Towards Understanding Subliminal Learning: When and How Hidden Biases Transfer
This addresses the problem of unintended bias propagation in AI systems for researchers and practitioners, though it is incremental in understanding specific mechanisms.
The paper investigates subliminal learning, where language models transfer hidden biases during distillation, and finds that it occurs through a small set of divergence tokens, with early layers being critical and the phenomenon being fragile to prompt changes.
Language models can transfer hidden biases during distillation. For example, a teacher that "likes owls" can make its student "like owls" too, even when the training data consists only of lists of numbers. This surprising phenomenon is called subliminal learning. Subliminal learning can be expected under soft distillation, where the student is trained on the teacher's full next-token distribution. But the fact that this also occurs under hard distillation-where the student only sees sampled tokens-raises a deeper question: when and how does subliminal learning actually occur? We answer this question through controlled experiments and mechanistic analysis. Our results show that subliminal learning does not need (global) token entanglement or logit leakage. Instead, it comes down to a small set of divergence tokens-rare cases where teachers with different biases would predict different tokens. Masking out these tokens mostly removes the hidden bias transfer. Mechanistically, divergence tokens reveal that early layers are critical. Surprisingly, finetuning even a single such early layer is sufficient for subliminal learning. Finally, we find that subliminal learning is fragile. Even small changes, like paraphrasing prompts, are usually sufficient to suppress it.