Lost in Inference: Rediscovering the Role of Natural Language Inference for Large Language Models
This work addresses the problem of evaluating LLMs for researchers by revisiting NLI tasks, showing incremental improvements in benchmarking methods.
The paper investigates whether natural language inference (NLI) tasks, which are rarely used for evaluating large language models (LLMs), can effectively discriminate between models of different sizes and training stages, finding that they perform well and are not saturated, with model distributions aligning more closely with human labels as scale increases.
In the recent past, a popular way of evaluating natural language understanding (NLU), was to consider a model's ability to perform natural language inference (NLI) tasks. In this paper, we investigate if NLI tasks, that are rarely used for LLM evaluation, can still be informative for evaluating LLMs. Focusing on five different NLI benchmarks across six models of different scales, we investigate if they are able to discriminate models of different size and quality and how their accuracies develop during training. Furthermore, we investigate the extent to which the softmax distributions of models align with human distributions in cases where statements are ambiguous or vague. Overall, our results paint a positive picture for the NLI tasks: we find that they are able to discriminate well between models at various stages of training, yet are not (all) saturated. Furthermore, we find that while the similarity of model distributions with human label distributions increases with scale, it is still much higher than the similarity between two populations of humans, making it a potentially interesting statistic to consider.