# Random Initialization Can't Catch Up: The Advantage of Language Model Transfer for Time Series Forecasting

arXiv: https://arxiv.org/abs/2506.21570 · HTML: https://mlsanity.com/paper/2506.21570/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 10 of its 13 references that are in the map.
- First posted on arXiv 12 Jun 2025.

## Builds on (most cited first, 10 of 10)

- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [Finetuned Language Models Are Zero-Shot Learners](https://mlsanity.com/paper/2109.01652/) (2021)
- [Unified Training of Universal Time Series Forecasting Transformers](https://mlsanity.com/paper/2402.02592/) (2024)
- [On the Stability of Fine-tuning BERT: Misconceptions, Explanations, and Strong Baselines](https://mlsanity.com/paper/2006.04884/) (2020)
- [Pretrained Transformers as Universal Computation Engines](https://mlsanity.com/paper/2103.05247/) (2021)
- [Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://mlsanity.com/paper/2109.10686/) (2021)
- [Context is Key: A Benchmark for Forecasting with Essential Textual Information](https://mlsanity.com/paper/2410.18959/) (2024)
- [LLM Processes: Numerical Predictive Distributions Conditioned on Natural Language](https://mlsanity.com/paper/2405.12856/) (2024)
- [PQMass: Probabilistic Assessment of the Quality of Generative Models using Probability Mass Estimation](https://mlsanity.com/paper/2402.04355/) (2024)

## Built on by (most cited first, 2 of 2)

- [LLM Pretraining Shapes a Generalizable Manifold: Insights into Cross-Modal Transfer to Time Series](https://mlsanity.com/paper/2605.20449/) (2026)
- [Procedural Core: A Compact Recurrent Initialization for Vision Transformers](https://mlsanity.com/paper/2609.37631/) (2026)

## Abstract (arXiv)

Recent works have demonstrated the effectiveness of adapting pre-trained language models (LMs) for forecasting time series in the low-data regime. We build upon these findings by analyzing the effective transfer from language models to time series forecasting under various design choices including upstream post-training, time series tokenizer and language backbone size. In the low-data regime, these design choices have a significant impact on the validation loss, with clear-cut choices that outperform others. Contrary to Hernandez et al. (2021), we observe that the validation loss of the LMs continues to smoothly decrease long after the validation loss of the randomly initialized models has converged, leading to a non-vanishing transfer gap that holds across design choices. These findings not only help shed light on the effective use of compute-efficient training for time series, but also open the way for the study of modality-agnostic properties of data distributions leveraged by these models.