Baselines and Metrics Update

[LLMs with Limited Resources for Slavic Languages 2025]

Dear participants,

We would like to inform you that we have provided for you the code to run baselines and evaluation scores. 

Please, check out our main repo:

https://github.com/TUM-NLP/llms-limited-resources2025

for Evaluation Methods and Baseline Results sections. 

The code is provided in this repository:

https://github.com/TUM-NLP/wmt25-lrsl-evaluation

where you can find:

1) the code for evaluation;

2) the code to run baselines. Our baseline model is based on Qwen2.5 instance:

https://huggingface.co/unsloth/Qwen2.5-3B-Instruct-unsloth-bnb-4bit

We would encourage you to try out baselines and try already to improve them! We also would be glad to ask any clarification questions, so do not hesitate to reach to us!

We wish you all good luck with your experiments during the dev phase and preparations for the final test set. It will be aligned with the WMT2025 main track and will start in the end of June.

All the best,

Organizers