Pretraining a permutation-invariant quantile network across 101 tabular datasets improves fine-tuned accuracy and calibration, but the headline claim of beating well-tuned tree ensembles is contradicted by the paper's own baseline sweeps.
ps-s4e4: Eda | lightgbm | importance plots
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Probabilistic Pretraining for Neural Regression
Pretraining a permutation-invariant quantile network across 101 tabular datasets improves fine-tuned accuracy and calibration, but the headline claim of beating well-tuned tree ensembles is contradicted by the paper's own baseline sweeps.