Likelihood Variance as Text Importance for Resampling Texts to Map Language Models
This enables scalable and efficient comparison of diverse language models for researchers and practitioners, though it is incremental as it builds on existing mapping techniques.
The paper tackles the high computational cost of constructing language model maps by proposing a resampling method that selects texts based on the variance of log-likelihoods across models, reducing the number of required texts by about half while preserving KL divergence accuracy.
We address the computational cost of constructing a model map, which embeds diverse language models into a common space for comparison via KL divergence. The map relies on log-likelihoods over a large text set, making the cost proportional to the number of texts. To reduce this cost, we propose a resampling method that selects important texts with weights proportional to the variance of log-likelihoods across models for each text. Our method significantly reduces the number of required texts while preserving the accuracy of KL divergence estimates. Experiments show that it achieves comparable performance to uniform sampling with about half as many texts, and also facilitates efficient incorporation of new models into an existing map. These results enable scalable and efficient construction of language model maps.