EveNet_Result_Plot

⭐ Final Naming Table (Nature-Style, Jargon-Free, Consistent)

1. Model Variants

We evaluate four variants of EveNet that differ in their pretraining strategy:

• Fully pretrained EveNet model Self-supervised + supervised pretraining Short-form: EveNet–Full

• Self-supervised–only EveNet model Stage-1 SSL pretraining only Short-form: EveNet–SSL

• Ablation-pretrained EveNet variant Pretrained without the supervised generative objective Short-form: EveNet–Abl.

• EveNet trained from random initialization No pretraining Short-form: EveNet–Scratch

These names refer exclusively to the pretrained backbone weights.

Concept	Long-form (for main text)	Short-form (for figures and tables)	Notes
Fully pretrained model (“Nominal”)	fully pretrained model (self-supervised + supervised stages)	Full	Nature prefers plain English (“full model”).
Self-supervised only (“SSL”)	self-supervised–only pretrained model	SSL	Standard ML term; safe.
Ablation-pretrained model	ablation-pretrained variant (pretrained with segmentation)	Abl. or Abl-Seg	Use “Abl.” unless you need explicit “Abl-SG.”
Random initialization (“Scratch”)	model trained from random initialization	Scratch	Universally understood.

---

2. Task Heads

Concept	Long-form (for main text)	Short-form (for captions)	Notes
Classification head	classification head	Cls	“Classification” is accessible; “Cls” is concise.
Assignment head	object–resonance assignment head	Asn	Avoid jargon (“assignment” alone may confuse beginners).
Segmentation head	final-state segmentation head	Seg	Standard terminology.
Supervised generative head	supervised generative head	SG	“SG” clearer than “SGEN” for figures.
Self-supervised generative head	self-supervised generative head	SSG	Parallel with SG.

Combinations for caption labels: Cls + Asn, Cls + SG, Asn + Seg, etc. (Best practice: Title-case them, not all caps.)

Category	Long-form (main text)	Short-form (figures/tables)
Fully pretrained EveNet	fully pretrained EveNet model	EveNet–Full
SSL-only EveNet	self-supervised–only EveNet model	EveNet–SSL
Ablation EveNet	ablation-pretrained EveNet variant	EveNet–Abl.
Scratch EveNet	EveNet trained from random initialization	EveNet–Scratch
Classification head	classification head	Cls
Assignment head	object–resonance assignment head	Asn
Segmentation head	final-state segmentation head	Seg
Supervised generation	supervised generative head	SG
Self-supervised generation	self-supervised generative head	SSG
Quantum-correlation benchmark	top-quark quantum-correlation analysis	QC
Rare-Higgs benchmark	rare Higgs-boson decay benchmark (H→2a→4b)	RH / H→2a→4b
Anomaly detection	dimuon anomaly-detection benchmark	AD

---

3. Downstream Benchmarks

Study	Long-form (main text)	Short-form (figures)	Notes
Top-quark quantum-correlation / entanglement study (“QE”)	top-quark quantum-correlation analysis (includes spin correlation and entanglement)	QC	Avoid “QE” (“quantum efficiency”). QC is clear.
Exotic Higgs / BSM decay benchmark	rare Higgs-boson decay benchmark (H→2a→4b)	RH or H→2a→4b	“Rare Higgs decay” avoids jargon like “exotic” or “BSM”.
Anomaly detection (“AD”)	dimuon anomaly-detection benchmark using real collision data	AD	Clear even outside HEP.

---

⭐ 4. Combined Overview Table (Full Final Table)

Below is the single integrated table you can drop into your Supplementary Methods or into a Consistency Appendix.

---

📘 Final Jargon-Free Naming Table for the Foundation-Model Paper

Category	Concept	Long-form description (main text)	Short-form (for figures/tables)
Model Variant	Fully pretrained	fully pretrained model	Full
	Self-supervised only	self-supervised–only pretrained model	SSL
	Ablation-pretrained	ablation-pretrained variant (without supervised generation)	Abl. / Abl-SG
	Scratch	model trained from random initialization	Scratch
Task Head	Classification	classification head	Cls
	Assignment	object–resonance assignment head	Asn
	Segmentation	final-state segmentation head	Seg
	Supervised generation	supervised generative head	SG
	Self-supervised generation	self-supervised generative head	SSG
Downstream Task	Quantum correlation (formerly “QE”)	top-quark quantum-correlation analysis	QC
	Rare Higgs decay (formerly “BSM”)	rare Higgs-boson decay benchmark (H→2a→4b)	RH or H→2a→4b
	Anomaly detection	dimuon anomaly-detection benchmark using real data	AD

🔗 Plot gallery website

This repository now includes a small static gallery that renders the paper plots into PNGs and publishes them to GitHub Pages. The CI workflow builds the gallery on every push to main.

Local preview

python -m pip install -r requirements.txt
python website/build_gallery.py --output-dir site --format png --dpi 200
python -m http.server --directory site 8000

Open http://localhost:8000 to browse the same gallery that will appear on GitHub Pages.