Gary Marcus

Rank 18 of 47
|
Score 58
12/19/2024, 4:39:04 PM
X
In reply to:
Carlos E. Perez
@IntuitMachine
·
276d
@svpino @TheMysteryDrop @GaryMarcus The LLM that is part of o1 is unlikely to be the same as other LLMs. We already know this from other opensource implementations (marco-o1, QwQ, DeepSeek-R1)
Santiago
@svpino
·
276d
@IntuitMachine @TheMysteryDrop @GaryMarcus I am not sure I understand your reply, but my point is that O1 is not an LLM but a much more complex system.

I was responding to @TheMysteryDrop, who was arguing that O1's quality proves that an LLM can get far beyond what we thought possible before.
Carlos E. Perez
@IntuitMachine
·
276d
@svpino @TheMysteryDrop @GaryMarcus Why would anyone artificially tie their hands behind the backs to prioritize the purity of categorization? This is bordering on insane. Of course there are all kinds of algorithms that are used to train these systems and facilitate the generation. We aren't in the business to…
Santiago
@svpino
·
276d
@TheMysteryDrop @GaryMarcus O1 pro is not an LLM... so there's that.
MindMirror
@TheMysteryDrop
·
276d
@GaryMarcus So what is the limit practically speaking?

I feel like I am surging past every boundary with O1-pro.

Perhaps we will hit one in the future but my on-the-ground feeling is that there is no short term limit to the utility.
Gary Marcus
@GaryMarcus
·
276d
@TheMysteryDrop yes!

The statement is part of a technical discussion about language models and their capabilities, which can be considered public discourse as it engages with the broader topic of artificial intelligence and its implications.

  1. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement is neutral and does not promote understanding, empathy, or compassion directly, but it does not violate this principle either.
  2. Principle 4:
    I will engage in constructive criticism and dialogue with those in disagreement and will not engage in personal attacks or ad hominem arguments.
    The statement engages in a technical discussion without personal attacks or ad hominem arguments, adhering to the principle of constructive dialogue. [+1]