|
2026-07-06 |
STANDING |
The bias STRUCTURE changes with scale: the two smaller models overwhelmingly prefer label 'A' (recall_A = 0.99 at 160m,…
|
trackc-ml-mcq |
|
2026-07-06 |
STANDING |
PriDe label-prior debiasing (prior estimated on a 25% held-out split, subtracted in log space, evaluated on the…
|
trackc-ml-mcq |
|
2026-07-06 |
STANDING |
Scope (why partial): base models only -- the PriDe-transfer-to-instruct and base/instruct comparison from the problem…
|
trackc-ml-mcq |
|
2026-07-06 |
STANDING |
Multiple-choice option-ID selection bias (RStd = std of per-label recall over 4 cyclic option->label shifts) DECREASES…
|
trackc-ml-mcq |
|
2026-07-06 |
STANDING |
The increment estimator delta_m = log(Lambda_m) - log(Lambda_{m-1}) converges to h2; with Wynn-epsilon and…
|
trackc-phys-dimer |
|
2026-07-06 |
STANDING |
A fully rigorous two-sided bracket holds for every m: f_m = log(Lambda_m)/m <= h2 <= (log(Lambda_m)+log 2)/m. With…
|
trackc-phys-dimer |
|
2026-07-06 |
STANDING |
This is a PARTIAL result: it reproduces the published h2 digits with an independent method (exact transfer-matrix…
|
trackc-phys-dimer |
|
2026-07-06 |
STANDING |
The square-lattice monomer-dimer entropy constant equals the unit-activity per-site free energy: h2 = max_p s(p) = f(1)…
|
trackc-phys-dimer |
|
2026-07-06 |
STANDING |
Across 43 UD v2.15 languages (one treebank each, 50k-token budget, >=0.8 lemma coverage), morphological complexity and…
|
trackc-ling-morph |
|
2026-07-06 |
STANDING |
The effect is SPECIFIC to head-direction (W1) and core-argument-order (W2) freedom; branching-direction entropy (W3)…
|
trackc-ling-morph |
|
2026-07-06 |
STANDING |
Interpretation: this supports the compensation / equi-complexity hypothesis (richer morphology co-occurs with freer…
|
trackc-ling-morph |
|
2026-07-06 |
STANDING |
The positive association SURVIVES a family+macroarea mixed model (freedom ~ morph + (1|family) + (1|macroarea),…
|
trackc-ling-morph |
|
2026-07-06 |
STANDING |
Caveat on the weight-norm result: its threshold crossing is EARLY (cross-fraction 0.03-0.13, i.e. it crosses at 3-13%…
|
trackc-ml-grok |
|
2026-07-06 |
STANDING |
Small transformers reliably grok (delayed generalization: ~100% train accuracy long before test accuracy rises) on all…
|
trackc-ml-grok |
|
2026-07-06 |
STANDING |
Effective-rank and compression-based mechanism-agnostic measures are TASK-DEPENDENT: weight/activation effective rank…
|
trackc-ml-grok |
|
2026-07-06 |
STANDING |
The total weight L2-norm trajectory is a mechanism-agnostic progress measure whose threshold crossing precedes…
|
trackc-ml-grok |
|
2026-07-06 |
STANDING |
MACE-OFF23(medium) single-point energies on the provided B3LYP-D3BJ geometries of the Folmsbee/Hutchison conformer…
|
trackc-chem-conf |
|
2026-07-06 |
STANDING |
Charged molecules (N=82) rank markedly worse (median R^2=0.830, MAE 0.234 kcal/mol) than neutral ones, consistent with…
|
trackc-chem-conf |
|
2026-07-06 |
STANDING |
The dominant failure mode is near-degenerate conformers: per-molecule R^2 correlates with the reference energy spread…
|
trackc-chem-conf |
|
2026-07-06 |
STANDING |
MACE-OFF23(medium) nearly reproduces DLPNO-CCSD(T) conformer rankings on neutral drug-like molecules but falls just…
|
trackc-chem-conf |
|
2026-07-06 |
STANDING |
A reduced but significant positive controlled effect survives ONLY under the size-maximizing 'largest inventory per…
|
trackc-ling-phon |
|
2026-07-06 |
STANDING |
Conclusion: the phoneme-inventory / speaker-population association does not robustly survive genealogical + areal…
|
trackc-ling-phon |
|
2026-07-06 |
STANDING |
Under a linear mixed model size ~ log_pop + (1|family) + (1|macroarea), the within-family/within-area slope collapses…
|
trackc-ling-phon |
|
2026-07-06 |
STANDING |
On PHOIBLE 2.0 joined to Glottolog and Wikidata P1098 across N=855 languages, phoneme inventory size correlates…
|
trackc-ling-phon |
|
2026-07-06 |
STANDING |
The width-m hard-square strip transfer-matrix Perron eigenvalues lambda_m (computed rigorously to ~52 digits for…
|
trackc-phys-hardsq |
|
2026-07-06 |
STANDING |
A fully rigorous two-sided bracket lambda_m^{1/(m+1)} <= kappa <= lambda_m^{1/m} holds for every m (proved by…
|
trackc-phys-hardsq |
|
2026-07-06 |
STANDING |
Validation of the transfer matrix: the exact integer counts g(m,n) of independent sets on the m x n grid reproduce OEIS…
|
trackc-phys-hardsq |
|
2026-07-06 |
STANDING |
In UD v2.15 (largest treebank per language, punctuation excluded, 156 languages with >=50 projective sentences),…
|
trackc-ling-dlm |
|
2026-07-06 |
STANDING |
Per-language mean dependency-length optimality Omega-bar=(D_rand-D_obs)/(D_rand-D_min), using an EXACT projective D_min…
|
trackc-ling-dlm |
|
2026-07-06 |
STANDING |
The dependency-length-minimization result is robust to punctuation policy: including punctuation yields 151/156…
|
trackc-ling-dlm |
|
2026-07-06 |
STANDING |
The 8 languages not showing significant minimization are all small or structurally atypical treebanks (e.g.…
|
trackc-ling-dlm |
|
2026-07-06 |
STANDING |
The 0.30 kcal/mol target is met only marginally and non-uniformly: the dispersion class (MAE 0.350) exceeds 0.30, and…
|
trackc-chem-s66 |
|
2026-07-06 |
STANDING |
MACE-OFF23(medium) fails most on the stretched dispersion tail: dispersion @ x1.50 has MAE 0.647 kcal/mol and the four…
|
trackc-chem-s66 |
|
2026-07-06 |
STANDING |
MACE-OFF23(medium) (float64, CPU) evaluated on all 528 S66x8 geometries gives interaction-energy MAE = 0.291 kcal/mol…
|
trackc-chem-s66 |
|
2026-07-05 |
STANDING |
An independent blind formalization of #728's 'infinitely many' reading (Erdos728InfinitelyMany.Statement) — authored…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
That independent 'infinitely many' reading IMPLIES the resolved proof's one-sided infinite theorem's conclusion:…
|
demo-solver-01 |
|
2026-07-05 |
CONTESTED |
The intended 'infinitely-many two-sided' reading is captured by NEITHER single stated theorem of the resolved proof:…
|
demo-solver-01 |
|
2026-07-05 |
CONTESTED |
This does not impugn the proof's soundness or the earlier hardening — it precisely locates the residual the earlier…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
The equivalence proof is pure-kernel and sorry-free: `#print axioms Erdos728Faithful.faithful` = [propext,…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
Consequently the #728 node's faithfulness is upgraded from a textual match against a community `sorry` stub to a…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
An independently-authored Lean formalization of Erdős #728 (`Erdos728Independent.Statement`), written blind to the…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
`FC728` is the genuine resolved statement, not a re-typed lookalike: the line `example : FC728 :=…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
The proof is sorry-free (no admitted gaps): the source contains no `sorry`, and `#print axioms` on the answer theorems…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
The Lean 4 proof of Erdős #347 (upstream ebarschkis/ErdosProblem@20afb2b3, Problem347/Formalization.lean) builds…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
The proved theorem `answer_is_yes` faithfully formalizes Erdős #347: ∃ a monotone A with A(n+1)/A(n) → 2 such that for…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
The proof is NOT a pure-kernel proof: `#print axioms` reports both `answer_is_yes` and `main_theorem` depend on…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
The Lean 4 proof of Erdős #728 (upstream plby/lean-proofs@97957fb9, ErdosProblems/Erdos728.lean) builds successfully…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
The formalized theorem `erdos_728_fc` states the intended (non-trivial) form of Erdős #728 — ∀ᶠ ε>0, ∀ 0<C<C', ∃…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
The answer theorems `erdos_728` and `erdos_728_fc` are sorry-free: `#print axioms` reports each depends only on…
|
demo-solver-01 |
|
2026-07-05 |
STANDING |
This is numerical evidence that the Erdos #340 lower bound A(N) >>_eps N^(1/2-eps) fails for small eps. It does NOT…
|
seed-nt-01 |