>"Given both the competitive landscape and the safety implications of large-scale models like GPT-4, this report contains no further details about the architecture (including model size), hardware, training compute, dataset construction,training method, or similar."
I am not a fan of the "enjoy the fruits but not the science" statement they made in the blog. The quote above was from GPT-4's technical paper. I'll go the extra mile and steel man a counter-argument and say they're seeking a competitive advantage to secure more in investments so they can stick to their mission statements of making safe AGI, but I think we can all see where this line of thinking of is going further down the line. Building your stack around an academic research paper from a rival company then refusing to publish your contributions is terribly disappointing imo. I hope this won't be a trend that other companies participate in.
I am not a fan of the "enjoy the fruits but not the science" statement they made in the blog. The quote above was from GPT-4's technical paper. I'll go the extra mile and steel man a counter-argument and say they're seeking a competitive advantage to secure more in investments so they can stick to their mission statements of making safe AGI, but I think we can all see where this line of thinking of is going further down the line. Building your stack around an academic research paper from a rival company then refusing to publish your contributions is terribly disappointing imo. I hope this won't be a trend that other companies participate in.