Hungarian-centered 12-lingual finetuned M2M100_1.2B model

For further details, see or our demo site.

• Source language: Bulgarian (bg), Czech (cs), German (de), English (en), Croatian (hr), Polish, (pl), Romanian (ro), Russian (ru), Slovak (sk), Slovene (sl), Serbian (sr), Ukrainian (uk)

• Target language: Hungarian (hu)

• Finetuned on subcorpora from OPUS

  • Segments: 3 million per language

Limitations

• max_source_length: 256
• max_target_length: 256

Citation

If you use this model, please cite the following paper:

@article{laki-yang-12lang,
    title = {Solving Hungarian natural language processing tasks with multilingual generative models},
	journal = {Annales Mathematicae et Informaticae},
	year = {2023},
	author = {Yang, Zijian Győző and Laki László János},
    volume = {57},
	pages = {92–-106},
    doi = {10.33039/ami.2022.11.001}
}