# Ranking Preservation on CoreBench Hard (test)

Source: [Ranking Preservation on CoreBench Hard (test)](<https://www.sota2.com/research/sota/ranking-preservation-on-corebench-hard-test>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-26T13:37:25.447Z

Evaluation of model performance on the Ranking Preservation task using the test split of the CoreBench Hard dataset, measured using Spearman correlation coefficients.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Ranking Preservation |
| Dataset (including any stated split) | CoreBench Hard (test) |
| Metric | Mean Spearman Rho |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Mean Spearman Rho](<https://www.sota2.com/research/sota/ranking-preservation-on-corebench-hard-test.md?metric=Mean%20Spearman%20Rho>): Higher is better; unit: SCALAR.

- [Std Dev Spearman Rho](<https://www.sota2.com/research/sota/ranking-preservation-on-corebench-hard-test.md?metric=Std%20Dev%20Spearman%20Rho>): Lower is better; unit: SCALAR.

- [Worst Spearman Rho](<https://www.sota2.com/research/sota/ranking-preservation-on-corebench-hard-test.md?metric=Worst%20Spearman%20Rho>): Higher is better; unit: SCALAR.

- [Best Spearman Rho](<https://www.sota2.com/research/sota/ranking-preservation-on-corebench-hard-test.md?metric=Best%20Spearman%20Rho>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Mean Spearman Rho | Result date | Report paper |
| --- | --- | --- | --- |
| MR | 0.901 | Not recorded | 69c4bd730636e5bf4afc74ed |
| Easiest | 0.896 | Not recorded | 69c4bd730636e5bf4afc74ed |
| Greedy | 0.869 | Not recorded | 69c4bd730636e5bf4afc74ed |
| Random | 0.78 | Not recorded | 69c4bd730636e5bf4afc74ed |
| Stratified | 0.741 | Not recorded | 69c4bd730636e5bf4afc74ed |

### MR

Mean Spearman Rho: 0.901. Result ID: 69c5366147f34ce924aed2ac.

[Reporting paper](<https://www.sota2.com/research/paper/69c4bd730636e5bf4afc74ed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c4bd730636e5bf4afc74ed.md>)

Recorded evaluation configuration: {"k":"14","Selection strategy":"Mid-Range","Evaluation protocol":"mean across all protocols"}. This record does not assert independent reproduction.

### Easiest

Mean Spearman Rho: 0.896. Result ID: 69c5366147f34ce924aed2ae.

[Reporting paper](<https://www.sota2.com/research/paper/69c4bd730636e5bf4afc74ed.md>)

Recorded evaluation configuration: {"k":"14","Selection strategy":"Easiest","Evaluation protocol":"mean across all protocols"}. This record does not assert independent reproduction.

### Greedy

Mean Spearman Rho: 0.869. Result ID: 69c5366147f34ce924aed2ad.

[Reporting paper](<https://www.sota2.com/research/paper/69c4bd730636e5bf4afc74ed.md>)

Recorded evaluation configuration: {"k":"14","Selection strategy":"Greedy","Evaluation protocol":"mean across all protocols"}. This record does not assert independent reproduction.

### Random

Mean Spearman Rho: 0.78. Result ID: 69c5366147f34ce924aed2b0.

[Reporting paper](<https://www.sota2.com/research/paper/69c4bd730636e5bf4afc74ed.md>)

Recorded evaluation configuration: {"k":"14","Selection strategy":"Random","Evaluation protocol":"mean across all protocols"}. This record does not assert independent reproduction.

### Stratified

Mean Spearman Rho: 0.741. Result ID: 69c5366147f34ce924aed2af.

[Reporting paper](<https://www.sota2.com/research/paper/69c4bd730636e5bf4afc74ed.md>)

Recorded evaluation configuration: {"k":"14","Selection strategy":"Stratified","Evaluation protocol":"mean across all protocols"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
