Saved in:
Bibliographic Details
Main Authors: Bellos, Filippos, Premkumar, NaveenJohn, Avrithis, Yannis, Nguyen, Nam H., Corso, Jason J.
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2602.16188
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911453964926976
author Bellos, Filippos
Premkumar, NaveenJohn
Avrithis, Yannis
Nguyen, Nam H.
Corso, Jason J.
author_facet Bellos, Filippos
Premkumar, NaveenJohn
Avrithis, Yannis
Nguyen, Nam H.
Corso, Jason J.
contents LLM-for-time series (TS) methods typically treat time shallowly, injecting positional or prompt-based cues once at the input of a largely frozen decoder, which limits temporal reasoning as this information degrades through the layers. We introduce Temporal-Prior Conditioning (TPC), which elevates time to a first-class modality that conditions the model at multiple depths. TPC attaches a small set of learnable time series tokens to the patch stream; at selected layers these tokens cross-attend to temporal embeddings derived from compact, human-readable temporal descriptors encoded by the same frozen LLM, then feed temporal context back via self-attention. This disentangles time series signal and temporal information while maintaining a low parameter budget. We show that by training only the cross-attention modules and explicitly disentangling time series signal and temporal information, TPC consistently outperforms both full fine-tuning and shallow conditioning strategies, achieving state-of-the-art performance in long-term forecasting across diverse datasets. Code available at: https://github.com/fil-mp/Deep_tpc
format Preprint
id arxiv_https___arxiv_org_abs_2602_16188
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Deep TPC: Temporal-Prior Conditioning for Time Series Forecasting
Bellos, Filippos
Premkumar, NaveenJohn
Avrithis, Yannis
Nguyen, Nam H.
Corso, Jason J.
Machine Learning
LLM-for-time series (TS) methods typically treat time shallowly, injecting positional or prompt-based cues once at the input of a largely frozen decoder, which limits temporal reasoning as this information degrades through the layers. We introduce Temporal-Prior Conditioning (TPC), which elevates time to a first-class modality that conditions the model at multiple depths. TPC attaches a small set of learnable time series tokens to the patch stream; at selected layers these tokens cross-attend to temporal embeddings derived from compact, human-readable temporal descriptors encoded by the same frozen LLM, then feed temporal context back via self-attention. This disentangles time series signal and temporal information while maintaining a low parameter budget. We show that by training only the cross-attention modules and explicitly disentangling time series signal and temporal information, TPC consistently outperforms both full fine-tuning and shallow conditioning strategies, achieving state-of-the-art performance in long-term forecasting across diverse datasets. Code available at: https://github.com/fil-mp/Deep_tpc
title Deep TPC: Temporal-Prior Conditioning for Time Series Forecasting
topic Machine Learning
url https://arxiv.org/abs/2602.16188