11 papers
The Piggyback Hypothesis of Generalization: Explaining and Mitigating Emergent Misalignment
Jiachen Zhao, Zhengxuan Wu, Aryaman Arora +3
The mechanisms behind LLMs' broad over-generalization beyond training examples remain unclear. Emergent misalignment (EM) offers a striking case study: finetuning on narrow tasks i…
Language Model Circuits Are Sparse in the Neuron Basis
Aryaman Arora, Zhengxuan Wu, Jacob Steinhardt +1
The high-level concepts that a neural network uses to perform computation need not be aligned to individual neurons (Smolensky, 1986). Language model interpretability research has…
PreFT: Prefill-only finetuning for efficient inference
Andrew Lanpouthakoun, Aryaman Arora, Zhengxuan Wu +4
Large language models can now be personalised efficiently at scale using parameter efficient finetuning methods (PEFTs), but serving user-specific PEFTs harms throughput, even with…
Verbalizing LLMs' assumptions to explain and control sycophancy
Myra Cheng, Isabel Sieh, Humishka Zope +7
LLMs can be socially sycophantic, affirming users when they ask questions like "am I in the wrong?" rather than providing genuine assessment. We hypothesize that this behavior aris…
ADAG: Automatically Describing Attribution Graphs
Aryaman Arora, Zhengxuan Wu, Jacob Steinhardt +1
In language model interpretability research, \textbf{circuit tracing} aims to identify which internal features causally contributed to a particular output and how they affected eac…
Mechanistic evaluation of Transformers and state space models
Aryaman Arora, Neil Rathi, Nikil Roashan Selvam +3
State space models (SSMs) for language modelling promise an efficient and performant alternative to quadratic-attention Transformers, yet show variable performance on recalling bas…