Large language models (LLMs) have been shown to be able to perform new tasks based on a few demonstrations or natural language instructions. While these capabilities have led to widespread adoption, most LLMs are developed by resource-rich organizations and are frequently kept from the public. As a step towards democratizing this powerful technology, we present BLOOM, a 176B-parameter open-access language model designed and built thanks to a collaboration of hundreds of researchers. BLOOM is a decoder-only Transformer language model that was trained on the ROOTS corpus, a dataset comprising hundreds of sources in 46 natural and 13 programming languages (59 in total). We find that BLOOM achieves competitive performance on a wide variety of benchmarks, with stronger results after undergoing multitask prompted finetuning. To facilitate future research and applications using LLMs, we publicly release our models and code under the Responsible AI License.
Evaluating change in ranked term importance in a growing corpus is a powerful tool for understanding changes in vocabulary usage. In this paper, we analyze a corpus of free-response answers where 33,993 LGBTQ Generation Z respondents from age 13 to 24 in the United States are asked to self-describe their sexual orientation. We observe that certain labels, such as bisexual, pansexual, and lesbian, remain equally important across age groups. The importance of other labels, such as homosexual, demisexual, and omnisexual, evolve across age groups. Although Generation Z is often stereotyped as homogenous, we observe noticeably different label usage when self-describing sexual orientation within it. We urge that interested parties must routinely survey the most important sexual orientation labels to their target audience and refresh their materials (such as demographic surveys) to reflect the constantly evolving LGBTQ community and create an inclusive environment.
What are the units of text that we want to model? From bytes to multi-word expressions, text can be analyzed and generated at many granularities. Until recently, most natural language processing (NLP) models operated over words, treating those as discrete and atomic tokens, but starting with byte-pair encoding (BPE), subword-based approaches have become dominant in many areas, enabling small vocabularies while still allowing for fast inference. Is the end of the road character-level model or byte-level processing? In this survey, we connect several lines of work from the pre-neural and neural era, by showing how hybrid approaches of words and characters as well as subword-based approaches based on learned segmentation have been proposed and evaluated. We conclude that there is and likely will never be a silver bullet singular solution for all applications and that thinking seriously about tokenization remains important for many applications.