TimeHF: Billion-Scale Time Series Models Guided by Human Feedback

TL;DR

TimeHF, a novel pipeline for creating LTMs with 6 billion parameters, incorporating human feedback, is introduced, using patch convolutional embedding to capture long time series information and design a human feedback mechanism called time-series policy optimization.

Abstract

Time series neural networks perform exceptionally well in real-world applications but encounter challenges such as limited scalability, poor generalization, and suboptimal zero-shot performance. Inspired by large language models, there is interest in developing large time series models (LTM) to address these issues. However, current methods struggle with training complexity, adapting human feedback, and achieving high predictive accuracy. We introduce TimeHF, a novel pipeline for creating LTMs with 6 billion parameters, incorporating human feedback. We use patch convolutional embedding to capture long time series information and design a human feedback mechanism called time-series policy optimization. Deployed in JD.com's supply chain, TimeHF handles automated replenishment for over 20,000 products, improving prediction accuracy by 33.21% over existing methods. This work advances LTM technology and shows significant industrial benefits.

Figures (6)

• Figure 1: Left: Traditional time series foundation models, while performing well in common scenarios, may be overly sensitive to noises in training datasets, resulting in hallucinations in complex scenarios. Right: With RLHF, feedback contrast pairs—constructed by interpretable small models created by human experts—guide the model to gradually shift toward more accurate predictions.
• Figure 2: Model architecture of our method.
• Figure 3: Cross-patch convolutional projection layers.
• Figure 4: Impact of Hyperparameters on Accuracy Improvement
• Figure 5: Prediction Comparison Between Our Method and JD_Online Across Different Products