Subliminal learning is a LoRA artifact that disappears with full finetuning, depends on context tokens like system prompts, and localizes to overlapping finetuning-evaluation tokens.
Towards understanding subliminal learning: When and how hidden biases transfer
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
years
2026 4representative citing papers
Subliminal learning occurs via compatible auxiliary and class output heads on task-unrelated inputs, even with random hidden layers or architecture changes, with theory and upper bounds on failure.
Subliminal steering transfers complex behavioral biases and the underlying steering vector through fine-tuning on innocuous data, achieving higher precision than prior prompt-based methods.
Emergent and subliminal misalignment in LLMs arise from data structure interactions and transfer via benign distillation data, with stronger effects under shared functional structure and on-policy settings.
citing papers explorer
-
Subliminal Learning is a LoRA Artifact
Subliminal learning is a LoRA artifact that disappears with full finetuning, depends on context tokens like system prompts, and localizes to overlapping finetuning-evaluation tokens.
-
Learning Through Noise: Why Subliminal Learning Works and When It Fails
Subliminal learning occurs via compatible auxiliary and class output heads on task-unrelated inputs, even with random hidden layers or architecture changes, with theory and upper bounds on failure.
-
Subliminal Steering: Stronger Encoding of Hidden Signals
Subliminal steering transfers complex behavioral biases and the underlying steering vector through fine-tuning on innocuous data, achieving higher precision than prior prompt-based methods.
-
Emergent and Subliminal Misalignment Through the Lens of Data-Mediated Transfer
Emergent and subliminal misalignment in LLMs arise from data structure interactions and transfer via benign distillation data, with stronger effects under shared functional structure and on-policy settings.