News for hackers. Stuff that matters.

A comment on It's Time to Stop Taking Sam Altman at His Word. Parent comment

Re:It's Time to Stop Taking Sam Altman at His Word by jerpint on Saturday October 05, 2024 @01:20PM (#41751372)
Video data is still very much untapped and likely to unlock a step function worth of data. Current image-language models are trained mostly on {image, caption} pairs with a bit of extra fine tuning
[ Parent | on HN ]
Re:It's Time to Stop Taking Sam Altman at His Word by KPGv2 on Saturday October 05, 2024 @03:42PM (#41752371)
Do you think it matters that there's orders of magnitude less video (and audio) data than text data?
[ Parent | on HN ]
Re:It's Time to Stop Taking Sam Altman at His Word by jerpint on Sunday October 06, 2024 @05:36AM (#41755955)
I’m not sure I agree, text is full of compressed information but lacks all of the visual cues we all use to navigate and understand our world. Video data also has temporal components which text is really bad at.
[ Parent | on HN ]
Re:It's Time to Stop Taking Sam Altman at His Word by mike_hearn on Sunday October 06, 2024 @09:18AM (#41756977)
OpenAI has been training on YouTube for years.
[ Parent | on HN ]