Skip to main content
AIMSAI Measurement Science

ML Engineering & Research

ML4CO-Bench-101

ML4CO-Bench-101: a modular benchmark of neural combinatorial-optimization solvers under a paradigm-model-learning taxonomy, over 7 graph CO problems (TSP, ATSP, CVRP, MIS, MCl, MVC, MCut). This build ingests the released per-instance neural-solver outputs (ML4CO/ML4CO-Bench-101-SL, ml4co_result/) and computes the achieved objective value per instance, with the reference (optimal/near-optimal) objective as correct_answer.

35,578items
1subjects
100%observed
Modelsubject type
CC-BY-4.0license
generaldomain
textmodality
Original sourcePaperBuild script← All benchmarks

Response matrix

Every model, scored item by item.

Each row is an AI model and each column an item, ordered so the strongest models and easiest items gather toward one corner. 1 subjects × 35,578 items, 100% of cells evaluated.

Fit to width. Hover for subject & item; click a cell for details.

ML4CO-Bench-101 response matrix: AI models (rows) against items (columns)
lowhighUnobserved

Scale: Achieved objective value per instance — tour cost (TSP/ATSP/CVRP), set size (MIS/MVC/MCl), or cut value (MCut); each combinatorial problem is shown on its own scale.

Sample items

What the questions look like — and how subjects answer.

A spread of items across the difficulty range. This benchmark does not publish per-answer traces, so each item shows which subjects succeeded.

Item 199% solve rateanswer: 0.986935

ML4CO-Bench-101 ATSP instance, scale 50, #1782

Subject outcomes

  • ML4CO-SL-solverscore 0.988
Item 2772% solve rateanswer: 7.723952

ML4CO-Bench-101 TSP instance, scale 100, #662

Subject outcomes

  • ML4CO-SL-solverscore 7.724
Item 31084% solve rateanswer: 10.685417

ML4CO-Bench-101 CVRP instance, scale 50, #9278

Subject outcomes

  • ML4CO-SL-solverscore 10.836
Item 41477% solve rateanswer: 14.476074

ML4CO-Bench-101 CVRP instance, scale 100, #6429

Subject outcomes

  • ML4CO-SL-solverscore 14.767
Item 51922% solve rateanswer: 18.993048

ML4CO-Bench-101 CVRP instance, scale 100, #6151

Subject outcomes

  • ML4CO-SL-solverscore 19.216
Item 6893000% solve rateanswer: 8734.000000

ML4CO-Bench-101 MCut instance, scale ba-giant, #5

Subject outcomes

  • ML4CO-SL-solverscore 8930

Subjects

The models, agents, and reward models evaluated.

1 subjects, ranked by mean response across this benchmark's items.

  1. 1ML4CO-SL-solver102.815
Full data on Hugging FaceBack to the gallery