GRDD+: An Extended Greek Dialectal Dataset with Cross-Architecture Fine-tuning Evaluation

TL;DR

This paper introduces GRDD+, a comprehensive Greek dialectal dataset with over 6 million words across 10 varieties, and evaluates the impact of dialectal data on large language models through extensive fine-tuning experiments.

Contribution

The paper provides the largest and most diverse Greek dialect dataset to date and assesses how high-quality dialectal data influences LLM performance.

Findings

Dialectal data improves LLM fine-tuning results.

Extended dataset covers 10 Greek varieties.

Fine-tuning enhances model understanding of dialectal nuances.

Abstract

We present an extended Greek Dialectal Dataset (GRDD+) 1that complements the existing GRDD dataset with more data from Cretan, Cypriot, Pontic and Northern Greek, while we add six new varieties: Greco-Corsican, Griko (Southern Italian Greek), Maniot, Heptanesian, Tsakonian, and Katharevusa Greek. The result is a dataset with total size 6,374,939 words and 10 varieties. This is the first dataset with such variation and size to date. We conduct a number of fine-tuning experiments to see the effect of good quality dialectal data on a number of LLMs. We fine-tune three model architectures (Llama-3-8B, Llama-3.1-8B, Krikri-8B) and compare the results to frontier models (Claude-3.7-Sonnet, Gemini-2.5, ChatGPT-5).