[2206.04615] Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models

Abstract

Language models demonstrate both quantitative improvement and new qualitative capabilities with increasing scale. Despite their potentially transformative impact, these new capabilities are as yet poorly characterized. In order to inform future research, prepare for disruptive new model capabilities, and ameliorate socially harmful effects, it is vital that we understand the present and near-future capabilities and limitations of language models. To address this challenge, we introduce the Beyond the Imitation Game benchmark (BIG-bench). BIG-bench currently consists of 204 tasks, contributed by 450 authors across 132 institutions. Task topics are diverse, drawing problems from linguistics, childhood development, math, common-sense reasoning, biology, physics, social bias, software development, and beyond. BIG-bench focuses on tasks that are believed to be beyond the capabilities of current language models. We evaluate the behavior of OpenAI's GPT models, Google-internal dense transformer architectures, and Switch-style sparse transformers on BIG-bench, across model sizes spanning millions to hundreds of billions of parameters. In addition, a team of human expert raters performed all tasks in order to provide a strong baseline. Findings include: model performance and calibration both improve with scale, but are poor in absolute terms (and when compared with rater performance); performance is remarkably similar across model classes, though with benefits from sparsity; tasks that improve gradually and predictably commonly involve a large knowledge or memorization component, whereas tasks that exhibit "breakthrough" behavior at a critical scale often involve multiple steps or components, or brittle metrics; social bias typically increases with scale in settings with ambiguous context, but this can be improved with prompting.

Summary

Introduces BIG-bench, a collaborative benchmark of 204 diverse tasks designed to probe language model capabilities beyond standard benchmarks, including tasks believed to be beyond current model abilities. The paper evaluates models across scales and finds that performance is often unpredictable, with some tasks showing discontinuous 'emergent' improvements at certain model sizes, while others remain flat regardless of scale.

Key Points

• •BIG-bench comprises 204 tasks contributed by 450+ researchers, covering reasoning, knowledge, creativity, and social understanding to stress-test LLMs
• •Documents 'emergent' capabilities—behaviors that appear suddenly at certain model scales rather than improving gradually, raising concerns for capability forecasting
• •Finds that larger models generally outperform smaller ones but with high variance; some tasks show inverse scaling (larger models perform worse)
• •Introduces BIG-bench Hard, a subset of 23 challenging tasks where even large models fail to exceed human performance, for future benchmarking
• •Highlights limitations of current evaluations and provides tools for extrapolating future model capabilities from scaling trends