Auditability of subliminal learning is constrained by channel location, with initialization-dependent body channels allowing pre-training screens while vocabulary geometry and conditional body channels evade them.
Alex Warstadt, Amanpreet Singh, and Samuel R
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
verdicts
UNVERDICTED 5representative citing papers
FedSmoothLoRA improves federated LoRA fine-tuning by constructing local initializations from a round-matching matrix for cross-round continuity and a gradient-aligned matrix for client-specific guidance, yielding faster convergence than prior methods in image and text tasks.
HyperAdapt performs parameter-efficient fine-tuning by row- and column-wise diagonal scaling to induce high-rank updates with only n+m trainable parameters.
FACT is a three-phase hierarchical finetuning framework that reports over 20% gains on ViT models under low sampling ratios across image classification benchmarks.
TLoRA+ augments LoRA with a dedicated optimizer to improve fine-tuning performance on GLUE tasks without meaningful added compute.
citing papers explorer
-
Channel Location Constrains the Auditability of Subliminal Learning
Auditability of subliminal learning is constrained by channel location, with initialization-dependent body channels allowing pre-training screens while vocabulary geometry and conditional body channels evade them.
-
FedSmoothLoRA: Toward Smoother and Faster Convergence in Federated Low-Rank Adaptation
FedSmoothLoRA improves federated LoRA fine-tuning by constructing local initializations from a round-matching matrix for cross-round continuity and a gradient-aligned matrix for client-specific guidance, yielding faster convergence than prior methods in image and text tasks.
-
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.
-
FACT: A Simple and Efficient Framework for Active Finetuning
FACT is a three-phase hierarchical finetuning framework that reports over 20% gains on ViT models under low sampling ratios across image classification benchmarks.
-
TLoRA+: A Low-Rank Parameter-Efficient Fine-Tuning Method for Large Language Models
TLoRA+ augments LoRA with a dedicated optimizer to improve fine-tuning performance on GLUE tasks without meaningful added compute.