Master Thesis: Calibrating LLM-as-a-Judge for Swedish Fluency vs. Adequacy
AI Sweden
Summary
Automated evaluation is vital for tracking LLM post-training progress. "LLM-as-a-judge" paradigms are highly scalable but often exhibit severe biases, particularly in non-English languages where they may conflate grammatical fluency with factual adequacy, heavily penalizing minor linguistic errors.
Job description
Introduction
Automated evaluation is vital for tracking LLM post-training progress. "LLM-as-a-judge" paradigms are highly scalable but often exhibit severe biases, particularly in non-English languages where they may conflate grammatical fluency with factual adequacy, heavily penalizing minor linguistic errors.
Project Background and Problem Statement
In WP5, OpenEuroLLM is relying on the JudgeArena framework. For Swedish, we face the challenge that English-centric judge models misjudge local knowledge or nuanced fluency. This thesis explores whether providing an LLM judge with few-shot, human-calibrated Swedish examples can effectively decouple fluency from adequacy, aligning the automated Elo scores with human expert ratings.
Outline
The goal is to refine and validate JudgeArena for Swedish post-training evaluation.
- Literature study: Review automated evaluation paradigms, LLM-as-a-judge biases, and metrics distinguishing fluency from adequacy.
- Implementation: Annotate a high-quality Swedish dataset with separate scores for fluency and adequacy. Use these as calibration prompts for the Judge model in JudgeArena.
- Evaluation: Correlate the calibrated JudgeArena outputs against human judgments on a holdout set of Prelude 9B's generated responses, quantifying the improvement in Pearson/Spearman correlation.
Who we're looking for
We are seeking curious, self-driven MSc students eager to work at the frontier of open-weight European AI research (LLMs). You thrive on empirical discovery, design rigorous experiments, and let data challenge your assumptions.
- Ongoing Master's studies in Computer Science, Data Science, Machine Learning, Engineering Physics, or a related quantitative field.
- Proficiency in Python and hands-on experience with modern deep learning frameworks (PyTorch, Hugging Face ecosystem).
- Familiarity with LLM post-training alignment (e.g., SFT, DPO, RLHF/RLVR) or context-extension, alongside comfort running distributed GPU training in Linux/HPC environments.
Requirements
- Ongoing Master's studies in Computer Science, Data Science, Machine Learning, Engineering Physics, or a related quantitative field.
- Proficiency in Python and hands-on experience with modern deep learning frameworks (PyTorch, Hugging Face ecosystem).
- Familiarity with LLM post-training alignment (e.g., SFT, DPO, RLHF/RLVR) or context-extension, alongside comfort running distributed GPU training in Linux/HPC environments.
Company overview
AI Sweden is Sweden's national center for applied artificial intelligence, established in 2019 with a mission to accelerate the adoption and implementation of AI technologies for societal benefit, economic competitiveness, and national advancement. As a collaborative research institute, it brings together over 170 partners from diverse sectors including government agencies, universities, research institutions, startups, and global corporations. The organization operates as a catalyst for AI innovation, focusing on practical applications across healthcare, energy systems, municipalities, and various industries. With a core team of over 80 professionals comprising AI experts, data scientists, research engineers, linguists, mathematicians, policy experts, and change leaders, AI Sweden facilitates knowledge sharing, talent development, and infrastructure building. The institute operates on a not-for-profit basis, funded by Sweden's innovation agency Vinnova, the Swedish Agency for Economic and Regional Growth, European Regional Development Fund, Västra Götalandsregionen, and partner contributions. Through its AI Labs, talent programs, workshops, and collaborative projects, AI Sweden addresses critical societal challenges while bridging the gap between cutting-edge research and real-world implementation, positioning Sweden as a leader in responsible and impactful AI development.