Gary Marcus

Rank 35 of 47
|
Score -20
7/23/2024, 4:37:28 PM
X
In reply to:
Jason Gu
@jasonprompts
·
105d
@GaryMarcus You do realize our current evals and benchmarks are nearly maxed out? We can't go above 100% on these tests?
Gary Marcus
@GaryMarcus
·
105d
@jasonprompts And yet barely beat gpt-4. certainly the advance here is not remotely like the advance from GPT-3 to GPT-4 (or 2 to 3).
Jason Gu
@jasonprompts
·
105d
@GaryMarcus What wall? Meta literally just threw data and compute at this
Gary Marcus
@GaryMarcus
·
105d
LLaMa 3 450B is huge. Here’s why:
– Roughly as good as GPT-4, but open source.
- There goes OpenAI’s business model
- But it’s not GPT-5 level either, despite truly massive resources.
- Sure looks deep learning is hitting a wall, pending some kind of new breakthrough. Synthetic…
Marco Mascorro @ ICML 2024
@Mascobot
·
105d
LLaMa 3.1 benchmarks side by side.

This is truly a SOTA model.

Beats GPT4 almost on every single benchmark.

Continuously trained with a 128K context length.

Pre-trained on 15.6T tokens (405B).

The fine-tuning data includes publicly available instruction datasets, as well as…

The statement engages in a technical discussion about the performance and evaluation of AI models, specifically referencing benchmarks and diminishing returns. It contributes to the public discourse on the development and evaluation of AI technologies.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not cause harm and is focused on a technical discussion, which aligns with the principle of striving to do no harm. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement promotes understanding by referencing data and inviting readers to a more detailed discussion on Substack, which aligns with promoting understanding, empathy, and compassion. [+1]
  3. 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 dialogue by providing additional resources for further discussion, which aligns with engaging in constructive criticism and dialogue. [+1]