HETA is a new attribution framework for decoder-only LLMs that combines semantic transition vectors, Hessian-based sensitivity scores, and KL divergence to produce more faithful and human-aligned token attributions than prior methods.
arXiv preprint arXiv:2002.10689 , year=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
PPM injects parametric structural priors into generative models via a learnable mapping to improve probabilistic forecasts on non-stationary MTS data.
citing papers explorer
-
Hessian-Enhanced Token Attribution (HETA): Interpreting Autoregressive LLMs
HETA is a new attribution framework for decoder-only LLMs that combines semantic transition vectors, Hessian-based sensitivity scores, and KL divergence to produce more faithful and human-aligned token attributions than prior methods.
-
Parametric Prior Mapping Framework for Non-stationary Probabilistic Time Series Forecasting
PPM injects parametric structural priors into generative models via a learnable mapping to improve probabilistic forecasts on non-stationary MTS data.