Time Series Reasoning via Process-Verifiable Thinking Data Synthesis and Scheduling for Tailored LLM Reasoning

TL;DR

VeriTime is a framework that enhances LLMs for time series reasoning by synthesizing data, scheduling training samples effectively, and applying RL finetuning, leading to significant performance improvements.

Contribution

It introduces a novel data synthesis pipeline, a hierarchical data scheduling mechanism, and a two-stage RL finetuning process specifically for time series reasoning in LLMs.

Findings

Small models achieve reasoning performance comparable to larger models.

The framework significantly improves LLM performance on diverse time series tasks.

VeriTime's data synthesis and scheduling strategies are effective for training time series reasoning models.

Abstract

Time series is a pervasive data type across various application domains, rendering the reasonable solving of diverse time series tasks a long-standing goal. Recent advances in large language models (LLMs), especially their reasoning abilities unlocked through reinforcement learning (RL), have opened new opportunities for tackling tasks with long Chain-of-Thought (CoT) reasoning. However, leveraging LLM reasoning for time series remains in its infancy, hindered by the absence of carefully curated time series CoT data for training, limited data efficiency caused by underexplored data scheduling, and the lack of RL algorithms tailored for exploiting such time series CoT data. In this paper, we introduce VeriTime, a framework that tailors LLMs for time series reasoning through data synthesis, data scheduling, and RL training. First, we propose a data synthesis pipeline that constructs a TS-text multimodal dataset with process-verifiable annotations. Second, we design a data scheduling mechanism that arranges training samples according to a principled hierarchy of difficulty and task taxonomy. Third, we develop a two-stage reinforcement finetuning featuring fine-grained, multi-objective rewards that leverage verifiable process-level CoT data. Extensive experiments show that VeriTime substantially boosts LLM performance across diverse time series reasoning tasks. Notably, it enables compact 3B, 4B models to achieve reasoning capabilities on par with or exceeding those of larger proprietary LLMs.