Performance Prediction for Large Systems via Text-to-Text Regression
Abstract
A text-to-text regression model achieves high accuracy in predicting resource efficiency for Google's Borg system, surpassing tabular methods, and demonstrates adaptability and uncertainty quantification.
In many industries, predicting metric outcomes of large systems is a fundamental problem, driven largely by traditional tabular regression. However, such methods struggle on complex systems data in the wild such as configuration files or system logs, where feature engineering is often infeasible. We propose text-to-text regression as a general, scalable alternative. For predicting resource efficiency on Borg, Google's massive compute cluster scheduling system, a 60M parameter encoder-decoder, trained from random initialization, achieves up to a near perfect 0.99 (0.9 average) rank correlation across the entire fleet, and 100x lower MSE than tabular approaches. The model also easily adapts to new tasks in only 500 few-shot examples and captures the densities of complex outcome distributions. Ablation studies highlight the importance of using encoders, increasing sequence length, and the model's inherent uncertainty quantification. These findings pave the way for universal simulators of real-world outcomes.
Community
Hi everyone! Sharing our recent work. A lot of our motivation comes from trying to reward model real world feedback, necessary for super-intelligence ("Era of Experience" essay by Richard Sutton and David Silver is a great reference!)
Seeing text-to-text regression work for Google’s massive compute cluster (billion $$ problem!) blew our minds and convinced us that we can reward model literally any world feedback.
Paper: http://arxiv.org/abs/2506.21718
Code: http://github.com/google-deepmind/regress-lm
It's very simple: just train a simple encoder-decoder from scratch to read any complex "x" as text, then generate numeric tokens as "y".
We’re also seeing strong results on classic tabular data and "exotic" inputs like graphs, system logs, and even code snippets. Feature engineering will no longer exist!
Models citing this paper 0
No model linking this paper
Datasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper
Collections including this paper 0
No Collection including this paper