Agent skill · research science · aradotso
tribev2-brain-encoding
Use TRIBE v2, Meta's multimodal foundation model for predicting fMRI brain responses to video, audio, and text stimuli
Why this skill is useful
Includes executable scripts for predicting brain responses and visualizing results that extend the AI's capabilities in neuroscience modeling.
What it needs
Requires pip installed locally. Requires huggingface account access. About 4k tokens when loaded. Last updated 2026-07-09. 70 stars on the source repository.
What this skill does
TRIBE v2 Brain Encoding Model Skill by ara.so — Daily 2026 Skills collection TRIBE v2 is Meta's multimodal foundation model that predicts fMRI brain responses to naturalistic stimuli (video, audio, text). It combines LLaMA 3.2 (text), V-JEPA2 (video), and Wav2Vec-BERT (audio) encoders into a unified Transformer architecture that maps multimodal representations onto the cortical surface (fsaverage5, ~20k vertices). Installation Quick Start — Inference Load pretrained model and predict from video Multimodal input — video + audio + text Text-only prediction Brain Visualization Training a Model from Scratch 1. Set environment variables 2. Authenticate with HuggingFace (required for LLaMA 3.2) 3. Local test run 4. Full grid search on Slurm Key API — TribeModel Project Structure Configuration — Defaults Edit tribev2/grids/defaults.py or set environment variables: Custom Experiment with PyTorch Lightning Working with fMRI Surfaces Common Patterns Batch prediction over multiple videos Extract predictions for specific brain region Access segment timing metadata Troubleshooting LLaMA 3.2 access denied CUDA out of memory during inference Missing visualization dependencies Slurm training not submitting Video without audio track causes error Citation Resources Paper Interactive Demo HuggingFace Weights Colab Notebook
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills aradotso/tribev2-brain-encoding