A new probing framework detects moderate parametric memorization signals in tabular in-context learning models under single-task fine-tuning, strongest on low-cardinality tasks, but signals largely disappear under realistic training.
Soufiane Hayou, Nikhil Ghosh, and Bin Yu
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
verdicts
UNVERDICTED 3representative citing papers
HyperAdapt performs parameter-efficient fine-tuning by row- and column-wise diagonal scaling to induce high-rank updates with only n+m trainable parameters.
Alpha in LoRA outperforms learning-rate scaling, follows a square-root law with rank, and enables a minimalist LoRA-alpha method that improves performance across tasks.
citing papers explorer
-
Probing Memorization of Tabular In-Context Learning
A new probing framework detects moderate parametric memorization signals in tabular in-context learning models under single-task fine-tuning, strongest on low-cardinality tasks, but signals largely disappear under realistic training.
-
HyperAdapt: Simple High-Rank Adaptation
HyperAdapt performs parameter-efficient fine-tuning by row- and column-wise diagonal scaling to induce high-rank updates with only n+m trainable parameters.
-
The Hidden Power of Scaling Factor in LoRA Optimization
Alpha in LoRA outperforms learning-rate scaling, follows a square-root law with rank, and enables a minimalist LoRA-alpha method that improves performance across tasks.