Reproduce “TokaMind: A Multi-Modal Transformer Foundation Model for Tokamak Plasma Dynamics” 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
Tobia Boschi, Andrea Loreti, Nicola C. Amorisco, Rodrigo H. Ordonez-Hurtado, Cécile Rousseau, George K. Holt, Eszter Székely, Alexander Whittle and 8 more
TokaMind reconstructs and predicts tokamak plasma states from heterogeneous missing diagnostics. It tokenizes sensor chunks, marks unavailable inputs and targets, and adapts a pretrained Transformer to task-specific outputs. On MAST’s TokaMark benchmark, fine-tuned Base reaches NRMSE 0.080 on Task 4-3, beating Scratch and CNN+LSTM; it beats the baseline on all but Task 4-5, with evidence limited to MAST.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.