Skip To Main Content
Yaroslav Aksenov on Muck Rack

Yaroslav Aksenov

As seen in: arXiv
Is this you? As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work. Claim your profile

Get in touch with Yaroslav

Contact Yaroslav, search articles and posts on X, monitor coverage, and track replies from one place.

Learn more about Muck Rack

Actions

Share this page

Is this you?

As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.
Claim your profile

Articles

[2507.12990] Teach Old SAEs New Domain Tricks with Boosting

|
By Yaroslav Aksenov, Daniil Laptev, Gleb Gerasimov, Nikita Balagansky
| arXiv
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Open in Who Shared View Duplicates (397) Issue with article?

[2505.22255] Train Sparse Autoencoders Efficiently by Utilizing Features Correlation

|
By Yaroslav Aksenov, Daniil Laptev, Daniil Gavrilov, Nikita Balagansky
| arXiv
arXiv:2505.22255 (cs) View PDF HTML (experimental) Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) Cite as: arXiv:2505.22255 [cs.LG] (or arXiv:2505.22255v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2505.22255 Submission history From: Yaroslav Aksenov [ view email] [v1] Wed, 28 May 2025 11:41:11 UTC (2,680 KB) Bibliographic Tools Bibliographic Explorer Toggle Bibliographic Explorer () Connected Papers Toggle Connected Papers (What is Connected Papers?)...
Open in Who Shared View Duplicates (84) Issue with article?

Paper page - Analyze Feature Flow to Enhance Interpretation and Steering in Language Models

|
By Nikita Balagansky, Yaroslav Aksenov, Daniil Gavrilov, Daniil Laptev
| huggingface.co
Papers Published on Feb 5 · Submitted by kefirski on Feb 7 #1 Paper of the day Abstract We introduce a new approach to systematically map features discovered by sparse autoencoder across consecutive layers of large language models, extending earlier work that examined inter-layer feature links. By using a data-free cosine similarity technique, we trace how specific features persist, transform, or first appear at each stage.
Open in Who Shared Issue with article?
See all 5 articles