DataComp-LM: In search of the next generation of training sets for language models

Jeffrey Li, Alex Fang, Georgios Smyrnis, Maor Ivgi, Matt Jordan, Samir Yitzhak Gadre, Hritik Bansal, Etash Kumar Guha, Sedrick Keh, Kushal Arora, Saurabh Garg, Rui Xin, Niklas Muennighoff, Reinhard Heckel, Jean Mercat, Mayee F Chen, Suchin Gururangan, Mitchell Wortsman, Alon Albalak, Yonatan Bitton et al. (39 additional authors not shown)

Keywords: language models, data comp, datacomp, dataset design, data-centric AI, data-centric benchmarks, benchmarks, data curation

TL;DR: We introduce a benchmark to build better language models by improving dataset design.

Abstract:

We introduce DataComp for Language Models, a testbed for controlled dataset experiments with the goal of improving language models.
As part of DCLM, we provide a standardized corpus of 240T tokens extracted from Common Crawl, effective pretraining recipes based on the OpenLM framework, and a broad suite of 53 downstream evaluations.
Participants in the DCLM benchmark can experiment with data curation strategies such as deduplication, filtering, and data mixing at
model scales ranging from 412M to 7B parameters.
As a baseline for DCLM, we conduct extensive experiments and find that model-based filtering is key to assembling a high-quality training set.
The resulting dataset, DCLM-Baseline, enables training a 7B parameter language model from scratch to 63% 5-shot accuracy on MMLU with 2T training tokens.
Compared to MAP-Neo, the previous state-of-the-art in open-data language models, DCLM-Baseline represents a 6 percentage point improvement on MMLU while being trained with half the compute.
Our results highlight the importance of dataset design for training language models and offer a starting point for further research on data curation. We release the \dclm benchmark, framework, models, and datasets at https://www.datacomp.ai/dclm/

Submission Number: 1114