TimeBraid unifies language and time series for better forecasting
TimeBraid: Unifying Time Series and Language for Understanding and Forecasting
Computation and LanguageArtificial IntelligenceComputational Engineering, Finance, and Science
Summary
Forecasting future events based on data that changes over time is hard, especially when descriptions in language need to connect with time-based numbers. The authors introduce TimeBraid, a method that combines models trained on language and on time-series data into one system that understands and generates both text and time records. They designed it carefully to balance learning from the two types of data together and trained it on millions of paired examples. TimeBraid performs well on tasks like interpreting time data, following instructions, reasoning, and predicting future values using context.
What this means in practice
- •For financial analysts: Enhance forecasting accuracy by combining textual reports and time-series data in one model for market trend prediction.
- •For energy grid operators: Use integrated time-series and language models to improve interpretation and prediction of energy usage patterns with textual event context.
Authors
Xinyue Wang, Jiacheng Pang, Kun Zhou, Kexin Zhang, Defu Cao, Fan Feng, Faisal, Songyao Jin, Yan Liu, Biwei Huang
Abstract
We present TimeBraid, a series of unified time-series and language models that align pretrained language models and pretrained time-series foundation models through interleaved global residual attention layers. Each model inherits knowledge, instruction following, and reasoning from one side, continuous-signal perception and zero-shot forecasting from the other, and fuses the two in a shared representation space where both modalities are understood and generated. We study the design choices that make such unified modeling work: where to align the two representation spaces, how to ground language in temporal structure, how to balance understanding with generation, and how to keep joint optimization stable. The resulting recipe combines a unified prompting scheme for diverse time-series and text tasks, stabilized joint training, and supervision from 2.2M curated series--text pairs and 4.9M instruction-tuning samples. Across benchmarks spanning time-series perception, understanding, reasoning, and both context-aided and unimodal forecasting, TimeBraid remains competitive with far larger general-purpose models and task-specific counterparts.