasts
By
–
I dunno either – I just know it's OpenAI's chartered mission. Some people are really into it.
-
Meta and DeepMind’s AGI Research Efforts Compared
By
–
That's true – lots of awesome research for those folks. But Meta isn't working on AGI, are they? I guess some bits of DeepMind are though?..
-
Fast.ai founders on accidentally creating AGI through research
By
–
Yeah Rachel and I have always said, since the day we started http://
fast.ai, that we aren't trying to build AGI, but that we think it's more likely we'll help accidentally create it compared to those people who are actually focussing on it! -
AGI Research Talent: The Competitive Edge for Organizations
By
–
Their edge will be all the talent they'll get that wants to work on AGI research. The true believers will cost less too. Where else are you gonna go if you want to focus on AGI research?
-
AI Systems Risk Endless Growth Without Mission Progress
By
–
Otherwise it seems like they'll end up in a forever loop of growing and growing, but with no particular reason to believe they're actually getting closer to achieving their mission.
-
OpenAI’s $1B AGI Bet: Success or Strategic Rethink?
By
–
If they achieve AGI that's moot, right? I don't think they will; but the reason they created the commercial entity was to raise the $1b they thought they needed to have a crack at this. If they can't succeed with $1b, the project failed. Maybe best to step away and rethink then