Paul Michel

Research scientist working on large language models, scaling, alignment, and language; public work includes machine translation robustness, data selection, CodeGemma, and reward overoptimization in LLMs.