Gary Marcus

Rank 14 of 47
|
Score 83
12/22/2024, 1:04:45 AM
X
In reply to:
David Balfour
@hal_incandenza4
·
375d
@GaryMarcus Chollet himself confirming this is all aboveboard
Gary Marcus
@GaryMarcus
·
375d
@hal_incandenza4
David Balfour
@hal_incandenza4
·
375d
@GaryMarcus It would be worth calling out the zero-shot performance of the model if that’s what they did. There’s no obligation to detail all the rules of what the challenge allows if they’re just talking about it as is. Anyone else can then learn more about what arc-agi is
David Balfour
@hal_incandenza4
·
375d
@GaryMarcus I did. The performance improvement is relative to other models - which are all trained on the public dataset. Training on the training set available for precisely that purpose is part of the challenge and if you’re not aware of the challenge parameters that’s on you
Gary Marcus
@GaryMarcus
·
375d
@hal_incandenza4 Watch the video
David Balfour
@hal_incandenza4
·
375d
@GaryMarcus They did exactly what every other model taking the arc benchmark does - the training set is meant to be trained on - there was nothing underhanded here - stop fishing

The statement engages in a discussion about the interpretation of a model's performance and the implications of training data usage, which is relevant to public discourse on AI transparency and ethics.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not appear to cause harm and focuses on clarifying misunderstandings. [+1]
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    It respects the privacy and dignity of others, avoiding personal attacks. [+1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement aims to promote understanding by clarifying the context of the model's performance. [+1]
  4. 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.
    Engages in constructive dialogue by addressing potential misunderstandings without resorting to personal attacks. [+1]