$\sigma$Transfer: Uncertainty Transfer from Small to Large Networks under $\mu\mathrm{P}$
arXiv:2610.11668v1 Announce Type: new Abstract: Reliable predictive uncertainty in Laplace approximations depends critically on the prior precision, yet selecting it requires a posterior sweep that is prohibitively expensive for neural networks with billions of parameters. Under the Maximal Update…
Read the full story at arXiv stat.ML ↗
Timeline · 1 report
- 2026-10-09 04:00 · arXiv stat.ML
$\sigma$Transfer: Uncertainty Transfer from Small to Large Networks under $\mu\mathrm{P}$