Author

Paweł Mąka

1 paper indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

#artificial intelligence Preprint Sep 2026

Scaling Attention Head Analysis via Gradient-Based Attribution in Context-Aware Machine Translation

A gradient-based head attribution strategy where the Token-level Max-Margin loss is backpropagated to the attention maps enables a large-scale causal analysis of attention heads, making it suitable for LLMs.

Paweł Mąka, Yusuf Can Semerci, Jan Scholtes et al. · 0 citations