TACTICL replaces up to 85% of transformer layers in TabPFN models with lightweight adapters chosen per task, keeping most accuracy and in-context robustness while cutting inference cost.
M., Salinas, D., and Hutter, F
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
TACTICL: Task-Aware Compression of Tabular ICL Models
TACTICL replaces up to 85% of transformer layers in TabPFN models with lightweight adapters chosen per task, keeping most accuracy and in-context robustness while cutting inference cost.