Gary Marcus

Rank 13 of 47
|
Score 83
12/22/2024, 4:37:02 AM
X
In reply to:
Merovingian
@f4talStrategies
·
387d
@GaryMarcus Gary, I see you’re now framing this as an issue of transparency. But OpenAI clearly disclosed the use of ARC’s public training set—fine-tuning on training data is standard in ML benchmarks. Misinterpretations of their process stem more from bad framing by critics than a lack of…
Gary Marcus
@GaryMarcus
·
387d
@f4talStrategies Sir, you should read what i actually wrote earlier tonight, 100% clarifying that the my problem was NOT with ARC and not with OpenAI’s administration of ARC, but with the report which lacked transparency.

your argument against what you think i said would be fine – but it’s not…
Merovingian
@f4talStrategies
·
387d
@GaryMarcus Fair enough, Gary. Here’s my main critique: You keep insisting that fine-tuning on publicly available training data invalidates ARC—yet ARC’s Challenge Set was expressly designed to filter out shortcut-solvable questions.

You also cite critiques (Nie et al., Bengio et al.) aimed…
Gary Marcus
@GaryMarcus
·
387d
@f4talStrategies what was your best direct critique? I didn’t say 100% and didn’t single you out. if you are in the 2%, good for you!
Merovingian
@f4talStrategies
·
387d
@GaryMarcus Lol, Gary, I’ve made multiple direct critiques of your claims, and you’ve dodged all of them. Maybe engage with the 98% of critiques you’re conveniently ignoring before complaining about ad hominem.
Gary Marcus
@GaryMarcus
·
387d
Same same: “a lot of criticism of Gary but little critique of his specific claims.”

I love the 2% that actually engages with real claims.

The fact the rest is ad hominem speaks volumes.

The statement is part of a public discourse on the transparency of OpenAI's processes and the critique of Gary Marcus's claims. It engages in a debate about the disclosure of information and the interpretation of critiques, which are relevant public issues in the field of machine learning and AI ethics.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not appear to cause harm directly, but the tone is somewhat confrontational, which could be seen as slightly negative. [-1]
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    The statement does not engage in cyberbullying, harassment, or hate speech, but the tone is dismissive, which could be seen as slightly disrespectful. [-1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement does not promote understanding, empathy, or compassion, as it is more focused on challenging the other party. [-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.
    The statement engages in dialogue but is somewhat confrontational, which could be seen as not fully constructive. [-1]