Gary Marcus

Rank 20 of 47
|
Score 54
5/6/2024, 1:52:38 PM
X
In reply to:
Asankhaya Sharma
@asankhaya
·
422d
@tunguz @GaryMarcus Very similar performance across models. If anything things seem to be converging at GPT-4 class of models.
Asankhaya Sharma
@asankhaya
·
422d
@tunguz @GaryMarcus This^ I don't see the issues he is mentioning either. There is no need for prompt contortions. In fact same prompts work well across models now, e.g. in https://github.com/patched-codes… we allow users to switch models and most of time the workflow works as is.
Bojan Tunguz
@tunguz
·
422d
@GaryMarcus Skill issue.
Gary Marcus
@GaryMarcus
·
422d
Encouraged by a friend I tried out some LLMs this weekend, first time in a while.

They are just so erratic; hardly anything feels replicable, within or between models.

How anyone is supposed to engineer real-world solutions on that foundation, I really don’t know.

The statement in question is part of a technical discussion on the performance similarities and differences between various machine learning models. The user is questioning the accuracy and similarity of results presented in a table, emphasizing the need for clarity and source citation for the data provided. This contributes to a broader conversation about the reliability and comparability of machine learning models.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement aims to clarify and improve the understanding of the data presented, which aligns with the principle of doing no harm through misinformation. [+1]
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    The user respects the dignity of others by focusing on the content of the data rather than personal attributes or qualifications of those involved in the discussion. [+1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The critique is presented in a way that promotes understanding and seeks further explanation, which fosters a more informed discussion. [+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 constructive criticism by pointing out the lack of sources and detailed explanation, which encourages more rigorous standards in public discourse. [+1]
  5. Principle 5:
    I will acknowledge and correct my mistakes.
    The user acknowledges potential issues in the data presentation and seeks correction or clarification, aligning with the principle of acknowledging and correcting mistakes. [+1]
  6. Principle 6:
    I will use my influence for the betterment of society.
    By questioning the data's accuracy and presentation, the user uses their influence to advocate for transparency and accuracy in public discourse related to machine learning. [+1]