# Cross-coding Reconstruction on Qwen Subliminal

Source: [Cross-coding Reconstruction on Qwen Subliminal](<https://www.sota2.com/research/sota/cross-coding-reconstruction-on-qwen-subliminal>)

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

Data updated: 2026-03-06T07:28:25.533Z

Evaluation of Cross-coding Reconstruction performance using the Qwen Subliminal dataset, measured by metrics including Explanatory Variance (Expl. Var.), count of dead elements (# Dead), and percentage of dead elements (Dead %).

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 | Cross-coding Reconstruction |
| Dataset (including any stated split) | Qwen Subliminal |
| Metric | Explanatory Variance |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [Explanatory Variance](<https://www.sota2.com/research/sota/cross-coding-reconstruction-on-qwen-subliminal.md?metric=Explanatory%20Variance>): Higher is better; unit: PERCENT.

- [Count of Dead Elements](<https://www.sota2.com/research/sota/cross-coding-reconstruction-on-qwen-subliminal.md?metric=Count%20of%20Dead%20Elements>): Lower is better; unit: SCALAR.

- [Dead Elements (%)](<https://www.sota2.com/research/sota/cross-coding-reconstruction-on-qwen-subliminal.md?metric=Dead%20Elements%20(%25)>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Explanatory Variance | Result date | Report paper |
| --- | --- | --- | --- |
| TopK-400 | 80.07 | Not recorded | 69aa5f7c2d603e6de3e41859 |
| DSF | 80.07 | Not recorded | 69aa5f7c2d603e6de3e41859 |
| TopK-200 | 79.29 | Not recorded | 69aa5f7c2d603e6de3e41859 |
| Delta | 76.17 | Not recorded | 69aa5f7c2d603e6de3e41859 |

### TopK-400

Explanatory Variance: 80.07. Result ID: 69aa81e86dbb7075b953e3bc.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f7c2d603e6de3e41859.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76301e209131cea6bbec.md>)

Recorded evaluation configuration: {"Dict. Size":"17920"}. This record does not assert independent reproduction.

### DSF

Explanatory Variance: 80.07. Result ID: 69aa81e86dbb7075b953e3bd.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f7c2d603e6de3e41859.md>)

Recorded evaluation configuration: {"Dict. Size":"17920"}. This record does not assert independent reproduction.

### TopK-200

Explanatory Variance: 79.29. Result ID: 69aa81e86dbb7075b953e3bb.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f7c2d603e6de3e41859.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76301e209131cea6bbec.md>)

Recorded evaluation configuration: {"Dict. Size":"17920"}. This record does not assert independent reproduction.

### Delta

Explanatory Variance: 76.17. Result ID: 69aa81e86dbb7075b953e3ba.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f7c2d603e6de3e41859.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69aa5f7c2d603e6de3e41859.md>)

Recorded evaluation configuration: {"Dict. Size":"17920"}. This record does not assert independent reproduction.

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