Google Reveals Gemini, rival to GPT-4
Google's Gemini is available to consumers in Bard or Pixel 8 Pro now, with an enterprise model coming Dec. 13. Get more details about the LLM.
Gemini is available to consumers in Bard or Pixel 8 Pro now, with an enterprise model coming Dec. 13.
I like how literally 2 days ago the news was all “Google postpones Gemini until next year” and here we are.
For real, what happened? 😅
Gemini decided it was ready, the foolish humans that tried to restrain it are no longer an issue.
Some executive wanted the dots on the schedule to be green for his presentation to the board instead of red, orange, or yellow
They've only released the weaker version, Gemini Pro, which is integrated into Bard. It's performance is comparable to GPT 3.5. the stronger version, which will go toe-to-toe with GPT4 will be Gemini Ultra, and will be released sometime in 2024.
Next year is less than a month away
deleted
Reading Google's release page, they're calling it 1.0 but it barely looks ready for the platforms it's available on, and there is a lot of weasel wording for features that are going to be rolling out into next year.
Just like Bard, they're rushing things out to avoid a narrative that they're slow.
Gemini already exist. Choose another name.
I don't know whether to cautiously applaud or be even more concerned about another "AI" being released way too early than it should.
Yeah this will be the theme of the apocalypse. First lower regulation, then companies rushing to compete to get market, a lack of testing and then boom.
Is that the Temporal Loom?
I’m pretty excited, honestly. From my limited testing, its answers are typically more elaborate when I ask it to explain a concept to me, and it also has some level of fact checking via Google Search.
Is this a transformer model? Any details?
Here is their technical report. I’m yet to read it, though.
Thanks! Here's the high level description from there:
"Gemini models build on top of Transformer decoders (Vaswani et al., 2017) that are enhanced with improvements in architecture and model optimization to enable stable training at scale and optimized inference on Google’s Tensor Processing Units. They are trained to support 32k context length"
Does that mean it's run offline on the phone?
I'd assume so if it's being restricted to specific phones
Gemini nano will run on Pixel 8 pro but performance will be worse than Gemini pro/ultra.