DMDIntel: Interpreting Large Language Models via Dynamic Mode Decomposition
In this work, we introduce DMDIntel which uses dynamic mode decomposition (DMD) to make the predictions made by LLMs in a classification task interpretable. It develops an input attribution pipeline, that first decomposes the hidden states of an LLM into prominent patterns, also known as modes, and then associates ranks to the input tokens based on the projection values on those modes. Rigorous experiments across three datasets and three model families consistently show that the ranked attribution of input tokens obtained using DMDIntel by far outperforms state-of-the-art techniques such as principal component analysis, integrated gradients and SHAP.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Transparency at the Source: Evaluating and Interpreting Language Models With Access to the True Distribution
We present a setup for training, evaluating and interpreting neural language models, that uses artificial, language-like data. The data is generated using a massive probabilistic grammar (based on state-split PCFGs), tha…
Language ModellingInterpreting and Technology: Is the Sky Really the Limit?
Nowadays there is a pressing need to develop interpreting-related technolo-gies, with practitioners and other end-users increasingly calling for tools tai-lored to their needs and their new interpreting scenarios. But, a…
Self-Supervised Decomposition, Disentanglement and Prediction of Video Sequences while Interpreting Dynamics: A Koopman Perspective
Human interpretation of the world encompasses the use of symbols to categorize sensory inputs and compose them in a hierarchical manner. One of the long-term objectives of Computer Vision and Artificial Intelligence is t…
DisentanglementProbing Multimodal LLMs as World Models for Driving
We provide a sober look at the application of Multimodal Large Language Models (MLLMs) in autonomous driving, challenging common assumptions about their ability to interpret dynamic driving scenarios. Despite advances in…
Autonomous DrivingTrajectory PlanningMockConf: A Student Interpretation Dataset: Analysis, Word- and Span-level Alignment and Baselines
In simultaneous interpreting, an interpreter renders a source speech into another language with a very short lag, much sooner than sentences are finished. In order to understand and later reproduce this dynamic and compl…