I am a doctoral student advised by Prof. Fanny Yang, working on efficient evaluation methods for Large Language Models (LLMs) and Reinforcement Learning with Verifiable Rewards (RLVR) for LLM post‑training. My research focuses on developing principled approaches to assess and improve large models using measurable and trustworthy signals.
I received my MSc in Statistics from ETH Zürich and a BSc in Mathematics, Statistics, and Business from the London School of Economics. With a strong mathematical and statistical background, I enjoy tackling real‑world machine learning problems guided by theory and rigorous analysis.
Outside of research, I am a regular gym‑goer and have a strong interest in psychology.
Cutting LLM Evaluation Costs with SySRs: A Bandit Algorithm That Provably Exploits Model Similarity
Zifan Lyu*,
Chahine Nejma*,
Tobias Wegel,
Fanny Yang,
and Florian E. Dorner
International Conference on Machine Learning (ICML),
2026
Large Language Models are typically benchmarked by evaluating every model on every test query. For practitioners seeking the best model to deploy, this is often wasteful: if a model clearly performs worse than others, there is no need to precisely estimate its performance. Best-arm identification algorithms can be naturally applied to drastically reduce costs by adaptively allocating evaluation budget. Further, language models often respond similarly to the same prompt-a property previous work has tried to leverage with mixed success. We propose Synchronized Successive Rejects (SySRs), augmenting the classical Successive Rejects algorithm with paired comparisons. Unlike prior attempts to leverage model similarity in best-model identification, our approach is hyperparameter-free and enjoys performance guarantees that improve with the degree of similarity between evaluated models. Empirically, our method outperforms all baselines in terms of average error rate across 15 standard benchmarks, and in terms of worst-case budget for reliably identifying the best model.
</div>
</li></ol>
</div> –>
Papers
-
Cutting LLM Evaluation Costs with SySRs: A Bandit Algorithm That Provably Exploits Model Similarity
Zifan Lyu*,
Chahine Nejma*,
Tobias Wegel,
Fanny Yang,
and Florian E. Dorner
International Conference on Machine Learning (ICML),
2026
Large Language Models are typically benchmarked by evaluating every model on every test query. For practitioners seeking the best model to deploy, this is often wasteful: if a model clearly performs worse than others, there is no need to precisely estimate its performance. Best-arm identification algorithms can be naturally applied to drastically reduce costs by adaptively allocating evaluation budget. Further, language models often respond similarly to the same prompt-a property previous work has tried to leverage with mixed success. We propose Synchronized Successive Rejects (SySRs), augmenting the classical Successive Rejects algorithm with paired comparisons. Unlike prior attempts to leverage model similarity in best-model identification, our approach is hyperparameter-free and enjoys performance guarantees that improve with the degree of similarity between evaluated models. Empirically, our method outperforms all baselines in terms of average error rate across 15 standard benchmarks, and in terms of worst-case budget for reliably identifying the best model.
Preprints
zifan.lyu@inf.ethz.ch
CAB G17 ETH Zürich
