Publications

* Equal contribution

Key figure from Towards Robust Model Organisms for Eliciting Latent Knowledge, redrawn

Preprint

Towards Robust Model Organisms for Eliciting Latent Knowledge

Lennart Finke*, Hardik Bhatnagar*, Euodia Dodd*, Francis Rhys Ward, Yves-Alexandre de Montjoye, Matthias Bethge, Mary Phuong

Key figure from PostTrainBench: Can LLM Agents Automate LLM Post-Training?, redrawn

ICML 2026

PostTrainBench: Can LLM Agents Automate LLM Post-Training?

Ben Rank*, Hardik Bhatnagar*, Matthias Bethge, Maksym Andriushchenko

Paper Code Website

Best Paper · ICLR 2026 RSI Workshop

Key figure from A Sober Look at Progress in Language Model Reasoning: Pitfalls and Paths to Reproducibility, redrawn

COLM 2025

A Sober Look at Progress in Language Model Reasoning: Pitfalls and Paths to Reproducibility

Andreas Hochlehnert*, Hardik Bhatnagar*, Vishaal Udandarao, Samuel Albanie, Ameya Prabhu, Matthias Bethge

Paper Code Website

Key figure from A is for Absorption: Studying Feature Splitting and Absorption in Sparse Autoencoders, redrawn

NeurIPS 2025Oral

A is for Absorption: Studying Feature Splitting and Absorption in Sparse Autoencoders

David Chanin*, James Wilken-Smith*, Tomas Dulka*, Hardik Bhatnagar*, Joseph Bloom

Paper Code Blog Website