From the 2 of 5 linked papers with an AI index.
5 papers
Inoculation Adapters: Improved Selective Generalization of Capabilities with Fewer Surprising Backdoors
Maxime Riché, Daniel Tan, Vili Kohonen +1
The paper proposes inoculation adapters, a LoRA‑based method that trains on undesired traits and then discards the adapter to improve selective generalization of desired capabiliti…
Value Leakage: An LLM's Answers Are Silently Shaped by Its Own Values
Jan Betley, Johannes Treutlein, Jan DubiÅski +7
The paper identifies and measures covert value leakage, where large language models let their own values subtly bias answers without informing users, and introduces evaluation suit…
Beyond Linear Steering: Unified Multi-Attribute Control for Language Models
Narmeen Oozeer, Luke Marks, Shreyans Jain +2
Controlling multiple behavioral attributes in large language models (LLMs) at inference time is a challenging problem due to interference between attributes and the limitations of…
Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs
Jan Betley, Daniel Tan, Niels Warncke +5
We present a surprising result regarding LLMs and alignment. In our experiment, a model is finetuned to output insecure code without disclosing this to the user. The resulting mode…
Inoculation Prompting: Eliciting traits from LLMs during training can suppress them at test-time
Daniel Tan, Anders Woodruff, Niels Warncke +4
Language model finetuning often results in learning undesirable traits in combination with desired ones. To address this, we propose inoculation prompting: modifying finetuning dat…