|
# Phase 2 update: flopscope v0.12.3, deadline extension, and final submission nomination
|
|
0
|
12
|
October 17, 2026
|
|
Approximate cumulant propagation with exact compute reductions
|
|
0
|
17
|
October 16, 2026
|
|
[Phase 2 write-up] A third-cumulant chain at its boundary: random projections, a skew law, and where the next order's accuracy lives (submission #333534)
|
|
0
|
13
|
October 16, 2026
|
|
[Phase 2 write-up] Removing the global scale mode makes a carried κ4 core pay: extraction on/off ×0.380 on the grader inside e30_a's engine (submission #336317)
|
|
0
|
24
|
October 15, 2026
|
|
[Phase 2 write-up] The (3,1) slice of κ₄: where V29 loses accuracy, and a two-term closure confirmed in a pre-registered test on 77 held-out networks (submission #334821)
|
|
0
|
27
|
October 15, 2026
|
|
How the hidden-set evaluation invokes whest run
|
|
3
|
87
|
October 12, 2026
|
|
Flopscope: fix pull requests, with tests, for the open billing issues
|
|
0
|
35
|
October 12, 2026
|
|
Eligibility of solutions built on another participant's MIT-licensed forum code (§4 / §6)
|
|
1
|
95
|
October 5, 2026
|
|
Everything we tried: a factorized K=3 cumulant-propagation estimator at 0.25 x B, where its FLOPs go, and 25 measured dead ends (team 504aldo, rank ~10)
|
|
5
|
341
|
October 4, 2026
|
|
Phase 1 final evaluation question: Overfitting to public instances
|
|
5
|
846
|
October 4, 2026
|
|
Leaderboard vs Overview leading scores
|
|
0
|
75
|
October 3, 2026
|
|
Phase 2 of the ARC White-Box Estimation Challenge is live 🚀
|
|
3
|
308
|
October 1, 2026
|
|
[Phase 1 write-up] Fixed grid of directions instead of random sampling (submission #320802) and five things that did not work
|
|
1
|
360
|
September 22, 2026
|
|
[Phase 2] The compute budget is not a lever, and the scoring rule is why: `adjusted = mse x max(0.1, util)`
|
|
0
|
114
|
September 17, 2026
|
|
Errors are leaking information
|
|
1
|
144
|
September 9, 2026
|
|
[Phase 2 write-up] Zero fitted constants, measured: what the 64 calibration numbers in a ReLU-MLP mean propagator are actually absorbing (submission #330001)
|
|
0
|
120
|
September 6, 2026
|
|
Phase 2 write-up: exact bivariate ReLU covariance plus per-layer calibration, 3x over the kit baseline, with negative results (submission 329951)
|
|
0
|
102
|
September 5, 2026
|
|
Interesting talks by the organizers
|
|
2
|
191
|
September 4, 2026
|
|
Covariance propagation baseline scores 3.93e-7/3.93e-6
|
|
2
|
120
|
September 4, 2026
|
|
Request to Claude and GPT
|
|
2
|
187
|
August 31, 2026
|
|
Shipped covariance_propagation example trips the Phase 2 residual cap locally (0.16.0) but grades fine - what happens when the evaluator upgrades?
|
|
4
|
151
|
August 29, 2026
|
|
[Phase 1 write-up] Variance reduction beats closure refinement: a moment-matched sampling estimator (submission #324497)
|
|
0
|
79
|
August 26, 2026
|
|
[Phase 2 write-up] Complementary errors beat either branch: a fixed 0.75/0.25 covariance + whitened antithetic MC blend (submission 328274)
|
|
0
|
130
|
August 26, 2026
|
|
"Unusually easy leaderboard set" effect is gone
|
|
1
|
121
|
August 24, 2026
|
|
Phase 2 — Official Start Date and Details?
|
|
2
|
114
|
August 23, 2026
|
|
Phase 1 SOLUTION, ideas for Phase 2 and the honest public frontier (all methods combined and explained)
|
|
4
|
234
|
August 23, 2026
|
|
Queries regarding phase 2
|
|
1
|
120
|
August 20, 2026
|
|
How did you use AI?
|
|
2
|
128
|
August 20, 2026
|
|
[Phase 1 research atlas] How I reached public #16 — and why the next 56 experiments failed
|
|
3
|
230
|
August 20, 2026
|
|
Phase 1 Report, Submission 327801
|
|
5
|
153
|
August 20, 2026
|