New adapter TFM-Retouche improves tabular foundation models without fine-tuning

Researchers have developed TFM-Retouche, a novel adapter designed to enhance tabular foundation models (TFMs) without requiring computationally expensive full fine-tuning. This lightweight, architecture-agnostic adapter operates in the input space, learning a small residual correction to better align data with the TFM's existing inductive biases. When applied to TabICLv2, the framework, named TabICLv2-Retouche, achieved top rankings on the TabArena-Lite benchmark, significantly improving aggregate Elo scores and maintaining efficiency. AI

IMPACT Introduces a more efficient method for adapting tabular foundation models, potentially improving their performance on diverse datasets without extensive retraining.

RANK_REASON This is a research paper introducing a new method for adapting tabular foundation models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• ConTextTab
• LoRA
• Mitra
• TabArena-Lite
• TabDPT
• TabICLv2
• TabICLv2-Retouche
• TabPFN-2.6
• Tabular Foundation Models
• TFM-Retouche

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →