Evaluation Library for the Hungarian Language Understanding Benchmark (HuLU)
This paper introduces Hungary’s first fine-tuning and evaluation Python library for the Hungarian Language Understanding Benchmark Kit (HuLU), designed to streamline prototyping and support effective language modeling in Hungarian. The library lets researchers quickly validate their results on the o...
Elmentve itt :
| Szerzők: | |
|---|---|
| Testületi szerző: | |
| Dokumentumtípus: | Könyv része |
| Megjelent: |
Szegedi Tudományegyetem TTIK, Informatikai Intézet
Szeged
2025
|
| Sorozat: | Magyar Számítógépes Nyelvészeti Konferencia
21 |
| Kulcsszavak: | Nyelvi modell, Nyelvészet - számítógép alkalmazása |
| Tárgyszavak: | |
| Online Access: | http://acta.bibl.u-szeged.hu/88786 |
| Tartalmi kivonat: | This paper introduces Hungary’s first fine-tuning and evaluation Python library for the Hungarian Language Understanding Benchmark Kit (HuLU), designed to streamline prototyping and support effective language modeling in Hungarian. The library lets researchers quickly validate their results on the official HuLU leaderboard, handling the full pipeline from training to submission. It supports a range of architectures, including encoders, decoders, and custom PyTorch modules, and it’s ready for both smaller models with standard fine-tuning and larger models using low-rank adaptation and reduced-precision fine-tuning. The library is highly configurable and supports huggingface integration, giving researchers the flexibility of modern training frameworks. Two interfaces are included: a simple command-line interface and a programmatic API for more in-depth customization. |
|---|---|
| Terjedelem/Fizikai jellemzők: | 257-268 |
| ISBN: | 978-963-688-034-7 |