A Foundation Model for Instruction-Conditioned In-Context Time Series Tasks

TL;DR

iAmTime is a novel instruction-conditioned time-series foundation model that improves zero-shot task adaptation across diverse domains by leveraging structured prompts and specialized transformer architectures.

Contribution

The paper introduces iAmTime, a new time-series foundation model trained with instruction-conditioned meta-learning, enabling explicit task inference from demonstrations.

Findings

iAmTime outperforms baseline models on probabilistic and point forecasting benchmarks.

The model achieves competitive results on classification and other non-forecasting tasks.

It effectively infers task structure from demonstrations across diverse domains.

Abstract

In-context learning (ICL) enables task adaptation at inference time by conditioning on demonstrations rather than updating model parameters. Although recent time-series foundation models incorporate contextual conditioning, retrieval, or example-based prompting, they typically rely on implicit positional structure or task-specific objectives rather than explicit instruction-conditioned input-output demonstrations. We introduce iAmTime, a time-series foundation model trained with instruction-conditioned amortized meta-learning to infer tasks directly from example demonstrations. iAmTime represents each episode as a structured prompt over historical context and future-known variables using specialized semantic tokens that attend to designated time-series regions, exchange information across demonstrations, and inject task information into the query representation. The model combines a Hierarchical Multi-Scope Transformer Encoder, which captures temporal and covariate dynamics while inferring latent task structure from demonstrated input-output mappings, with a Task-Conditioned Patch Decoder, which adapts decoding through expert-based routing. We train iAmTime on large-scale real and synthetic corpora using supervised and self-supervised instruction-conditioned tasks, including forecasting, imputation, reconstruction, classification, anomaly detection, and source de-mixing. Across diverse domains, frequencies, and horizons, iAmTime improves zero-shot adaptation over strong time-series foundation baselines on probabilistic and point forecasting benchmarks, while achieving competitive performance on non-forecasting tasks such as classification.