A survey on data selection for language models

A Albalak, Y Elazar, SM Xie, S Longpre… - arXiv preprint arXiv …, 2024 - arxiv.org
A major factor in the recent success of large language models is the use of enormous and
ever-growing text datasets for unsupervised pre-training. However, naively training a model …

Holistic evaluation of language models

P Liang, R Bommasani, T Lee, D Tsipras… - arXiv preprint arXiv …, 2022 - arxiv.org
Language models (LMs) are becoming the foundation for almost all major language
technologies, but their capabilities, limitations, and risks are not well understood. We present …

[PDF][PDF] DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models.

B Wang, W Chen, H Pei, C Xie, M Kang, C Zhang, C Xu… - NeurIPS, 2023 - blogs.qub.ac.uk
Abstract Generative Pre-trained Transformer (GPT) models have exhibited exciting progress
in their capabilities, capturing the interest of practitioners and the public alike. Yet, while the …

Doremi: Optimizing data mixtures speeds up language model pretraining

SM Xie, H Pham, X Dong, N Du, H Liu… - Advances in …, 2024 - proceedings.neurips.cc
The mixture proportions of pretraining data domains (eg, Wikipedia, books, web text) greatly
affect language model (LM) performance. In this paper, we propose Domain Reweighting …

On the opportunities and risks of foundation models

R Bommasani, DA Hudson, E Adeli, R Altman… - arXiv preprint arXiv …, 2021 - arxiv.org
AI is undergoing a paradigm shift with the rise of models (eg, BERT, DALL-E, GPT-3) that are
trained on broad data at scale and are adaptable to a wide range of downstream tasks. We …

Towards artificial general intelligence (agi) in the internet of things (iot): Opportunities and challenges

F Dou, J Ye, G Yuan, Q Lu, W Niu, H Sun… - arXiv preprint arXiv …, 2023 - arxiv.org
Artificial General Intelligence (AGI), possessing the capacity to comprehend, learn, and
execute tasks with human cognitive abilities, engenders significant anticipation and intrigue …

Just train twice: Improving group robustness without training group information

EZ Liu, B Haghgoo, AS Chen… - International …, 2021 - proceedings.mlr.press
Standard training via empirical risk minimization (ERM) can produce models that achieve
low error on average but high error on minority groups, especially in the presence of …

Measuring robustness to natural distribution shifts in image classification

R Taori, A Dave, V Shankar, N Carlini… - Advances in …, 2020 - proceedings.neurips.cc
We study how robust current ImageNet models are to distribution shifts arising from natural
variations in datasets. Most research on robustness focuses on synthetic image …

Distributionally robust neural networks for group shifts: On the importance of regularization for worst-case generalization

S Sagawa, PW Koh, TB Hashimoto, P Liang - arXiv preprint arXiv …, 2019 - arxiv.org
Overparameterized neural networks can be highly accurate on average on an iid test set yet
consistently fail on atypical groups of the data (eg, by learning spurious correlations that …

Wilds: A benchmark of in-the-wild distribution shifts

PW Koh, S Sagawa, H Marklund… - International …, 2021 - proceedings.mlr.press
Distribution shifts—where the training distribution differs from the test distribution—can
substantially degrade the accuracy of machine learning (ML) systems deployed in the wild …