Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Grantner, Tobias, Sallinger, Emanuel, Flechl, Martin
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2604.18199
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911608021712896
author Grantner, Tobias
Sallinger, Emanuel
Flechl, Martin
author_facet Grantner, Tobias
Sallinger, Emanuel
Flechl, Martin
contents Transformer-based embedding models suffer from quadratic computational and linear memory complexity, limiting their utility for long sequences. We propose recurrent architectures as an efficient alternative, introducing a vertically chunked inference strategy that enables fast embedding generation with memory usage that becomes constant in the input length once it exceeds the vertical chunk size. By fine-tuning Mamba2 models, we demonstrate their viability as general-purpose text embedders, achieving competitive performance across a range of benchmarks while maintaining a substantially smaller memory footprint compared to transformer-based counterparts. We empirically validate the applicability of our inference strategy to Mamba2, RWKV, and xLSTM models, confirming consistent runtime-memory trade-offs across architectures and establishing recurrent models as a compelling alternative to transformers for efficient embedding generation.
format Preprint
id arxiv_https___arxiv_org_abs_2604_18199
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Linear-Time and Constant-Memory Text Embeddings Based on Recurrent Language Models
Grantner, Tobias
Sallinger, Emanuel
Flechl, Martin
Computation and Language
Transformer-based embedding models suffer from quadratic computational and linear memory complexity, limiting their utility for long sequences. We propose recurrent architectures as an efficient alternative, introducing a vertically chunked inference strategy that enables fast embedding generation with memory usage that becomes constant in the input length once it exceeds the vertical chunk size. By fine-tuning Mamba2 models, we demonstrate their viability as general-purpose text embedders, achieving competitive performance across a range of benchmarks while maintaining a substantially smaller memory footprint compared to transformer-based counterparts. We empirically validate the applicability of our inference strategy to Mamba2, RWKV, and xLSTM models, confirming consistent runtime-memory trade-offs across architectures and establishing recurrent models as a compelling alternative to transformers for efficient embedding generation.
title Linear-Time and Constant-Memory Text Embeddings Based on Recurrent Language Models
topic Computation and Language
url https://arxiv.org/abs/2604.18199