A Prior-data Fitted Network with a scaling-law-specific prior gives better point and uncertainty predictions for neural scaling law extrapolation than MCMC, BNSL, and LC-PFN baselines.
LC-PFN includes its own normalization method for y-values, enabling it to predict learning curves across various ranges and directions
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Bayesian Neural Scaling Law Extrapolation with Prior-Data Fitted Networks
A Prior-data Fitted Network with a scaling-law-specific prior gives better point and uncertainty predictions for neural scaling law extrapolation than MCMC, BNSL, and LC-PFN baselines.