Subha Nawer Pushpita
@pushpita1729.bsky.social
Cognitive Computational Neuroscientist in Training
Neuro + ML PhD @ CMU'29 | BS in Math and in CS @ MIT' 23 & MEng' 24
Neuro + ML PhD @ CMU'29 | BS in Math and in CS @ MIT' 23 & MEng' 24
Pinned
Excited to share our work on mechanisms of naturalistic audiovisual processing in the human brain 🧠🎬!!
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
www.biorxiv.org
The statistical frameworks we introduced- DIP and PIP- go beyond audio&video, they’re general tools for quantifying how two feature-based models differ or converge in predicting a common target over time, revealing patterns of dominance and joint encoding across brains🧠, models🤖, and modalities🎬🎧.
Reposted by Subha Nawer Pushpita
🤖📈 How do LLMs use their depth?
Akshat Gupta led a fun project to find out! We leverage TunedLens (~linear decoding of tokens) to explore how LLMs' internal representations change from layer to layer.
Preprint: arxiv.org/abs/2510.18871
1/
Akshat Gupta led a fun project to find out! We leverage TunedLens (~linear decoding of tokens) to explore how LLMs' internal representations change from layer to layer.
Preprint: arxiv.org/abs/2510.18871
1/
October 23, 2025 at 4:17 PM
🤖📈 How do LLMs use their depth?
Akshat Gupta led a fun project to find out! We leverage TunedLens (~linear decoding of tokens) to explore how LLMs' internal representations change from layer to layer.
Preprint: arxiv.org/abs/2510.18871
1/
Akshat Gupta led a fun project to find out! We leverage TunedLens (~linear decoding of tokens) to explore how LLMs' internal representations change from layer to layer.
Preprint: arxiv.org/abs/2510.18871
1/
The statistical frameworks we introduced- DIP and PIP- go beyond audio&video, they’re general tools for quantifying how two feature-based models differ or converge in predicting a common target over time, revealing patterns of dominance and joint encoding across brains🧠, models🤖, and modalities🎬🎧.
Excited to share our work on mechanisms of naturalistic audiovisual processing in the human brain 🧠🎬!!
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
www.biorxiv.org
November 7, 2025 at 11:42 PM
The statistical frameworks we introduced- DIP and PIP- go beyond audio&video, they’re general tools for quantifying how two feature-based models differ or converge in predicting a common target over time, revealing patterns of dominance and joint encoding across brains🧠, models🤖, and modalities🎬🎧.
We uncover two complementary cortical mechanisms underlying how the brain processes naturalistic audiovisual information - one system flexibly switches what sensory modality it encodes; another one jointly encodes auditory + visual inputs.
⚡ Quick thread with the main findings 👇
⚡ Quick thread with the main findings 👇
Excited to share our work on mechanisms of naturalistic audiovisual processing in the human brain 🧠🎬!!
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
www.biorxiv.org
November 7, 2025 at 11:20 PM
We uncover two complementary cortical mechanisms underlying how the brain processes naturalistic audiovisual information - one system flexibly switches what sensory modality it encodes; another one jointly encodes auditory + visual inputs.
⚡ Quick thread with the main findings 👇
⚡ Quick thread with the main findings 👇
Excited to share our work on mechanisms of naturalistic audiovisual processing in the human brain 🧠🎬!!
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
www.biorxiv.org
November 7, 2025 at 4:01 PM
Excited to share our work on mechanisms of naturalistic audiovisual processing in the human brain 🧠🎬!!
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...