Gary Marcus

Rank 19 of 47
|
Score 64
Naomi Saphra 🧈🪰
@nsaphra
·
199d
Do all roads lead to hierarchical generalization OOD? While in-distribution behavior is stable during training, OOD behavior is very messy! When training data mixes examples with and without center embeddings, rules compete, causing training instability and inconsistent outcomes.
12/21/2024, 1:12:51 PM
X

The statement and conversation engage in a technical discussion about language models, hierarchical generalization, and training dynamics, contributing to public discourse on AI and machine learning. The tone is academic and informative, aiming to share insights and research findings.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement strives to do no harm by focusing on academic research and findings. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    It promotes understanding by explaining complex concepts in AI and machine learning. [+2]
  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.
    Engages in constructive dialogue by referencing studies and encouraging further reading. [+1]