Salvato in:
Dettagli Bibliografici
Autori principali: Kocabay, Şuayp Talha, Akkuş, Talha Rüzgar
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2603.20466
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917355419860992
author Kocabay, Şuayp Talha
Akkuş, Talha Rüzgar
author_facet Kocabay, Şuayp Talha
Akkuş, Talha Rüzgar
contents Masked Diffusion Language Models (MDLMs) have emerged as a compelling non-autoregressive alternative to standard large language models; however, their application to morphologically rich languages remains limited. In this paper, we introduce $\textit{Diffutron}$, a masked diffusion language model specifically designed for Turkish. Our approach leverages a resource-efficient training pipeline, starting with LoRA-based continual pre-training of a multilingual encoder on a large-scale corpus. To enable generative capabilities, we employ a progressive instruction-tuning strategy, sequentially adapting the model on general and task-specific instruction sets. Experimental results across comprehensive benchmarks demonstrate that, despite its compact size, our model achieves competitive performance compared to existing multi-billion-parameter baselines. These findings validate the effectiveness of masked diffusion modeling combined with multi-stage tuning for non-autoregressive text generation in Turkish.
format Preprint
id arxiv_https___arxiv_org_abs_2603_20466
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Diffutron: A Masked Diffusion Language Model for Turkish Language
Kocabay, Şuayp Talha
Akkuş, Talha Rüzgar
Computation and Language
Artificial Intelligence
Masked Diffusion Language Models (MDLMs) have emerged as a compelling non-autoregressive alternative to standard large language models; however, their application to morphologically rich languages remains limited. In this paper, we introduce $\textit{Diffutron}$, a masked diffusion language model specifically designed for Turkish. Our approach leverages a resource-efficient training pipeline, starting with LoRA-based continual pre-training of a multilingual encoder on a large-scale corpus. To enable generative capabilities, we employ a progressive instruction-tuning strategy, sequentially adapting the model on general and task-specific instruction sets. Experimental results across comprehensive benchmarks demonstrate that, despite its compact size, our model achieves competitive performance compared to existing multi-billion-parameter baselines. These findings validate the effectiveness of masked diffusion modeling combined with multi-stage tuning for non-autoregressive text generation in Turkish.
title Diffutron: A Masked Diffusion Language Model for Turkish Language
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2603.20466