Research map / Label-efficient and robust learning / Self-supervised and contrastive
FM4NPP: A Scaling Foundation Model for Nuclear and Particle Physics
David Keetae Park, Shuhang Li, Yi Huang, Xihaier Luo, Haiwang Yu, Yeonju Go, Christopher Pinkenburg, Yuewei Lin, Shinjae Yoo, Joseph D. Osborn, Jin Huang, Yihui Ren
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add pretrained FM4NPP(m6) features to adapter (ablation) | adapter · Track Finding | ARI: 0.7243 -> 0.9448 | Improved |
| add pretrained FM4NPP(m6) features to adapter (ablation) | adapter · Particle Identification | accuracy: 0.6631 -> 0.9039 | Improved |
| add pretrained FM4NPP(m6) features to adapter (ablation) | adapter · Noise Tagging | accuracy: 0.9111 -> 0.9713 | Improved |
| FM4NPP with 2.39M-parameter adapter versus Exa.TrkX GNN | Exa.TrkX · Track Finding | purity (%): 66.42 -> 93.08 | Improved |
| FM4NPP versus OneFormer3D on noise tagging recall | OneFormer3D · Noise Tagging | macro recall: 0.9404 -> 0.9367 | No clear effect |
| increase pretrained FM size (ablation) | FM4NPP · downstream tasks | performance | Improved |
| increase FM size from m5 to m6 (ablation) | FM4NPP · pretraining validation MSE | validation MSE | No clear effect |
| increase pretraining dataset size (ablation) | FM4NPP m3 · pretraining validation MSE | validation MSE | Improved |
| pretrained FM with few labeled examples versus many (ablation) | FM4NPP · Track Finding | relative ARI gain: 2.9x vs 1.3x | Improved |
| FM4NPP versus sPHENIX Cellular Automaton + Kalman pipeline | sPHENIX pipeline · Track Finding (high-pT tracks) | tracking efficiency (%): 94.6 -> 99.6 | Improved |
Closest papers by meaning
- Mitigating Premature Exploitation in Particle-based Monte Carlo for Inference-Time Scaling (ICML 2026 · similarity 0.85)
- Towards Fast, Specialized Machine Learning Force Fields: Distilling Foundation Models via Energy Hessians (ICLR 2025 · similarity 0.84)
- Specialized Foundation Models Struggle to Beat Supervised Baselines (ICLR 2025 · similarity 0.84)
- Scaling Physical Reasoning with the PHYSICS Dataset (NeurIPS 2025 · similarity 0.84)
- Rollout Roulette: A Probabilistic Inference Approach to Inference-Time Scaling of LLMs using Particle-Based Monte Carlo Methods (NeurIPS 2025 · similarity 0.83)
- Neural Interpretable PDEs: Harmonizing Fourier Insights with Attention for Scalable and Interpretable Physics Discovery (ICML 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
