List your company

Formal Theorem Proving on miniF2F (val)

42.2Pass@1POETRY
23.16828.10933.0537.991Feb 3, 2022Jun 23, 2022Nov 10, 2022Mar 30, 2023Aug 17, 2023Jan 4, 2024May 23, 2024

Evaluation results

MethodLinks
POETRYenvironment=Isabelle2024.0542.2——
Thor + expert iterationenvironment=Isabelle2024.0537.3——
Thor + Magnushammerenvironment=Isabelle2024.0536.9——
θ_full (expert iterated on full curriculum)search width (d)=512, expansion factor (e)=8, parameters=774M2022.0233.641.247.3
FMSCLenvironment=Lean2024.0533.6——
θ_mathlib (expert iterated on mathlib-train)search width (d)=512, expansion factor (e)=8, parameters=774M2022.0231.338.344.1
θ₁ (value-function based search)search width (d)=512, expansion factor (e)=8, parameters=774M2022.0228.535.541.2
θ1d (expansions)=512, e (samples per expansion)=8, objective=proofsize, search=best-first search2022.0228.535.5—
θ0d (expansions)=512, e (samples per expansion)=8, search=cumulative logprob priority best-first search2022.0228.433.6—
θ1 (outcome objective)d (expansions)=512, e (samples per expansion)=8, objective=outcome, search=best-first search2022.0228.334.7—
Thorenvironment=Isabelle2024.0528.3——
θ0 (MiniF2F setup)d (expansions)=128, e (samples per expansion)=16, search=cumulative logprob priority best-first search2022.0227.631.8—
PACTsearch width (d)=512, expansion factor (e)=82022.0223.929.3—
MiniF2Fd (expansions)=128, e (samples per expansion)=162022.0223.929.3—
PACTenvironment=Lean2024.0523.9——