5 papers
Interpreting and Steering a Text-to-Speech Language Model with Sparse Autoencoders
Nikita Koriagin, Georgii Aparin, Nikita Balagansky +1
Language models increasingly serve as the backbone of text-to-speech (TTS) systems, yet we understand little about the representations they build when text and generated speech tok…
Optimality of FSQ Tokens for Continuous Diffusion for Categorical Data with Application to Text-to-Speech
Vadim Popov, Wenju Gu, Tasnima Sadekova +2
Continuous diffusion for categorical data is a framework belonging to the diffusion family and aiming at generating discrete data. The scientific interest to such models has been c…
A Geometric Account of Activation Steering through Angle-Norm Decomposition
Georgii Aparin, Tatiana Gaintseva
Linear activation steering has gained popularity as a simple and empirically effective way to control language model behavior. More recently, spherical steering paradigms have been…
Whisper Hallucination Detection and Mitigation via Hidden Representation Steering and Sparse AutoEncoders
Georgii Aparin, Vadim Popov, Tasnima Sadekova +1
Whisper, a widely adopted ASR model, is known to suffer from hallucinations - coherent transcriptions generated for non-speech audio entirely disconnected from the input. We invest…
AudioSAE: Towards Understanding of Audio-Processing Models with Sparse AutoEncoders
Georgii Aparin, Tasnima Sadekova, Alexey Rukhovich +5
Sparse Autoencoders (SAEs) are powerful tools for interpreting neural representations, yet their use in audio remains underexplored. We train SAEs across all encoder layers of Whis…