AI.info

Simon Ostermann

Explore Simon Ostermann on AI.info.

• Separating Syntax from Language: A Mechanistic Account of Translation in Multilingual LLMs
• When Tokenization is Secretly Output Supervision
• CLaS-Bench: A Cross-Lingual Alignment and Steering Benchmark
• Can Large Language Models Still Explain Themselves? Investigating the Impact of Quantization on Self-Explanations