Schedule & rewards
A three-month development phase, a blind final, and recognition that goes well beyond the cash awards.
Key dates
- Feb 2026 Complete
Dataset assembly and curation
- Mar 2026 Complete
Baseline code, evaluation scripts, dFL visualizer
- Apr 2026 Complete
Website online
- Jul 2026 Complete
Public release of training & public test data
- Jul 27 – Oct 18, 2026
Phase 1 (development) — continuous public leaderboard
- Oct 19, 2026
Blind test data released; Phase 2 (final) opens
- Oct 26, 2026
Phase 2 (final) closes — last call for submissions
- Nov 2026
Verification, peer review of top entries, prize allocation
- Dec 2026
NeurIPS competition session & results
- Q1 2027
Lessons-learned community paper with selected winners
$1,000 — one $500 award per challenge
There are two challenges and two awards, matched one to one. You may enter either or both; entering both makes you eligible for both.
Best DIII-D intra-machine reconstruction
Highest composite S_model on the hidden DIII-D test set.
Best DIII-D → MAST cross-machine generalization
Highest G_ratio among entries reaching R²ψ > 0.6 on DIII-D.
Needs predictions for both machines — a DIII-D-only entry scores
G_ratio G_ratio The score that decides Challenge 2: G = S_model(MAST) / S_model(DIII-D) — the fraction of your DIII-D performance you keep when moving zero-shot to MAST. 1.0 would mean perfect transfer. It needs predictions for both machines: a DIII-D-only entry scores 0. = 0.
Live leaderboard Phase 1 — development
Mirrored from Codabench, which is the source of truth for every score. 18 entries · as of Aug 11, 2026
| # | Participant | Ch1: DIII-D S | Ch2: G_ratio |
|---|---|---|---|
| 1 | mnn31 | 0.9907 | 0.5316 |
| 2 | dremovd | 0.9829 | 0.068 |
| 3 | ruiqi_feng | 0.9758 | 0.0 |
| 4 | sigmaborov | 0.9598 | 0.5241 |
| 5 | ekaterinauu | 0.9568 | 0.6134 |
Four honorable mentions
Recognition & support
Co-authorship & talks
Certificates & talks
Compute for everyone
Accessible by design
Both awards are reachable without dedicated hardware. The compute-light badge Compute-light badge A self-declared badge for models that train end-to-end in under two hours on a single GPU or CPU. It is a recognition that applies across both challenges — not a separate track, and not a separate leaderboard. celebrates entries that train in under two hours on a single commodity GPU or CPU — it applies across both challenges, and every reference baseline qualifies. See the challenge and the baselines for what that looks like in practice.