AK
Anton Korznikov
cs.AIcs.CLcs.LG
On Valency
published · living versionsW_8pemzcn3·v1 · currentpublished
Context or Knowledge: How LLMs Attribute Their Own Answers
with Kirill Afendulev, Alexey Dontsov, Alla Chepurova, Elena Tutubalina
1 version
W_5khfgb7z·v1published
Preprints & journals
6 papers in the corpus · 2025–2026From Stories to Features: Scaling Concept Vectors for AI Alignment & Interpretability2610.24556v1 · Dmitry Scherbakov, Oleg Chernikov, Victoria Khamash et al.2026 · 0 citationsarXivon Valency
Sparse Transformer: a Controlled Testbed for Mechanistic Interpretability2610.19907v1 · Anton Korznikov, Alexey Dontsov, Elena Tutubalina2026 · 0 citationsarXiv
Your Transformer Can Hold Two Thoughts at Once: Evidence of Linear Superposition in LLMs2609.29845v1 · Pavel Tikhonov, Anton Korznikov, Matvey Mikhalchuk et al.2026 · 0 citationsarXiv
What Attention Recalls and Recurrence Controls in Hybrid Language Models2609.04434v1 · Kirill Afendulev, Alexey Dontsov, Elena Tutubalina et al.2026 · 0 citationsarXiv
OrtSAE: Orthogonal Sparse Autoencoders Uncover Atomic Features2509.22033v1 · Anton Korznikov, Andrey Galichin, Alexey Dontsov et al.2025 · 0 citationsarXiv
The Rogue Scalpel: Activation Steering Compromises LLM Safety2509.22067v2 · Anton Korznikov, Andrey Galichin, Alexey Dontsov et al.2025 · 0 citationsarXiv
Career total: 11 works. 6 are in this corpus.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.