Reproduce “Scaling Parameter-Efficient Fine-Tuning for Brain Encoding” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhuofan Josh Ying, Guanglong Sun, Nolan Ho, Kieran Liu, Rui Yang, Andrew Chen, Nikolaus Kriegeskorte, Pony Ma and 1 more
BrainLoRA adapts pretrained multimodal models to predict fMRI responses to naturalistic movies using low-rank updates and a causal single-pass feature-extraction scheme. On the CNeuroMod/Algonauts 2025 task, its Qwen3-Omni linear-readout model trains 61.2M parameters and scores 0.322 in-distribution and 0.222 out-of-distribution, exceeding published MIRAGE scores with 1.1B trainable parameters. The evidence is limited to narrative-film settings.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.