Find research datasets worth reusing

MLPF results on the simulated CLIC dataset

<p>Updates over the previous version:</p> <ul> <li>updated validation outputs for the cluster-based model <ul> <li>fixed a bug with how the PF candidates were stored</li> <li>added single particle gun samples to validation</li> </ul> </li> <li>added new timing runs <ul> <li>for the baseline algo, included memory information</li> <li>run the GNN model up to ~10k inputs</li> </ul> </li> <li>added hypertuning summary tables</li> </ul> <p>Trained models and evaluation results for the upcoming paper "Improved particle-flow event reconstruction with scalable neural networks for current and future particle detectors",&nbsp;https://doi.org/10.48550/arXiv.2309.06782.</p> <p>The archive contains the following subfolders:</p> <ul> <li><strong>clusters_best_tuned_gnn_clic_v130</strong> <ul> <li>MLPF GNN model configs and weight files after hypertuning</li> <li>the inputs are reconstructed tracks and Pandora clusters</li> <li>the outputs are reconstructed PF candidates</li> <li>trained on tt and qq v1.3.0 (1M events each)</li> </ul> </li> <li><strong>hits</strong> <ul> <li>MLPF GNN model configs and weight files</li> <li>inputs are reconstructed tracks and calorimeter hits</li> <li>outputs are reconstructed PF candidates</li> <li>trained on tt, qq and gun samples (K0L, gamma, pi+-, pi0, neutron, ele, mu) v1.2.0</li> <li>training was restarted several times from previous checkpoints</li> </ul> </li> <li><strong>hypertuning</strong> <ul> <li>GNN and transformer model before and after hypertuning</li> <li>summary tables of the hypertuning runs</li> </ul> </li> <li><strong>timing</strong> <ul> <li>scaling study of baseline PF with number of gun particles on CPU</li> <li>scaling study of GNN model with number of input elements on GPU</li> </ul> </li> <li><strong>gpu_scaling</strong> <ul> <li>the scaling study of model training on multiple accelerator cards</li> </ul> </li> </ul> <p>&nbsp;</p> <p>The training dataset is available at</p> <ul> <li>Pata, Joosep, Wulff, Eric, Duarte, Javier, Mokhtar, Farouk, Zhang, Mengke, Girone, Maria, &amp; Southwick, David. (2023). Simulated datasets for detector and particle flow reconstruction: CLIC detector (1.1) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.8260741</li> </ul>