Alicia Curth and colleagues presented work arguing that double descent is an illusion caused by using parameter count as an inadequate measure of model complexity.
Notes on verification
Confirmed by arXiv preprint, NeurIPS poster page, and NeurIPS proceedings abstract, all consistent in describing the paper's argument that double descent arises from inadequate parameter-count-based complexity measures and resolves into classical U-shaped curves under an effective parameter count view. [tier=silver indep_score=0.925 clusters=2 claim_tier=notable]
Sources
- Large language models can do jaw-dropping things. But nobody ... (seed:technology_and_ai)
- https://arxiv.org/abs/2310.18988 (corroboration)
- https://neurips.cc/virtual/2023/poster/71830 (corroboration)
- https://proceedings.neurips.cc//paper_files/paper/2023/hash/aec5e2847c5ae90f939ab786774856cc-Abstract-Conference.html (corroboration)