Publications | Nikita Balagansky

2025

EMNLP

Steering LLM Reasoning Through Bias-Only Adaptation

Viacheslav Sinii, Alexey Gorbatovski, Artem Cherepanov, Boris Shaposhnikov, Nikita Balagansky, Daniil Gavrilov

EMNLP

Train One Sparse Autoencoder Across Multiple Sparsity Budgets to Preserve Interpretability and Accuracy

Nikita Balagansky, Yaroslav Aksenov, Daniil Laptev, Vadim Kurochkin, Gleb Gerasimov, Nikita Koriagin, Daniil Gavrilov

ICLR

Mechanistic Permutability: Match Features Across Layers

Nikita Balagansky, Ian Maksimov, Daniil Gavrilov

ICLR

Learn Your Reference Model for Real Good Alignment

Alexey Gorbatovski, Boris Shaposhnikov, Alexey Malakhov, Nikita Surnachev, Yaroslav Aksenov, Ian Maksimov, Nikita Balagansky, Daniil Gavrilov

ICML

Analyze Feature Flow to Enhance Interpretation and Steering in Language Models

Daniil Laptev, Nikita Balagansky, Yaroslav Aksenov, Daniil Gavrilov

COLM

Teach Old SAEs New Domain Tricks with Boosting

Nikita Koriagin, Yaroslav Aksenov, Daniil Laptev, Gleb Gerasimov, Nikita Balagansky, Daniil Gavrilov

ACL

Linear Transformers with Learnable Kernel Functions are Better In-Context Models

Yaroslav Aksenov, Nikita Balagansky, Sofia Maria Lo Cicero Vaina, Boris Shaposhnikov, Alexey Gorbatovski, Daniil Gavrilov

NeurIPS

PALBERT: Teaching ALBERT to Ponder

Nikita Balagansky, Daniil Gavrilov