In the language domain, as in other domains, neural explainability takes an
ever more important role, with feature attribution methods on the forefront.
Many such methods require considerable computational resources and expert
knowledge about implementation details and parameter choices. To facilitate
research, we present Thermostat which consists of a large collection of model
explanations and accompanying analysis tools. Thermostat allows easy access to
over 200k explanations for the decisions of prominent state-of-the-art models
spanning across different NLP tasks, generated with multiple explainers. The
dataset took over 10k GPU hours (> one year) to compile; compute time that the
community now saves. The accompanying software tools allow to analyse
explanations instance-wise but also accumulatively on corpus level. Users can
investigate and compare models, datasets and explainers without the need to
orchestrate implementation details. Thermostat is fully open source,
democratizes explainability research in the language domain, circumvents
redundant computations and increases comparability and replicability.

Thermostat 是一个大型的模型解释数据集和分析工具，拥有超过 200k 种来自不同 NLP 任务的杰出 state-of-the-art 模型决策解释，该数据集的编译时间超过了 1 年，但现在可以使社区节省计算时间，降低了冗余计算以及增加可比性和可复制性。

Thermostat：大规模 NLP 模型解释和分析工具集合

Thermostat: A Large Collection of NLP Model Explanations and Analysis  Tools

We discuss the use of a Langevin equation with a colored (correlated) noise
to perform constant-temperature molecular dynamics simulations. Since the
equations of motion are linear in nature, it is easy to predict the response of
a Hamiltonian system to such a thermostat and to tune at will the relaxation
time of modes of different frequency. This allows one to optimize the time
needed to thermalize the system and generate independent configurations. We
show how this frequency-dependent response can be exploited to control the
temperature of Car-Parrinello-like dynamics, keeping at low temperature the
electronic degrees of freedom, without affecting the adiabatic separation from
the vibrations of the ions.

本文研究使用具有相关噪声的 Langevin 方程进行恒温分子动力学模拟，并探讨了利用其频率依赖响应来控制 Car-Parrinello 动力学的温度的方法。