Dataset Viewer
Auto-converted to Parquet Duplicate
Search is not available for this dataset
event_id
int64
0
10k
pileup
int32
200
200
jets
listlengths
97
123
trks
listlengths
4.21k
7.8k
jet_trk_IDs
listlengths
97
123
6,252
200
[[188.30747985839844,0.40465378761291504,3.1189498901367188,82.25028991699219,0.560958206653595,0.18(...TRUNCATED)
[[23.7512149810791,-1.8050389289855957,-3.026211977005005,-1.0,0.21143803000450134,57.01585006713867(...TRUNCATED)
[[31498,34358,26764,7103,16860,28149,9756,14509,26215,2124,22384,11776,18831,14851,34177,17919,12626(...TRUNCATED)
4,684
200
[[176.46214294433594,2.0088720321655273,4.058175086975098,80.3828353881836,0.41735219955444336,0.138(...TRUNCATED)
[[33.32120132446289,2.1570935249328613,0.831310510635376,-1.0,0.09258443862199783,-13.57294464111328(...TRUNCATED)
[[16497,11223,15915,9409,4653,16362,17793,30944,30532,34552,589,4364,11602,33398,32866,16611,4306,81(...TRUNCATED)
1,731
200
[[253.46978759765625,-0.5253357291221619,5.056132793426514,99.46022033691406,0.5144069790840149,0.07(...TRUNCATED)
[[131.25018310546875,-0.4859497845172882,-1.2473067045211792,-1.0,-0.5146170854568481,4.388370037078(...TRUNCATED)
[[25125,9315,21943,5629,35344,36497,26318,1705,20266,5573,25462,3414,10629,22558,20676,32430,21479,9(...TRUNCATED)
4,742
200
[[287.6410217285156,-3.2100741863250732,2.9187633991241455,94.3255844116211,0.6922895312309265,0.319(...TRUNCATED)
[[60.510108947753906,1.8507089614868164,-0.4589442014694214,-1.0,0.31703975796699524,-1.623993039131(...TRUNCATED)
[[21077,8778,19740,14435,34168,3231,17675,24593,33198,21444,106,37276,32808,7558,30733,30289,29905,1(...TRUNCATED)
4,521
200
[[237.60684204101562,-1.9516308307647705,2.466914176940918,102.08302307128906,0.3552144765853882,0.0(...TRUNCATED)
[[89.53085327148438,-1.8858344554901123,2.5092365741729736,-1.0,-0.24757228791713715,-10.65832710266(...TRUNCATED)
[[20868,27136,14736,38475,26615,11759,10884,26376,4912,33414,31315,26883,31321,1027,31373,38290,2688(...TRUNCATED)
6,340
200
[[202.85098266601562,1.8450003862380981,1.6849673986434937,107.31182098388672,0.25181305408477783,0.(...TRUNCATED)
[[60.06742858886719,0.21831277012825012,-0.8228257298469543,-1.0,-0.10129044950008392,-13.0796318054(...TRUNCATED)
[[9215,21258,94,7538,8151,35703,40572,34433,3165,20982,36376,4087,29755,12040,18157,4221,38146,26715(...TRUNCATED)
576
200
[[188.74644470214844,2.211611270904541,5.9757232666015625,86.35066986083984,0.5247436165809631,0.277(...TRUNCATED)
[[10.963845252990723,-0.6806262135505676,1.462962031364441,-1.0,0.2802756428718567,23.51789474487304(...TRUNCATED)
[[22845,19892,22094,19864,3667,18683,20912,4889,19784,3225,30999,25658,37127,24130,20910,71,34955,17(...TRUNCATED)
5,202
200
[[169.31141662597656,0.7289472818374634,5.36779260635376,89.99187469482422,0.2516530752182007,0.1582(...TRUNCATED)
[[51.62885284423828,1.012660026550293,1.8072117567062378,-1.0,-0.24880732595920563,63.62326049804687(...TRUNCATED)
[[801,6822,30678,2511,10741,7937,36000,9909,31136,29574,29567,1711,17637,35976,17257,23370,28120,421(...TRUNCATED)
6,363
200
[[161.7676544189453,-2.387652635574341,0.4096512198448181,81.13631439208984,0.28096702694892883,0.10(...TRUNCATED)
[[12.584325790405273,2.7792885303497314,-0.11084053665399551,-1.0,0.5355644226074219,-9.837354660034(...TRUNCATED)
[[9815,31462,26231,20902,12149,20820,32610,13050,8637,17104,5192,882,6996,21252,4790,38664,16425,861(...TRUNCATED)
439
200
[[335.2276306152344,0.10672713816165924,1.4414582252502441,112.75796508789062,0.6638429760932922,0.1(...TRUNCATED)
[[68.89164733886719,-1.6830577850341797,-2.5704457759857178,-1.0,0.6030519604682922,-0.3349425196647(...TRUNCATED)
[[29081,5185,14789,8831,32194,1255,30770,17680,13747,26803,1254,20689,6485,1256,6798,7387,20002,1365(...TRUNCATED)
End of preview. Expand in Data Studio

PhyGHT Top Quark Events

Lossless Parquet redistribution of the official PhyGHT top-quark pair simulation dataset. Each configuration has 10,000 events: mu200 models extreme HL-LHC pileup, and mu60 the lower pileup scenario. These are simulation events, not measured detector collisions.

Load

from datasets import load_dataset
events = load_dataset("Rakib/PhyGHT-TopQuark", "mu200")
event = events["test"][0]
assert len(events["train"]) == 8000

Use "mu60" for the other configuration. streaming=True permits streaming without loading the entire configuration. Explicit train/validation/test split paths are in this card's metadata; no executable dataset script is required.

Schema and units

Each row is one original simulated event. event_id is its zero-based index in the corresponding official pickle; unique identity is (pileup, event_id). pileup is 200 or 60. All original rows, column ordering, variable-length lists, float64 values, and int64 associations are preserved without quantization.

  • jets: list of jets, each [pT, eta, phi, mass, Efrac, Mfrac] (float64).
  • trks: list of tracks, each [pT, eta, phi, mass, d0, z0, track_id, truth_label] (float64 as in the source, including integral identifiers and labels).
  • jet_trk_IDs: one list of int64 track identifiers for each jet, in original order. These refer to track column 6, not positions in the event track list.
  • The model uses the first four jet columns and the first six track columns, converts inputs to float32, and predicts the last two jet columns. Track auxiliary labels are truth_label == -1 for hard-scatter signal; other labels denote pileup. No truth labels or identifiers enter the feature encoders.

The official walkthrough explicitly labels track pT in GeV. Eta and the fraction targets are dimensionless; phi is an azimuthal angle. The release does not assign unverified units to mass, d0, or z0.

Reproducible event splits

The official preprocessing algorithm is preserved exactly: numpy.random.RandomState(42).permutation(10000), then the first 8,000 events for train, next 1,000 for validation, and last 1,000 for test. Shard row order matches these indices. Indices for each configuration are saved in splits/. All three splits are disjoint within a configuration. Event numbers across different pileup configurations do not establish independent physical provenance or alignment beyond the source simulation's own guarantees.

Citation and sources

@misc{rakib2026phyghtphysicsguidedhypergraphtransformer,
  title={PhyGHT: Physics-Guided HyperGraph Transformer for Signal Purification at the HL-LHC},
  author={Mohammed Rakib and Luke Vaughan and Shivang Patel and Flera Rizatdinova and Alexander Khanov and Atriya Sen},
  year={2026}, eprint={2602.20475}, archivePrefix={arXiv}, primaryClass={hep-ex},
  url={https://arxiv.org/abs/2602.20475}
}
Downloads last month
93

Models trained or fine-tuned on Rakib/PhyGHT-TopQuark

Paper for Rakib/PhyGHT-TopQuark